feat(nodes): make Lora Stack Combiner inputs dynamic

After b464fdc3 (preserve .git on release switch), the hasGit-based
channel detection is unreliable — .git now exists for both release
and nightly installs, so page refresh always reset the channel.

- Add _resolveChannelFromSettings() with migration heuristic:
  !hasGit → release (ZIP), detached HEAD → release (on tag),
  on branch → nightly. Uses gitInfo.branch from check-updates.
- Persist resolved channel to settings.json on first load
  (one-time migration) and on explicit switchChannel.
- Add update_channel validation (release|nightly) in backend
  update_settings handler.
- Remove hasGit-based guessing from initialize(); defer to
  checkForUpdates where full gitInfo is available.
- Channel resolution runs before checkForUpdates early-returns
  to avoid null channelMode on reload-within-interval.

Tests: 361 passed.

.docs/batch-import-design.md

@@ -1,153 +0,0 @@
-# Recipe Batch Import Feature Design
-
-## Overview
-Enable users to import multiple images as recipes in a single operation, rather than processing them individually. This feature addresses the need for efficient bulk recipe creation from existing image collections.
-
-## Architecture
-
-```
-┌─────────────────────────────────────────────────────────────────┐
-│                        Frontend                                  │
-├─────────────────────────────────────────────────────────────────┤
-│  BatchImportManager.js                                          │
-│  ├── InputCollector (收集URL列表/目录路径)                        │
-│  ├── ConcurrencyController (自适应并发控制)                       │
-│  ├── ProgressTracker (进度追踪)                                   │
-│  └── ResultAggregator (结果汇总)                                  │
-├─────────────────────────────────────────────────────────────────┤
-│  batch_import_modal.html                                        │
-│  └── 批量导入UI组件                                              │
-├─────────────────────────────────────────────────────────────────┤
-│  batch_import_progress.css                                      │
-│  └── 进度显示样式                                                │
-└─────────────────────────────────────────────────────────────────┘
-                              │
-                              ▼
-┌─────────────────────────────────────────────────────────────────┐
-│                        Backend                                   │
-├─────────────────────────────────────────────────────────────────┤
-│  py/routes/handlers/recipe_handlers.py                          │
-│  ├── start_batch_import() - 启动批量导入                         │
-│  ├── get_batch_import_progress() - 查询进度                      │
-│  └── cancel_batch_import() - 取消导入                            │
-├─────────────────────────────────────────────────────────────────┤
-│  py/services/batch_import_service.py                            │
-│  ├── 自适应并发执行                                              │
-│  ├── 结果汇总                                                    │
-│  └── WebSocket进度广播                                           │
-└─────────────────────────────────────────────────────────────────┘
-```
-
-## API Endpoints
-
-| 端点 | 方法 | 说明 |
-|------|------|------|
-| `/api/lm/recipes/batch-import/start` | POST | 启动批量导入，返回 operation_id |
-| `/api/lm/recipes/batch-import/progress` | GET | 查询进度状态 |
-| `/api/lm/recipes/batch-import/cancel` | POST | 取消导入 |
-
-## Backend Implementation Details
-
-### BatchImportService
-
-Location: `py/services/batch_import_service.py`
-
-Key classes:
-- `BatchImportItem`: Dataclass for individual import item
-- `BatchImportProgress`: Dataclass for tracking progress
-- `BatchImportService`: Main service class
-
-Features:
-- Adaptive concurrency control (adjusts based on success/failure rate)
-- WebSocket progress broadcasting
-- Graceful error handling (individual failures don't stop the batch)
-- Result aggregation
-
-### WebSocket Message Format
-
-```json
-{
-    "type": "batch_import_progress",
-    "operation_id": "xxx",
-    "total": 50,
-    "completed": 23,
-    "success": 21,
-    "failed": 2,
-    "skipped": 0,
-    "current_item": "image_024.png",
-    "status": "running"
-}
-```
-
-### Input Types
-
-1. **URL List**: Array of URLs (http/https)
-2. **Local Paths**: Array of local file paths
-3. **Directory**: Path to directory with optional recursive flag
-
-### Error Handling
-
-- Invalid URLs/paths: Skip and record error
-- Download failures: Record error, continue
-- Metadata extraction failures: Mark as "no metadata"
-- Duplicate detection: Option to skip duplicates
-
-## Frontend Implementation Details (TODO)
-
-### UI Components
-
-1. **BatchImportModal**: Main modal with tabs for URLs/Directory input
-2. **ProgressDisplay**: Real-time progress bar and status
-3. **ResultsSummary**: Final results with success/failure breakdown
-
-### Adaptive Concurrency Controller
-
-```javascript
-class AdaptiveConcurrencyController {
-    constructor(options = {}) {
-        this.minConcurrency = options.minConcurrency || 1;
-        this.maxConcurrency = options.maxConcurrency || 5;
-        this.currentConcurrency = options.initialConcurrency || 3;
-    }
-
-    adjustConcurrency(taskDuration, success) {
-        if (success && taskDuration < 1000 && this.currentConcurrency < this.maxConcurrency) {
-            this.currentConcurrency = Math.min(this.currentConcurrency + 1, this.maxConcurrency);
-        }
-        if (!success || taskDuration > 10000) {
-            this.currentConcurrency = Math.max(this.currentConcurrency - 1, this.minConcurrency);
-        }
-        return this.currentConcurrency;
-    }
-}
-```
-
-## File Structure
-
-```
-Backend (implemented):
-├── py/services/batch_import_service.py            # 后端服务
-├── py/routes/handlers/batch_import_handler.py     # API处理器 (added to recipe_handlers.py)
-├── tests/services/test_batch_import_service.py    # 单元测试
-└── tests/routes/test_batch_import_routes.py       # API集成测试
-
-Frontend (TODO):
-├── static/js/managers/BatchImportManager.js       # 主管理器
-├── static/js/managers/batch/                       # 子模块
-│   ├── ConcurrencyController.js                    # 并发控制
-│   ├── ProgressTracker.js                          # 进度追踪
-│   └── ResultAggregator.js                         # 结果汇总
-├── static/css/components/batch-import-modal.css   # 样式
-└── templates/components/batch_import_modal.html    # Modal模板
-```
-
-## Implementation Status
-
-- [x] Backend BatchImportService
-- [x] Backend API handlers
-- [x] WebSocket progress broadcasting
-- [x] Unit tests
-- [x] Integration tests
-- [ ] Frontend BatchImportManager
-- [ ] Frontend UI components
-- [ ] E2E tests

.gitignore

@@ -7,17 +7,24 @@ py/run_test.py
 .vscode/
 cache/
 civitai/
+stats/
+wildcards/
+backups/
+logs/
 node_modules/
 coverage/
 .coverage
 model_cache/
 
-# agent
+# agent / dev tooling
 .opencode/
 .claude/
 .sisyphus/
 .codex
 .omo
+reasonix.toml
+.reasonix/
+.codegraph/
 
 # Vue widgets development cache (but keep build output)
 vue-widgets/node_modules/
@@ -26,3 +33,10 @@ vue-widgets/dist/
 
 # Hypothesis test cache
 .hypothesis/
+
+# Working/research notes (not committed)
+.docs/
+
+# HF enrichment validation baseline snapshots (contain potentially
+# NSFW README content fetched from community model repos)
+tests/enrich_hf_validation/baselines/

AGENTS.md

@@ -102,6 +102,7 @@ npm run test:coverage       # Generate coverage report
 - ComfyUI: `app.registerExtension()`, `node.addDOMWidget(name, type, element, options)`
 - Event handlers via `addEventListener` or widget callbacks
 - Shared utilities: `web/comfyui/utils.js`
+- Dual-mode rendering patterns (canvas vs Vue): see `docs/comfyui-dual-mode-widgets.md`
 
 ### Vue Composables Pattern
 
@@ -136,7 +137,13 @@ npm run test:coverage       # Generate coverage report
 - Dual mode: ComfyUI plugin (folder_paths) vs standalone (settings.json)
 - Detection: `os.environ.get("LORA_MANAGER_STANDALONE", "0") == "1"`
 - Run `python scripts/sync_translation_keys.py` after adding UI strings to `locales/en.json`
-- Symlinks require normalized paths
+- Symlinks require normalized paths.
+  **Business paths vs real paths**: All stored paths and operation routing use the
+  original paths as they appear under configured model roots — symlinks are NOT
+  resolved. `os.path.realpath` is only for scanner dedup and the symlink cache.
+  Any path passed to `os.remove`/`os.rename`/`shutil.move` or validated by a
+  containment check MUST use the business path (i.e. `os.path.abspath`, not
+  `realpath`).
 
 ## Git / Commit Messages
 

__init__.py

@@ -15,6 +15,10 @@ try:  # pragma: no cover - import fallback for pytest collection
     from .py.nodes.lora_pool import LoraPoolLM
     from .py.nodes.lora_randomizer import LoraRandomizerLM
     from .py.nodes.lora_cycler import LoraCyclerLM
+    from .py.nodes.lora_info import LoraInfoLM
+    from .py.nodes.lora_syntax_to_path import LoraSyntaxToPath
+    from .py.nodes.create_hook_lora import CreateHookLoraLM
+    from .py.nodes.metadata_overwrite import MetadataOverwriteLM
     from .py.metadata_collector import init as init_metadata_collector
 except (
     ImportError
@@ -56,6 +60,16 @@ except (
         "py.nodes.lora_randomizer"
     ).LoraRandomizerLM
     LoraCyclerLM = importlib.import_module("py.nodes.lora_cycler").LoraCyclerLM
+    LoraInfoLM = importlib.import_module("py.nodes.lora_info").LoraInfoLM
+    LoraSyntaxToPath = importlib.import_module(
+        "py.nodes.lora_syntax_to_path"
+    ).LoraSyntaxToPath
+    CreateHookLoraLM = importlib.import_module(
+        "py.nodes.create_hook_lora"
+    ).CreateHookLoraLM
+    MetadataOverwriteLM = importlib.import_module(
+        "py.nodes.metadata_overwrite"
+    ).MetadataOverwriteLM
     init_metadata_collector = importlib.import_module("py.metadata_collector").init
 
 NODE_CLASS_MAPPINGS = {
@@ -75,6 +89,10 @@ NODE_CLASS_MAPPINGS = {
     LoraPoolLM.NAME: LoraPoolLM,
     LoraRandomizerLM.NAME: LoraRandomizerLM,
     LoraCyclerLM.NAME: LoraCyclerLM,
+    LoraInfoLM.NAME: LoraInfoLM,
+    LoraSyntaxToPath.NAME: LoraSyntaxToPath,
+    CreateHookLoraLM.NAME: CreateHookLoraLM,
+    MetadataOverwriteLM.NAME: MetadataOverwriteLM,
 }
 
 WEB_DIRECTORY = "./web/comfyui"

docs/agent_skills.md

@@ -0,0 +1,208 @@
+# Agent Skills System
+
+The LoRA Manager agent skills system enables LLM-powered metadata enrichment and other AI-driven tasks. Users configure their own LLM provider (BYOK), and skills are executed through right-click context menu actions.
+
+## Architecture
+
+```
+┌──────────────────────────────────────────────┐
+│              LoRA Manager Backend             │
+│                                               │
+│  ┌──────────────┐    ┌────────────────┐       │
+│  │ LLMService    │───▶│ LLM Provider   │       │
+│  │ (BYOK config, │◀───│ (OpenAI/Ollama │       │
+│  │  API calls)   │    │ /custom)       │       │
+│  └───────┬───────┘    └────────────────┘       │
+│          │                                     │
+│  ┌───────▼───────────────────────┐             │
+│  │     AgentService              │             │
+│  │  (orchestration: validate     │             │
+│  │   → LLM call → post-process   │             │
+│  │   → WebSocket broadcast)      │             │
+│  └───────┬───────────────────────┘             │
+│          │                                     │
+│  ┌───────▼───────────────────────┐             │
+│  │     SkillRegistry             │             │
+│  │  ┌─────────────────────────┐  │             │
+│  │  │ enrich_hf_metadata:     │  │             │
+│  │  │  - skill.yaml           │  │             │
+│  │  │  - prompt.md            │  │             │
+│  │  │  - handler.py           │  │             │
+│  │  └─────────────────────────┘  │             │
+│  └───────────────────────────────┘             │
+└──────────────────────────────────────────────┘
+```
+
+### Key Design Principle
+
+**Skills define *what* to do (prompt + post-processing). The AgentService handles *how* (LLM calls, validation, progress).**
+
+Skills never call the LLM directly. This keeps BYOK configuration centralized and provider-agnostic.
+
+## BYOK Configuration
+
+Users configure their LLM provider in **Settings → AI Provider**:
+
+| Setting | Description | Example |
+|---|---|---|
+| `llm_provider` | Provider type | `openai`, `ollama`, or `custom` |
+| `llm_api_key` | API key (not needed for local Ollama) | `sk-...` |
+| `llm_api_base` | Custom API base URL (empty = provider default) | `https://api.openai.com/v1` |
+| `llm_model` | Model name | `gpt-4o-mini` |
+
+Environment variable overrides: `LLM_API_KEY`, `LLM_MODEL`, `LLM_API_BASE`, `LLM_PROVIDER`.
+
+### Supported Providers
+
+- **OpenAI**: Uses `https://api.openai.com/v1` by default
+- **Ollama** (local): Uses `http://localhost:11434/v1`, no API key required
+- **Custom**: Any OpenAI-compatible endpoint (vLLM, LM Studio, etc.) — set `llm_api_base` explicitly
+
+## Available Skills
+
+### enrich_hf_metadata
+
+Enriches HuggingFace-downloaded models with metadata extracted by an LLM from the HF model card.
+
+**Entry point**: Right-click context menu → "Enrich Metadata (Agent)"
+
+**What it does**:
+1. Reads the model's `.metadata.json` to get the `hf_url`
+2. Fetches the README.md from the HuggingFace repository
+3. Sends the README + local metadata to the LLM for structured extraction
+4. Writes extracted fields to `.metadata.json`:
+   - `base_model` — only if current value is empty
+   - `trainedWords` — trigger words (LoRA only, if none exist)
+   - `modelDescription` — concise summary (if none exists)
+   - `tags` — merged with existing tags, deduplicated
+   - `metadata_source` — audit trail: `agent:enrich_hf_metadata`
+   - `llm_enriched_at` — ISO timestamp
+5. Downloads and optimizes preview image (if LLM found one in the README)
+6. Updates the scanner cache
+7. Broadcasts WebSocket progress events
+
+**Model types**: LoRA, Checkpoint, Embedding
+
+## Adding a New Skill
+
+### 1. Create the skill directory
+
+```
+py/services/agent/skills/<skill_name>/
+├── skill.yaml      # Skill metadata and schemas
+├── prompt.md       # LLM prompt template
+└── handler.py      # Pre-processing and post-processing
+```
+
+### 2. Write skill.yaml
+
+```yaml
+name: my_skill
+title: "My Skill"
+description: "What this skill does"
+llm_required: true
+model_type_filter: ["lora"]  # or null for all types
+input_schema:
+  type: object
+  properties:
+    model_paths:
+      type: array
+      items:
+        type: string
+  required:
+    - model_paths
+output_schema:
+  type: object
+  properties:
+    # ... JSON schema for LLM output
+permissions:
+  write_metadata: true
+  write_previews: false
+  network_domains:
+    - "example.com"
+```
+
+### 3. Write prompt.md
+
+Use `{{variable}}` placeholders that will be replaced with data from the `prepare` function:
+
+```markdown
+You are an expert assistant...
+
+Model URL: {{hf_url}}
+README content:
+{{readme_content}}
+
+Current metadata:
+{{current_metadata}}
+```
+
+### 4. Write handler.py
+
+```python
+async def prepare(model_path: str, input_data: dict) -> dict:
+    """Gather context for the LLM prompt. Returns variables for template rendering."""
+    return {
+        "model_path": model_path,
+        # ... other variables used in prompt.md
+    }
+
+async def post_process(context) -> dict:
+    """Apply the LLM-extracted data to the model."""
+    llm_response = context.llm_response
+    # ... write metadata, download previews, update cache
+    return {
+        "success": True,
+        "updated_fields": ["base_model", "tags"],
+        "errors": [],
+    }
+```
+
+**Important**: Use absolute imports (`from py.utils.metadata_manager import MetadataManager`) because skills are loaded via `importlib.util.spec_from_file_location`, which doesn't support relative imports.
+
+### 5. Test
+
+The skill is automatically discovered by `SkillRegistry` on startup. Test with:
+
+```python
+pytest tests/services/test_agent_service.py
+```
+
+## API Endpoints
+
+| Method | Path | Description |
+|---|---|---|
+| GET | `/api/lm/agent/skills` | List available skills |
+| POST | `/api/lm/agent/execute/{skill_name}` | Execute a skill (body: `{"model_paths": [...]}`) |
+| POST | `/api/lm/agent/cancel` | Cancel running skill (stub) |
+
+## WebSocket Events
+
+| Type | When | Key fields |
+|---|---|---|
+| `agent_progress` | Skill started/processing | `skill`, `status`, `total`, `processed`, `success`, `current_path` |
+| `agent_progress` | Skill completed | `skill`, `status`, `updated_models`, `errors`, `summary` |
+| `agent_progress` | Skill error | `skill`, `status`, `error` |
+
+## Security Model
+
+Skills declare permissions in `skill.yaml`:
+- `write_metadata` — can write `.metadata.json` files
+- `write_previews` — can download/replace preview images
+- `network_domains` — allowed domains for HTTP requests
+
+These are declarative constraints checked by `AgentService`. They are defense-in-depth, not a sandbox — the Python process can technically do anything, but the contract is clear and auditable.
+
+## File Locations
+
+| Component | Path |
+|---|---|
+| LLMService | `py/services/llm_service.py` |
+| AgentService | `py/services/agent/agent_service.py` |
+| SkillRegistry | `py/services/agent/skill_registry.py` |
+| SkillDefinition | `py/services/agent/skill_definition.py` |
+| Skills directory | `py/services/agent/skills/` |
+| Route handlers | `py/routes/handlers/agent_handlers.py` |
+| Frontend manager | `static/js/managers/AgentManager.js` |
+| Settings UI | `templates/components/modals/settings_modal.html` |
+| Context menu | `templates/components/context_menu.html` |

docs/comfyui-dual-mode-widgets.md

@@ -0,0 +1,65 @@
+# ComfyUI Dual-Mode Widget Rendering
+
+ComfyUI custom node widgets render in one of two modes. Patterns that work in one often fail silently in the other. Test both.
+
+## Mode Detection
+
+```js
+typeof LiteGraph !== 'undefined' && LiteGraph.vueNodesMode
+```
+
+In Vue SFCs, `window.LiteGraph` is unavailable — pass as a prop from `main.ts`.
+
+## Canvas Mode Layout
+
+Uses `computeLayoutSize()` + `distributeSpace()` to allocate widget height within the node. Widgets with `computeLayoutSize` participate in space distribution; those with `computeSize` have fixed height.
+
+- `getMinHeight()` in `addDOMWidget` options → minimum widget height
+- `widget.computeLayoutSize()` → `{ minHeight, minWidth, maxHeight? }`
+- Avoid `getMaxHeight()` unless the widget genuinely needs a fixed cap (prevents user resize)
+
+## Vue Mode Layout
+
+Uses CSS Grid (`grid-template-rows`) + `ResizeObserver`. The ResizeObserver watches the widget's DOM and feeds back into grid row sizing. This creates a feedback loop: content grows → row resizes → more space for content → content reflows/grows → row resizes again.
+
+### Height Containment
+
+The fix: `contain: layout size` on the widget root. This tells the browser the element's intrinsic size is CSS-determined, not driven by descendant content. The ResizeObserver sees a stable size and the loop is broken.
+
+```css
+.widget-root.lm-vue-node {
+  height: 100%;
+  min-height: var(--comfy-widget-min-height, 200px);
+  contain: layout size;
+}
+```
+
+Existing examples: `.lm-loras-container.lm-vue-node` and `.comfy-tags-container.lm-vue-node` in `web/comfyui/lm_styles.css`.
+
+**Do NOT** fix height issues with `maxHeight`, `getMaxHeight()`, or inline `max-height` — these prevent the user from resizing the node.
+
+## Scroll Wheel Isolation
+
+Both modes need to distinguish "user wants to scroll widget content" from "user wants to zoom canvas".
+
+**Canvas mode:** Add `@wheel` on widget root. Check `event.target.closest(selector)` for scrollable sub-areas. If scrollable → `event.stopPropagation()`. Otherwise → `app.canvas.processMouseWheel(event)`.
+
+**Vue mode:** Add CSS class `lm-wheel-scrollable` to scrollable elements. The global capture-phase hook in `web/comfyui/utils.js` (`enableListWheelScroll`) detects wheel events on marked elements and manually scrolls them via `element.scrollTop`, consuming the event before canvas zoom sees it.
+
+## DOM Structure
+
+`main.ts` creates an outer `<div>` container, then `vueApp.mount(container)`. The Vue app renders its own root element inside.
+
+- `container.id` / `container.style.*` → outer element
+- Vue scoped `<style>` → `[data-v-hash]` applies only to Vue root
+
+Classes needed by scoped Vue CSS must go on the Vue root element. Pass data as props and bind with `:class` rather than manipulating the DOM from `main.ts`.
+
+## Serialization
+
+For stateful widgets that need workflow persistence:
+
+- `serialize: true` in `addDOMWidget` options
+- `serializeValue()` → state snapshot (called on workflow save)
+- `onSetValue(v)` → restore state (called on workflow load)
+- Always handle missing keys in restored value for backward compatibility with old workflows

py/config.py

@@ -8,6 +8,8 @@ from typing import Any, Dict, Iterable, List, Mapping, Optional, Set, Tuple
 import logging
 import json
 import urllib.parse
+import sys as _sys
+import types as _types
 import time
 
 from .utils.cache_paths import CacheType, get_cache_file_path, get_legacy_cache_paths
@@ -175,8 +177,7 @@ class Config:
 
         # Load extra folder paths from active library settings before symlink scan
         # so both primary and extra paths are discovered in a single pass.
-        if not standalone_mode:
-            self._load_extra_paths_from_settings()
+        self._load_extra_paths_from_settings()
 
         # Scan symbolic links during initialization
         self._initialize_symlink_mappings()
@@ -191,7 +192,7 @@ class Config:
         Called during ``Config.__init__`` before the symlink scan so both primary and
         extra paths are discovered in a single pass.  Mirrors the extra-path
         portion of ``_apply_library_paths`` without replacing the primary roots
-        that were already resolved from ComfyUI's ``folder_paths``.
+        that were already resolved via ``folder_paths.get_folder_paths``.
         """
         try:
             from .services.settings_manager import get_settings_manager
@@ -207,6 +208,12 @@ class Config:
             if not isinstance(library_config, dict):
                 return
 
+            # Always read recipes_path — it is independent of extra folder paths
+            # and must be set before any early returns below.
+            recipes_path = library_config.get("recipes_path", "")
+            if isinstance(recipes_path, str) and recipes_path:
+                self.recipes_path = recipes_path
+
             extra_folder_paths = library_config.get("extra_folder_paths")
             if not isinstance(extra_folder_paths, dict):
                 return
@@ -232,10 +239,6 @@ class Config:
                 extra_embedding
             )
 
-            recipes_path = library_config.get("recipes_path", "")
-            if isinstance(recipes_path, str) and recipes_path:
-                self.recipes_path = recipes_path
-
             if self.extra_loras_roots:
                 logger.info(
                     "Found extra LoRA roots:"
@@ -356,6 +359,47 @@ class Config:
                         "Failed to rename legacy 'default' library: %s", rename_error
                     )
 
+            # Clean up a stale "default" library entry that has no meaningful
+            # paths configured (e.g. leftover bootstrap artifact).  This only
+            # fires when "comfyui" already exists so we never delete the last
+            # remaining library.
+            if (
+                "default" in libraries
+                and "comfyui" in libraries
+                and isinstance(default_library, Mapping)
+            ):
+                default_folder_paths = _normalize_library_folder_paths(
+                    default_library
+                )
+                default_extra_paths = default_library.get("extra_folder_paths", {})
+                has_meaningful_paths = bool(default_folder_paths) or bool(
+                    default_extra_paths
+                ) or any(
+                    default_library.get(key)
+                    for key in (
+                        "default_lora_root",
+                        "default_checkpoint_root",
+                        "default_unet_root",
+                        "default_embedding_root",
+                        "recipes_path",
+                    )
+                )
+                if not has_meaningful_paths:
+                    try:
+                        settings_service.delete_library("default")
+                        libraries_changed = True
+                        logger.info(
+                            "Removed stale 'default' library entry "
+                            "with no meaningful paths configured"
+                        )
+                        libraries = settings_service.get_libraries()
+                        comfy_library = libraries.get("comfyui", {})
+                    except Exception as delete_error:
+                        logger.debug(
+                            "Failed to remove stale 'default' library: %s",
+                            delete_error,
+                        )
+
             default_lora_root = _resolve_valid_default_root(
                 comfy_library.get("default_lora_root", ""),
                 list(self.loras_roots or []),
@@ -1380,4 +1424,20 @@ class Config:
 
 
 # Global config instance
-config = Config()
+# NOTE: Guard against re-import.  When ServiceRegistry.get_lora_scanner() triggers
+# a fresh import of lora_scanner → config, we must NOT re-execute Config.__init__()
+# (which re-scans all roots, re-registers libraries, etc.).
+#
+# Strategy: store the config instance in a dedicated sentinel module
+# ('_lm_config_cache') that is NEVER removed from sys.modules (its key does
+# NOT start with 'py.'), so it survives re-imports of py.* modules.
+_CONFIG_SENTINEL = "_lm_config_cache"
+if _CONFIG_SENTINEL in _sys.modules:
+    # Re-import: reuse the existing singleton from the sentinel.
+    config: Config = _sys.modules[_CONFIG_SENTINEL].config  # type: ignore[valid-type]
+else:
+    config: Config = Config()
+    # Register the sentinel so re-imports of py.config find us.
+    _sentinel_mod = _types.ModuleType(_CONFIG_SENTINEL)
+    _sentinel_mod.config = config
+    _sys.modules[_CONFIG_SENTINEL] = _sentinel_mod

py/lora_manager.py

@@ -208,6 +208,10 @@ class LoraManager:
             # Initialize WebSocket manager
             await ServiceRegistry.get_websocket_manager()
 
+            # Preload LLM model catalog (background task, non-blocking)
+            from .services.llm_service import LLMService
+            await LLMService.get_instance()
+
             # Initialize scanners in background
             lora_scanner = await ServiceRegistry.get_lora_scanner()
             checkpoint_scanner = await ServiceRegistry.get_checkpoint_scanner()
@@ -436,5 +440,21 @@ class LoraManager:
         try:
             logger.info("LoRA Manager: Cleaning up services")
 
+            # Cancel any in-flight scanner initialization tasks so thread-pool
+            # workers (e.g. _initialize_cache_sync) can break out of their loops
+            # when the server shuts down (e.g. Ctrl+C on WSL).
+            for name in ("lora_scanner", "checkpoint_scanner", "embedding_scanner"):
+                scanner = ServiceRegistry.get_service_sync(name)
+                if scanner is not None and hasattr(scanner, "cancel_task"):
+                    scanner.cancel_task()
+                    logger.debug("LoRA Manager: Cancelled %s", name)
+
+            # Close shared aiohttp sessions to avoid "Unclosed client session" warnings
+            try:
+                from py.routes.handlers.hf_handlers import close_hf_api_session
+                await close_hf_api_session()
+            except Exception as exc:
+                logger.debug("Error closing HF API session: %s", exc)
+
         except Exception as e:
             logger.error(f"Error during cleanup: {e}", exc_info=True)

py/metadata_collector/constants.py

@@ -1,5 +1,11 @@
 """Constants used by the metadata collector"""
 
+# Sentinel value for clip_skip to distinguish "unconnected / widget default"
+# from "user wired value 0".  Both ComfyUI CLIPSetLastLayer (-24..-1) and
+# A1111 conventions treat 0 as meaningless for clip skipping, but users may
+# explicitly wire 0 to the overwrite node to express "no clip skip / default".
+CLIP_SKIP_SENTINEL = -25
+
 # Metadata categories
 MODELS = "models"
 PROMPTS = "prompts"
@@ -9,6 +15,14 @@ EMBEDDINGS = "embeddings"
 SIZE = "size"
 IMAGES = "images"
 IS_SAMPLER = "is_sampler"  # New constant to mark sampler nodes
+OVERWRITE = "overwrite"  # Manual metadata overwrite from MetadataOverwriteLM node
+
+# Field names that the MetadataOverwriteLM node and its extractor share
+METADATA_OVERWRITE_FIELDS = (
+    "prompt", "negative_prompt", "seed", "steps", "cfg_scale",
+    "sampler", "scheduler", "model", "loras", "size",
+    "clip_skip", "additional_data",
+)
 
 # Complete list of categories to track
-METADATA_CATEGORIES = [MODELS, PROMPTS, SAMPLING, LORAS, EMBEDDINGS, SIZE, IMAGES]
+METADATA_CATEGORIES = [MODELS, PROMPTS, SAMPLING, LORAS, EMBEDDINGS, SIZE, IMAGES, OVERWRITE]

py/metadata_collector/metadata_hook.py

@@ -83,7 +83,8 @@ class MetadataHook:
                         
                         # Record inputs before execution
                         if node_id is not None:
-                            registry.record_node_execution(node_id, class_type, input_data_all, None)
+                            return_types = getattr(obj, 'RETURN_TYPES', None)
+                            registry.record_node_execution(node_id, class_type, input_data_all, None, return_types=return_types)
                 except Exception as e:
                     logger.error(f"Error collecting metadata (pre-execution): {str(e)}")
             
@@ -114,7 +115,8 @@ class MetadataHook:
                         
                         # Record outputs after execution
                         if node_id is not None:
-                            registry.update_node_execution(node_id, class_type, results)
+                            return_types = getattr(obj, 'RETURN_TYPES', None)
+                            registry.update_node_execution(node_id, class_type, results, return_types=return_types)
                 except Exception as e:
                     logger.error(f"Error collecting metadata (post-execution): {str(e)}")
             
@@ -135,10 +137,13 @@ class MetadataHook:
                 # Store the dynprompt reference for node lookups
                 if hasattr(prompt, 'original_prompt'):
                     registry.set_current_prompt(prompt)
-            
+
+                # Store extra_data for accessing full workflow node properties
+                registry.set_extra_data(extra_data)
+
             # Execute the original function
             return original_execute(*args, **kwargs)
-            
+
         # Replace the functions
         execution._map_node_over_list = map_node_over_list_with_metadata
         execution.execute = execute_with_prompt_tracking
@@ -163,7 +168,8 @@ class MetadataHook:
                         class_type = obj.__class__.__name__
                         node_id = unique_id
                         if node_id is not None:
-                            registry.record_node_execution(node_id, class_type, input_data_all, None)
+                            return_types = getattr(obj, 'RETURN_TYPES', None)
+                            registry.record_node_execution(node_id, class_type, input_data_all, None, return_types=return_types)
                 except Exception as e:
                     logger.error(f"Error collecting metadata (pre-execution): {str(e)}")
 
@@ -180,7 +186,8 @@ class MetadataHook:
                         class_type = obj.__class__.__name__
                         node_id = unique_id
                         if node_id is not None:
-                            registry.update_node_execution(node_id, class_type, results)
+                            return_types = getattr(obj, 'RETURN_TYPES', None)
+                            registry.update_node_execution(node_id, class_type, results, return_types=return_types)
                 except Exception as e:
                     logger.error(f"Error collecting metadata (post-execution): {str(e)}")
 
@@ -202,6 +209,9 @@ class MetadataHook:
                 if hasattr(prompt, 'original_prompt'):
                     registry.set_current_prompt(prompt)
 
+                # Store extra_data for accessing full workflow node properties
+                registry.set_extra_data(extra_data)
+
             # Execute the original function
             return await original_execute(*args, **kwargs)
 

py/metadata_collector/metadata_processor.py

@@ -1,15 +1,68 @@
 import json
+import logging
 import os
 from .constants import IMAGES
 
 # Check if running in standalone mode
 standalone_mode = os.environ.get("LORA_MANAGER_STANDALONE", "0") == "1" or os.environ.get("HF_HUB_DISABLE_TELEMETRY", "0") == "0"
 
-from .constants import MODELS, PROMPTS, SAMPLING, LORAS, SIZE, IS_SAMPLER
+from .constants import MODELS, PROMPTS, SAMPLING, LORAS, SIZE, IS_SAMPLER, OVERWRITE
+from .node_extractors import NODE_EXTRACTORS
+
+logger = logging.getLogger(__name__)
+
+# Keys that identify metadata hint marks stored in node.properties.lm_marker_role
+_META_MARK_PREFIX = "meta_"
+_MARK_PRIMARY_MODEL = "primary_model"
+_MARK_PRIMARY_SAMPLER = "primary_sampler"
+_MARK_POSITIVE_PROMPT = "positive_prompt"
+_MARK_NEGATIVE_PROMPT = "negative_prompt"
 
 class MetadataProcessor:
     """Process and format collected metadata"""
-    
+
+    @staticmethod
+    def _get_user_marks(metadata):
+        """Scan workflow nodes (from extra_data.extra_pnginfo.workflow) for user-assigned
+        metadata hint marks stored in node.properties.lm_marker_role.
+
+        Returns a dict mapping mark type keys to node IDs.
+        Example: {'primary_model': '42', 'primary_sampler': '17'}
+        """
+        marks: dict[str, str] = {}
+
+        # Primary source: extra_data.extra_pnginfo.workflow.nodes (has full properties)
+        extra_data = metadata.get("extra_data")
+        if extra_data and isinstance(extra_data, dict):
+            extra_pnginfo = extra_data.get("extra_pnginfo", {})
+            if isinstance(extra_pnginfo, dict):
+                workflow = extra_pnginfo.get("workflow", {})
+                nodes = workflow.get("nodes", [])
+                for node in nodes:
+                    node_id = str(node.get("id", ""))
+                    role = node.get("properties", {}).get("lm_marker_role", "")
+                    if role.startswith(_META_MARK_PREFIX):
+                        mark_type = role[len(_META_MARK_PREFIX):]
+                        if mark_type in marks:
+                            logger.warning(
+                                "Duplicate meta hint '%s': node %s (previous: %s), "
+                                "last match wins",
+                                mark_type, node_id, marks[mark_type],
+                            )
+                        marks[mark_type] = node_id
+
+        # Fallback: try prompt.original_prompt (API-only submissions may not have workflow)
+        if not marks:
+            prompt = metadata.get("current_prompt")
+            if prompt and getattr(prompt, "original_prompt", None):
+                for node_id, node_data in prompt.original_prompt.items():
+                    role = node_data.get("properties", {}).get("lm_marker_role", "")
+                    if role.startswith(_META_MARK_PREFIX):
+                        mark_type = role[len(_META_MARK_PREFIX):]
+                        marks[mark_type] = node_id
+
+        return marks
+
     @staticmethod
     def find_primary_sampler(metadata, downstream_id=None):
         """
@@ -471,20 +524,57 @@ class MetadataProcessor:
             "checkpoint": None,
             "loras": "",
             "size": None,
-            "clip_skip": None
+            "clip_skip": None,
+            "additional_data": "",
         }
         
         # Get the prompt object for node relationship tracing
         prompt = metadata.get("current_prompt")
-        
-        # Find the primary KSampler node
-        primary_sampler_id, primary_sampler = MetadataProcessor.find_primary_sampler(metadata, id)
-        
-        # Directly get checkpoint from metadata instead of tracing
-        # Pass primary_sampler_id to avoid redundant calculation
-        checkpoint = MetadataProcessor.find_primary_checkpoint(metadata, id, primary_sampler_id)
-        if checkpoint:
-            params["checkpoint"] = checkpoint
+
+        # ---- User marks: override heuristic inference with user-assigned hints ----
+        user_marks = MetadataProcessor._get_user_marks(metadata)
+
+        # Find the primary KSampler node (user mark takes priority)
+        primary_sampler_id = None
+        primary_sampler = None
+        if _MARK_PRIMARY_SAMPLER in user_marks:
+            marked_id = user_marks[_MARK_PRIMARY_SAMPLER]
+            sampler_data = metadata.get(SAMPLING, {}).get(marked_id)
+            if sampler_data and sampler_data.get(IS_SAMPLER):
+                primary_sampler_id = marked_id
+                primary_sampler = sampler_data
+            else:
+                logger.warning(
+                    "User-marked primary sampler %s has no runtime metadata, "
+                    "falling back to heuristic",
+                    marked_id,
+                )
+        if primary_sampler is None:
+            primary_sampler_id, primary_sampler = MetadataProcessor.find_primary_sampler(metadata, id)
+
+        # Resolve checkpoint / model (user mark takes priority)
+        if _MARK_PRIMARY_MODEL in user_marks:
+            marked_id = user_marks[_MARK_PRIMARY_MODEL]
+            if marked_id in metadata.get(MODELS, {}):
+                params["checkpoint"] = metadata[MODELS][marked_id].get("name")
+            else:
+                extra_data = metadata.get("extra_data")
+                extra_pnginfo = extra_data.get("extra_pnginfo", {}) if extra_data and isinstance(extra_data, dict) else {}
+                workflow = extra_pnginfo.get("workflow", {}) if isinstance(extra_pnginfo, dict) else {}
+                node_type = "unknown"
+                for n in workflow.get("nodes", []):
+                    if str(n.get("id", "")) == marked_id:
+                        node_type = n.get("type", "unknown")
+                        break
+                logger.warning(
+                    "User-marked primary model %s (type=%s, registered=%s) has no runtime metadata, "
+                    "falling back to heuristic",
+                    marked_id, node_type, node_type in NODE_EXTRACTORS,
+                )
+        if params["checkpoint"] is None:
+            checkpoint = MetadataProcessor.find_primary_checkpoint(metadata, id, primary_sampler_id)
+            if checkpoint:
+                params["checkpoint"] = checkpoint
         
         # Check if guidance parameter exists in any sampling node
         for node_id, sampler_info in metadata.get(SAMPLING, {}).items():
@@ -539,7 +629,22 @@ class MetadataProcessor:
                     
                     # For SamplerCustom, handle any additional parameters
                     MetadataProcessor.handle_custom_advanced_sampler(metadata, prompt, primary_sampler_id, params)
-                
+
+            # ---- User marks: override prompts with explicitly tagged nodes ----
+            prompts_data = metadata.get(PROMPTS, {})
+            if _MARK_POSITIVE_PROMPT in user_marks:
+                pos_id = user_marks[_MARK_POSITIVE_PROMPT]
+                if pos_id in prompts_data:
+                    prompt_text = prompts_data[pos_id].get("text") or prompts_data[pos_id].get("positive_text")
+                    if prompt_text:
+                        params["prompt"] = prompt_text
+            if _MARK_NEGATIVE_PROMPT in user_marks:
+                neg_id = user_marks[_MARK_NEGATIVE_PROMPT]
+                if neg_id in prompts_data:
+                    prompt_text = prompts_data[neg_id].get("text") or prompts_data[neg_id].get("negative_text")
+                    if prompt_text:
+                        params["negative_prompt"] = prompt_text
+
             # Size extraction is same for all sampler types
             # Check if the sampler itself has size information (from latent_image)
             if primary_sampler_id in metadata.get(SIZE, {}):
@@ -568,7 +673,26 @@ class MetadataProcessor:
                 break
         if params["clip_skip"] is None:
             params["clip_skip"] = "1"
-        
+
+        # ---- Apply manual metadata overwrites ----
+        for overwrite_info in metadata.get(OVERWRITE, {}).values():
+            overwrite_params = overwrite_info.get("parameters", {})
+            for key, value in overwrite_params.items():
+                if key == "clip_skip":
+                    # Accept any value from overwrite node (sentinel -25 already
+                    # filtered upstream).  Needed because falsy check treats 0
+                    # as "not set" even though 0 is a valid wired input here.
+                    params[key] = value
+                elif value:  # truthy check — only overwrite when user provided a real value
+                    params[key] = value
+
+        # Bridge: the overwrite node exposes the field as "model" (more accurate),
+        # but the internal pipeline key remains "checkpoint" for backward compatibility
+        # with A1111 metadata format and downstream consumers.
+        if params.get("model"):
+            params["checkpoint"] = params["model"]
+            del params["model"]
+
         return params
     
     @staticmethod

py/metadata_collector/metadata_registry.py

@@ -1,7 +1,7 @@
 import time
 from nodes import NODE_CLASS_MAPPINGS  # type: ignore
 from .node_extractors import NODE_EXTRACTORS, GenericNodeExtractor
-from .constants import METADATA_CATEGORIES, IMAGES
+from .constants import METADATA_CATEGORIES, IMAGES, OVERWRITE
 
 
 class MetadataRegistry:
@@ -61,6 +61,7 @@ class MetadataRegistry:
             {
                 "execution_order": [],
                 "current_prompt": None,  # Will store the prompt object
+                "extra_data": None,  # Will store the API extra_data for workflow metadata
                 "timestamp": time.time(),
             }
         )
@@ -75,6 +76,11 @@ class MetadataRegistry:
             # Store the prompt in the metadata for later relationship tracing
             self.prompt_metadata[self.current_prompt_id]["current_prompt"] = prompt
 
+    def set_extra_data(self, extra_data):
+        """Store the API extra_data (contains extra_pnginfo.workflow with node properties)"""
+        if self.current_prompt_id and self.current_prompt_id in self.prompt_metadata:
+            self.prompt_metadata[self.current_prompt_id]["extra_data"] = extra_data
+
     def get_metadata(self, prompt_id=None):
         """Get collected metadata for a prompt"""
         key = prompt_id if prompt_id is not None else self.current_prompt_id
@@ -122,20 +128,28 @@ class MetadataRegistry:
             cache_key = f"{node_id}:{class_type}"
 
             # Check if this node type is relevant for metadata collection
-            if class_type in NODE_EXTRACTORS:
+            if class_type in NODE_EXTRACTORS or cache_key in self.node_cache:
                 # Check if we have cached metadata for this node
                 if cache_key in self.node_cache:
                     cached_data = self.node_cache[cache_key]
 
+                    # Detect bypass (mode=4) / mute (mode=2) — these nodes
+                    # were intentionally disabled and should not contribute
+                    # overwrite values from a previous execution's cache.
+                    node_mode = node_data.get("mode", 0)
+                    node_is_disabled = node_mode in (2, 4)
+
                     # Apply cached metadata to the current metadata
                     for category in self.metadata_categories:
+                        if category == OVERWRITE and node_is_disabled:
+                            continue
                         if category in cached_data and node_id in cached_data[category]:
                             if node_id not in metadata[category]:
                                 metadata[category][node_id] = cached_data[category][
                                     node_id
                                 ]
 
-    def record_node_execution(self, node_id, class_type, inputs, outputs):
+    def record_node_execution(self, node_id, class_type, inputs, outputs, return_types=None):
         """Record information about a node's execution"""
         if not self.current_prompt_id:
             return
@@ -158,17 +172,18 @@ class MetadataRegistry:
 
         # Extract node-specific metadata
         extractor = NODE_EXTRACTORS.get(class_type, GenericNodeExtractor)
-        extractor.extract(
-            node_id,
-            processed_inputs,
-            outputs,
-            self.prompt_metadata[self.current_prompt_id],
-        )
+        if extractor is GenericNodeExtractor:
+            extractor.extract(node_id, processed_inputs, outputs,
+                              self.prompt_metadata[self.current_prompt_id],
+                              return_types=return_types)
+        else:
+            extractor.extract(node_id, processed_inputs, outputs,
+                              self.prompt_metadata[self.current_prompt_id])
 
         # Cache this node's metadata
         self._cache_node_metadata(node_id, class_type)
 
-    def update_node_execution(self, node_id, class_type, outputs):
+    def update_node_execution(self, node_id, class_type, outputs, return_types=None):
         """Update node metadata with output information"""
         if not self.current_prompt_id:
             return
@@ -179,9 +194,17 @@ class MetadataRegistry:
         # Use the same extractor to update with outputs
         extractor = NODE_EXTRACTORS.get(class_type, GenericNodeExtractor)
         if hasattr(extractor, "update"):
-            extractor.update(
-                node_id, processed_outputs, self.prompt_metadata[self.current_prompt_id]
-            )
+            if extractor is GenericNodeExtractor:
+                extractor.update(
+                    node_id, processed_outputs,
+                    self.prompt_metadata[self.current_prompt_id],
+                    return_types=return_types,
+                )
+            else:
+                extractor.update(
+                    node_id, processed_outputs,
+                    self.prompt_metadata[self.current_prompt_id],
+                )
 
         # Update the cached metadata for this node
         self._cache_node_metadata(node_id, class_type)

py/metadata_collector/node_extractors.py

@@ -2,7 +2,7 @@ import json
 import os
 import re
 
-from .constants import MODELS, PROMPTS, SAMPLING, LORAS, SIZE, IMAGES, IS_SAMPLER
+from .constants import CLIP_SKIP_SENTINEL, MODELS, PROMPTS, SAMPLING, LORAS, SIZE, IMAGES, IS_SAMPLER, OVERWRITE, METADATA_OVERWRITE_FIELDS
 
 
 def _store_checkpoint_metadata(metadata, node_id, model_name):
@@ -31,11 +31,78 @@ class NodeMetadataExtractor:
         pass
         
 class GenericNodeExtractor(NodeMetadataExtractor):
-    """Default extractor for nodes without specific handling"""
+    """Fallback extractor with type-signature-based detection.
+
+    When a node is not in the NODE_EXTRACTORS registry, the hook layer
+    passes ``return_types`` from ``obj.RETURN_TYPES``:
+
+    * ``MODEL`` output: common input fields (ckpt_name, unet_name, etc.)
+      are checked for a model file name and stored as checkpoint metadata.
+    * ``CONDITIONING`` output: common text input fields are checked for
+      prompt text and stored as prompt metadata.
+    """
+
+    # Input field names that carry a model path in loader-style nodes.
+    _MODEL_NAME_FIELDS = (
+        "ckpt_name", "unet_name", "model_path", "model_name", "gguf_name",
+    )
+
+    # Extensions used by checkpoint_scanner.py — only record values that look
+    # like real model filenames to avoid capturing unrelated string fields.
+    _MODEL_EXTENSIONS = {
+        ".ckpt", ".pt", ".pt2", ".bin", ".pth", ".safetensors", ".pkl", ".sft", ".gguf",
+    }
+
+    # Input field names that may carry prompt text in encoder-style nodes.
+    _TEXT_FIELDS = ("text", "clip_l", "t5xxl", "prompt", "positive", "negative")
+
     @staticmethod
-    def extract(node_id, inputs, outputs, metadata):
-        pass
-        
+    def extract(node_id, inputs, outputs, metadata, return_types=None):
+        if return_types is None:
+            return
+
+        # — MODEL loader detection (checkpoint / UNET / GGUF) —
+        if "MODEL" in return_types or any("MODEL" in str(t) for t in return_types):
+            for field in GenericNodeExtractor._MODEL_NAME_FIELDS:
+                val = inputs.get(field)
+                if val and isinstance(val, str) and val.strip():
+                    name = val.strip()
+                    if not any(name.lower().endswith(ext) for ext in GenericNodeExtractor._MODEL_EXTENSIONS):
+                        continue
+                    _store_checkpoint_metadata(metadata, node_id, name)
+                    return
+
+        # — CONDITIONING encoder detection (CLIPTextEncode, Flux, custom) —
+        if "CONDITIONING" in return_types or any("CONDITIONING" in str(t) for t in return_types):
+            text = None
+            for field in GenericNodeExtractor._TEXT_FIELDS:
+                val = inputs.get(field)
+                if val and isinstance(val, str) and val.strip():
+                    text = val.strip()
+                    break
+            if text:
+                prompt_data = metadata.setdefault(PROMPTS, {})
+                prompt_data[node_id] = {
+                    "text": text,
+                    "node_id": node_id,
+                }
+
+    @staticmethod
+    def update(node_id, outputs, metadata, return_types=None):
+        if return_types is None:
+            return
+        if "CONDITIONING" not in return_types and not any(
+            "CONDITIONING" in str(t) for t in return_types
+        ):
+            return
+        if node_id not in metadata.get(PROMPTS, {}):
+            return
+        if outputs and isinstance(outputs, list) and len(outputs) > 0:
+            if isinstance(outputs[0], tuple) and len(outputs[0]) > 0:
+                cond = outputs[0][0]
+                if cond is not None:
+                    metadata[PROMPTS][node_id]["conditioning"] = cond
+
 class CheckpointLoaderExtractor(NodeMetadataExtractor):
     @staticmethod
     def extract(node_id, inputs, outputs, metadata):
@@ -901,6 +968,55 @@ class LoraLoaderManagerExtractor(NodeMetadataExtractor):
                 "node_id": node_id
             }
 
+class LoraTextLoaderManagerExtractor(NodeMetadataExtractor):
+    """Extract LoRA metadata from LoraTextLoaderLM (LoRA Text Loader).
+
+    The node accepts a `lora_syntax` STRING containing <lora:name:strength> tags
+    (same format as the ComfyUI prompt), plus an optional `lora_stack`.
+    This extractor parses the syntax string using the same regex as the node.
+    """
+    @staticmethod
+    def extract(node_id, inputs, outputs, metadata):
+        if not inputs:
+            return
+
+        active_loras = []
+
+        # Process lora_stack if available (optional input)
+        if "lora_stack" in inputs:
+            lora_stack = inputs.get("lora_stack", [])
+            for item in lora_stack:
+                # lora_stack entries are (path, model_strength, clip_strength) tuples
+                if isinstance(item, (list, tuple)) and len(item) >= 2:
+                    lora_path = item[0]
+                    model_strength = item[1]
+                    lora_name = os.path.splitext(os.path.basename(lora_path))[0]
+                    active_loras.append({
+                        "name": lora_name,
+                        "strength": round(float(model_strength), 2)
+                    })
+
+        # Process lora_syntax string input
+        if "lora_syntax" in inputs:
+            lora_syntax = inputs.get("lora_syntax", "")
+            if lora_syntax and isinstance(lora_syntax, str):
+                pattern = r"<lora:([^:>]+):([^:>]+)(?::([^:>]+))?>"
+                matches = re.findall(pattern, lora_syntax, re.IGNORECASE)
+                for match in matches:
+                    lora_name = match[0]
+                    model_strength = float(match[1])
+                    active_loras.append({
+                        "name": lora_name,
+                        "strength": round(model_strength, 2)
+                    })
+
+        if active_loras:
+            metadata[LORAS][node_id] = {
+                "lora_list": active_loras,
+                "node_id": node_id
+            }
+
+
 class FluxGuidanceExtractor(NodeMetadataExtractor):
     @staticmethod
     def extract(node_id, inputs, outputs, metadata):
@@ -1105,6 +1221,35 @@ class CR_ApplyControlNetStackExtractor(NodeMetadataExtractor):
                 metadata[PROMPTS][node_id]["positive_encoded"] = transformed_positive
                 metadata[PROMPTS][node_id]["negative_encoded"] = transformed_negative
 
+class MetadataOverwriteExtractor(NodeMetadataExtractor):
+    """Extract manually specified metadata from MetadataOverwriteLM node.
+
+    Stores truthy input values under the OVERWRITE category so that
+    extract_generation_params can merge them over the inferred params.
+    """
+
+    @staticmethod
+    def extract(node_id, inputs, outputs, metadata):
+        if not inputs:
+            return
+
+        overwrite_params = {}
+        for key in METADATA_OVERWRITE_FIELDS:
+            value = inputs.get(key)
+            if key == "clip_skip":
+                if value != CLIP_SKIP_SENTINEL:
+                    overwrite_params[key] = value
+            elif value:  # truthy — only overwrite when user provided a real value
+                overwrite_params[key] = value
+
+        if overwrite_params:
+            metadata.setdefault(OVERWRITE, {})
+            metadata[OVERWRITE][node_id] = {
+                "parameters": overwrite_params,
+                "node_id": node_id,
+            }
+
+
 # Registry of node-specific extractors
 # Keys are node class names
 NODE_EXTRACTORS = {
@@ -1146,6 +1291,7 @@ NODE_EXTRACTORS = {
     "UNETLoaderLM": UNETLoaderExtractor,  # LoRA Manager
     "LoraLoader": LoraLoaderExtractor,
     "LoraLoaderLM": LoraLoaderManagerExtractor,
+    "LoraTextLoaderLM": LoraTextLoaderManagerExtractor,
     "RgthreePowerLoraLoader": RgthreePowerLoraLoaderExtractor,
     "TensorRTLoader": TensorRTLoaderExtractor,
     # Conditioning
@@ -1171,5 +1317,7 @@ NODE_EXTRACTORS = {
     "CFGGuider": CFGGuiderExtractor,            # Add CFGGuider
     # Image
     "VAEDecode": VAEDecodeExtractor,  # Added VAEDecode extractor
+    # Metadata overwrite
+    "MetadataOverwriteLM": MetadataOverwriteExtractor,
     # Add other nodes as needed
 }

py/metadata_ops/__init__.py

@@ -0,0 +1,233 @@
+"""Metadata operations — thin in-process wrappers around LoRA Manager internal services.
+
+All functions are simple Python async functions that delegate to the
+appropriate internal service.  They use **relative imports** within the
+``py`` package, so ``sys.modules`` caching works normally and there is no
+risk of double import or circular dependencies.
+
+Usage (in-process, primary)::
+
+    from py.metadata_ops import list_base_models, read_metadata
+
+    models = await list_base_models()
+    meta   = await read_metadata("/path/to/model.safetensors")
+
+Usage (subprocess, debugging / external)::
+
+    python -m py.metadata_ops base-models list
+    python -m py.metadata_ops metadata read /path/to/model.safetensors
+"""
+
+from __future__ import annotations
+
+import asyncio
+import logging
+import os
+from typing import Any, Dict, List, Optional
+
+logger = logging.getLogger(__name__)
+
+
+# ---------------------------------------------------------------------------
+# Helpers
+# ---------------------------------------------------------------------------
+
+SCANNER_TYPE_MAP: dict[str, str] = {
+    "get_lora_scanner": "lora",
+    "get_checkpoint_scanner": "checkpoint",
+    "get_embedding_scanner": "embedding",
+}
+
+SCANNER_GETTER_NAMES = tuple(SCANNER_TYPE_MAP.keys())
+
+
+async def _find_model_entry(
+    model_path: str,
+) -> tuple[object, object, str | None] | tuple[None, None, None]:
+    """Iterate all scanners and return the first (scanner, entry, getter_name)
+    that owns *model_path*.  Returns ``(None, None, None)`` when no scanner
+    claims it.
+    """
+    from ..services.service_registry import ServiceRegistry
+
+    normalized = os.path.normpath(model_path)
+    for getter_name in SCANNER_GETTER_NAMES:
+        getter = getattr(ServiceRegistry, getter_name, None)
+        if getter is None:
+            continue
+        try:
+            scanner = await getter()
+            if scanner is None:
+                continue
+            cache = await scanner.get_cached_data()
+            for entry in cache.raw_data:
+                if os.path.normpath(entry.get("file_path", "")) == normalized:
+                    return scanner, entry, getter_name
+        except Exception as exc:
+            logger.debug(
+                "Scanner %s check failed for %s: %s",
+                getter_name, model_path, exc,
+            )
+    return None, None, None
+
+
+async def _find_scanner_for_model(
+    model_path: str,
+) -> tuple[object, object] | tuple[None, None]:
+    """Find the (scanner, cache_entry) responsible for *model_path*."""
+    scanner, entry, _ = await _find_model_entry(model_path)
+    return scanner, entry
+
+
+async def identify_model_type(model_path: str) -> str:
+    """Determine the model type (``\"lora\"``, ``\"checkpoint\"``, or
+    ``\"embedding\"``) for *model_path*.
+
+    Falls back to ``\"lora\"`` when unknown.
+    """
+    _, _, getter_name = await _find_model_entry(model_path)
+    return SCANNER_TYPE_MAP[getter_name] if getter_name else "lora"
+
+
+# ---------------------------------------------------------------------------
+# Public API
+# ---------------------------------------------------------------------------
+
+
+async def list_base_models(limit: int = 0) -> List[str]:
+    """Return all valid CivitAI base model names.
+
+    Uses ``CivitaiBaseModelService.get_base_models()`` which merges a
+    hardcoded list (``SUPPORTED_DOWNLOAD_SKIP_BASE_MODELS``) with remote
+    models fetched from the CivitAI API.  Never empty — the hardcoded
+    fallback always provides a complete set.
+
+    The result is sorted alphabetically.  Pass *limit* = 0 for all models.
+    """
+    from ..services.civitai_base_model_service import (
+        CivitaiBaseModelService,
+    )
+
+    try:
+        service = await CivitaiBaseModelService.get_instance()
+        response = await service.get_base_models()
+        names: List[str] = response.get("models", [])
+    except Exception as exc:
+        logger.warning("list_base_models failed: %s", exc)
+        names = []
+    if limit > 0:
+        return names[:limit]
+    return names
+
+
+async def read_metadata(model_path: str) -> Dict[str, Any]:
+    """Load the full metadata payload for *model_path* from disk.
+
+    Returns an empty dict when the metadata file does not exist or cannot
+    be parsed — never raises.
+    """
+    from ..utils.metadata_manager import MetadataManager
+
+    try:
+        return await MetadataManager.load_metadata_payload(model_path) or {}
+    except Exception as exc:
+        logger.warning("read_metadata failed for %s: %s", model_path, exc)
+        return {}
+
+
+async def apply_metadata_updates(
+    model_path: str,
+    updates: Dict[str, Any],
+) -> List[str]:
+    """Merge *updates* into the model's on-disk metadata and persist.
+
+    Returns the list of field names that actually changed.
+    """
+    from ..utils.metadata_manager import MetadataManager
+
+    metadata = await read_metadata(model_path)
+    updated_fields: List[str] = []
+    for key, value in updates.items():
+        old = metadata.get(key)
+        if old != value:
+            metadata[key] = value
+            updated_fields.append(key)
+    if updated_fields:
+        await MetadataManager.save_metadata(model_path, metadata)
+    return updated_fields
+
+
+async def download_preview(
+    model_path: str,
+    url: str,
+    *,
+    target_width: int = 480,
+    quality: int = 85,
+) -> str | None:
+    """Download a preview image from *url*, optimise to .webp, and save it.
+
+    The output file is placed alongside the model file with a ``.webp``
+    extension.  Returns the local file path on success, ``None`` on failure.
+    """
+    from ..services.downloader import get_downloader
+    from ..utils.exif_utils import ExifUtils
+
+    if not url or not url.strip():
+        return None
+
+    base_name = os.path.splitext(os.path.basename(model_path))[0]
+    preview_dir = os.path.dirname(model_path)
+    output_path = os.path.join(preview_dir, base_name + ".webp")
+
+    downloader = await get_downloader()
+
+    # Try in-memory download + optimise first
+    success, content, _headers = await downloader.download_to_memory(
+        url, use_auth=False,
+    )
+    if success and content:
+        try:
+            optimized_data, _ = ExifUtils.optimize_image(
+                image_data=content,
+                target_width=target_width,
+                format="webp",
+                quality=quality,
+                preserve_metadata=False,
+            )
+            with open(output_path, "wb") as f:
+                f.write(optimized_data)
+            return output_path
+        except Exception as exc:
+            logger.warning("Preview optimisation failed, saving raw: %s", exc)
+            # Fall through to raw save
+
+    # Fallback: download directly to file
+    try:
+        ok, _ = await downloader.download_file(url, output_path, use_auth=False)
+        if ok:
+            return output_path
+    except Exception as exc:
+        logger.warning("Preview fallback download failed for %s: %s", model_path, exc)
+
+    return None
+
+
+async def refresh_cache(model_path: str) -> bool:
+    """Invalidate and reload the scanner cache entry for *model_path*.
+
+    Returns ``True`` when the model was found and the cache was refreshed.
+    """
+    scanner, entry = await _find_scanner_for_model(model_path)
+    if scanner is None:
+        logger.warning("refresh_cache: no scanner found for %s", model_path)
+        return False
+    try:
+        metadata = await read_metadata(model_path)
+        if not metadata:
+            logger.warning("refresh_cache: no metadata for %s", model_path)
+            return False
+        await scanner.update_single_model_cache(model_path, model_path, metadata)
+        return True
+    except Exception as exc:
+        logger.warning("refresh_cache failed for %s: %s", model_path, exc)
+        return False

py/metadata_ops/__main__.py

@@ -0,0 +1,113 @@
+"""Subprocess entry point for ``metadata_ops`` (debugging / external use).
+
+Usage::
+
+    python -m py.metadata_ops base-models list [--limit N]
+    python -m py.metadata_ops metadata read <path>
+    python -m py.metadata_ops metadata update <path> --json '{...}'
+    python -m py.metadata_ops preview download <path> --url <url>
+    python -m py.metadata_ops cache refresh <path>
+"""
+
+from __future__ import annotations
+
+import argparse
+import asyncio
+import json
+import sys
+from typing import Any, Dict, List
+
+
+def _build_parser() -> argparse.ArgumentParser:
+    parser = argparse.ArgumentParser(prog="lmcli", description="LoRA Manager Agent CLI")
+    sub = parser.add_subparsers(dest="command", required=True)
+
+    # base-models list
+    base_models = sub.add_parser("base-models", aliases=["bm"])
+    base_models_cmds = base_models.add_subparsers(dest="subcommand", required=True)
+    base_models_list = base_models_cmds.add_parser("list")
+    base_models_list.add_argument(
+        "--limit", type=int, default=0, help="Max number of models (0 = all)"
+    )
+
+    # metadata read
+    meta = sub.add_parser("metadata", aliases=["md"])
+    meta_cmds = meta.add_subparsers(dest="subcommand", required=True)
+    meta_read = meta_cmds.add_parser("read")
+    meta_read.add_argument("path", type=str, help="Model file path")
+
+    # metadata update
+    meta_update = meta_cmds.add_parser("update")
+    meta_update.add_argument("path", type=str, help="Model file path")
+    meta_update.add_argument(
+        "--json",
+        type=str,
+        required=True,
+        help='JSON object of fields to update, e.g. \'{"base_model": "SDXL 1.0"}\'',
+    )
+
+    # preview download
+    prev = sub.add_parser("preview", aliases=["pv"])
+    prev_cmds = prev.add_subparsers(dest="subcommand", required=True)
+    prev_dl = prev_cmds.add_parser("download")
+    prev_dl.add_argument("path", type=str, help="Model file path")
+    prev_dl.add_argument("--url", type=str, required=True, help="Preview image URL")
+
+    # cache refresh
+    cache = sub.add_parser("cache")
+    cache_cmds = cache.add_subparsers(dest="subcommand", required=True)
+    cache_refresh = cache_cmds.add_parser("refresh")
+    cache_refresh.add_argument("path", type=str, help="Model file path")
+
+    return parser
+
+
+async def _run(args: argparse.Namespace) -> Any:
+    from . import (  # lazy import so startup is fast
+        list_base_models,
+        read_metadata,
+        apply_metadata_updates,
+        download_preview,
+        refresh_cache,
+    )
+
+    cmd = args.command
+    sub = args.subcommand
+
+    if cmd in ("base-models", "bm") and sub == "list":
+        return await list_base_models(limit=args.limit)
+
+    if cmd in ("metadata", "md") and sub == "read":
+        return await read_metadata(args.path)
+
+    if cmd in ("metadata", "md") and sub == "update":
+        updates: Dict[str, Any] = json.loads(args.json)
+        return await apply_metadata_updates(args.path, updates)
+
+    if cmd in ("preview", "pv") and sub == "download":
+        return await download_preview(args.path, args.url)
+
+    if cmd == "cache" and sub == "refresh":
+        return await refresh_cache(args.path)
+
+    raise ValueError(f"Unknown command: {cmd} {sub}")
+
+
+def main() -> None:
+    parser = _build_parser()
+    args = parser.parse_args()
+
+    result = asyncio.run(_run(args))
+    # Always print as JSON so callers can parse reliably
+    if isinstance(result, list):
+        for item in result:
+            print(item)
+    elif isinstance(result, dict):
+        json.dump(result, sys.stdout, ensure_ascii=False, indent=2)
+        print()
+    else:
+        print(json.dumps(result))
+
+
+if __name__ == "__main__":
+    main()

py/middleware/cache_middleware.py

@@ -16,6 +16,8 @@ IMG_EXTENSIONS = (
     ".tif",
     ".tiff",
     ".webp",
+    ".avif",
+    ".jxl",
     ".mp4"
 )
 

py/middleware/error_middleware.py

@@ -41,7 +41,12 @@ async def api_json_error(
         if exc.status < 400:
             raise
 
-        logger.warning(
+        # Preview 404 is routine (file deleted from disk) — not worth a warning.
+        logger_method = logger.warning
+        if request.path.startswith("/api/lm/previews") and exc.status == 404:
+            logger_method = logger.debug
+
+        logger_method(
             "API %s %s returned HTTP %d: %s",
             request.method,
             request.path,

py/nodes/create_hook_lora.py

@@ -0,0 +1,117 @@
+"""Create Hook LoRA (LoraManager) — multi-LoRA hook node compatible with ComfyUI's built-in hook pipeline.
+
+Produces ``("HOOKS",)`` output that chains seamlessly with downstream hook consumers
+(ConditioningSetProperties, SetHookKeyframes, CombineHooks, SetClipHooks, etc.).
+"""
+
+from __future__ import annotations
+
+import logging
+import os
+
+from ..utils.utils import get_lora_info_absolute
+from .utils import (
+    FlexibleOptionalInputType,
+    any_type,
+    apply_lora_syntax_format,
+    get_loras_list,
+)
+
+logger = logging.getLogger(__name__)
+
+
+class CreateHookLoraLM:
+    NAME = "Create Hook LoRA (LoraManager)"
+    CATEGORY = "Lora Manager/hooks"
+
+    @classmethod
+    def INPUT_TYPES(cls):
+        return {
+            "required": {
+                "text": (
+                    "AUTOCOMPLETE_TEXT_LORAS",
+                    {
+                        "placeholder": "Search LoRAs to add...",
+                        "tooltip": (
+                            "Search and select LoRAs. Each LoRA gets its own "
+                            "model/clip strength. Hooks chain with prev_hooks."
+                        ),
+                    },
+                ),
+            },
+            "optional": FlexibleOptionalInputType(any_type),
+        }
+
+    RETURN_TYPES = ("HOOKS", "STRING", "STRING")
+    RETURN_NAMES = ("HOOKS", "trigger_words", "active_loras")
+    FUNCTION = "create_hook"
+
+    def create_hook(self, text: str, **kwargs):
+        """Create a HookGroup from the selected LoRAs, chained with prev_hooks.
+
+        Each active LoRA from the widget is loaded and wrapped in a WeightHook
+        via :func:`comfy.hooks.create_hook_lora`.  All hooks are combined into a
+        single group and returned alongside trigger words and a human-readable
+        summary of the active LoRAs.
+        """
+        del text  # used by the frontend widget only
+
+        # Lazy imports: comfy is not available in CI/test environment at module level
+        import comfy.hooks  # type: ignore  # noqa: C0415
+        import comfy.utils  # type: ignore  # noqa: C0415
+
+        prev_hooks: comfy.hooks.HookGroup | None = kwargs.get("prev_hooks")
+
+        hook_group = prev_hooks.clone() if prev_hooks is not None else comfy.hooks.HookGroup()
+
+        all_trigger_words: list[str] = []
+        active_loras: list[tuple[str, float, float]] = []
+
+        for lora in get_loras_list(kwargs):
+            if not lora.get("active", False):
+                continue
+
+            lora_name = apply_lora_syntax_format(lora["name"])
+            model_strength = float(lora["strength"])
+            clip_strength = float(lora.get("clipStrength", model_strength))
+
+            # Skip useless no-op entries (both strengths are zero)
+            if model_strength == 0.0 and clip_strength == 0.0:
+                continue
+
+            lora_path, trigger_words = get_lora_info_absolute(lora_name)
+            if not lora_path or not os.path.isfile(lora_path):
+                logger.warning("LoRA '%s' not found — skipping", lora_name)
+                continue
+
+            try:
+                lora_weights = comfy.utils.load_torch_file(lora_path, safe_load=True)
+
+                lora_hooks = comfy.hooks.create_hook_lora(
+                    lora=lora_weights,
+                    strength_model=model_strength,
+                    strength_clip=clip_strength,
+                )
+            except Exception:
+                logger.exception("Failed to load LoRA '%s' — skipping", lora_name)
+                continue
+            hook_group = hook_group.clone_and_combine(lora_hooks)
+
+            active_loras.append((lora_name, model_strength, clip_strength))
+            all_trigger_words.extend(trigger_words)
+
+        # Format trigger words (group mode separator)
+        trigger_words_text = ",, ".join(all_trigger_words) if all_trigger_words else ""
+
+        # Format active LoRAs summary
+        formatted_loras = []
+        for name, model_s, clip_s in active_loras:
+            if abs(model_s - clip_s) > 0.001:
+                formatted_loras.append(
+                    f"<lora:{name}:{model_s}:{clip_s}>"
+                )
+            else:
+                formatted_loras.append(f"<lora:{name}:{model_s}>")
+        active_loras_text = " ".join(formatted_loras)
+
+        return (hook_group, trigger_words_text, active_loras_text)

py/nodes/lora_info.py

@@ -0,0 +1,45 @@
+"""Lora Info display node — pure frontend node for showing selected LoRA info.
+
+This node does NOT participate in workflow execution. Its single optional
+"lora_source" input exists solely as a wire-connection anchor so that the
+frontend can traverse the graph and push selection data to connected info nodes.
+"""
+
+from __future__ import annotations
+
+
+class LoraInfoLM:
+    """Display node that shows filename and notes for the selected LoRA."""
+
+    NAME = "Lora Info (LoraManager)"
+    CATEGORY = "Lora Manager/utils"
+    DESCRIPTION = (
+        "Displays information (filename, notes) about the currently selected "
+        "LoRA. Connect any output from a LoRA Loader or Stacker to the "
+        "lora_source input, then select a LoRA in the source widget — the "
+        "info updates automatically. Does not affect workflow execution."
+    )
+
+    @classmethod
+    def INPUT_TYPES(cls):
+        return {
+            "required": {},
+        }
+
+    RETURN_TYPES = ()
+    RETURN_NAMES = ()
+    OUTPUT_NODE = False
+    FUNCTION = "noop"
+
+    def noop(self, **kwargs):
+        # This node is display-only — no workflow execution needed.
+        return ()
+
+
+NODE_CLASS_MAPPINGS = {
+    LoraInfoLM.NAME: LoraInfoLM,
+}
+
+NODE_DISPLAY_NAME_MAPPINGS = {
+    LoraInfoLM.NAME: "Lora Info (LoraManager)",
+}

py/nodes/lora_loader.py

@@ -1,6 +1,5 @@
 import importlib
 import logging
-import re
 
 import comfy.sd  # type: ignore
 import comfy.utils  # type: ignore
@@ -14,6 +13,7 @@ from .utils import (
     extract_lora_name,
     get_loras_list,
     nunchaku_load_lora,
+    parse_lora_syntax,
 )
 
 logger = logging.getLogger(__name__)
@@ -189,25 +189,10 @@ class LoraTextLoaderLM:
     RETURN_NAMES = ("MODEL", "CLIP", "trigger_words", "loaded_loras")
     FUNCTION = "load_loras_from_text"
 
-    def parse_lora_syntax(self, text):
-        """Parse LoRA syntax from text input."""
-        pattern = r"<lora:([^:>]+):([^:>]+)(?::([^:>]+))?>"
-        matches = re.findall(pattern, text, re.IGNORECASE)
-
-        loras = []
-        for match in matches:
-            model_strength = float(match[1])
-            loras.append({
-                "name": match[0],
-                "model_strength": model_strength,
-                "clip_strength": float(match[2]) if match[2] else model_strength,
-            })
-        return loras
-
     def load_loras_from_text(self, model, lora_syntax, clip=None, lora_stack=None):
         """Load LoRAs based on text syntax input."""
         lora_entries = _collect_stack_entries(lora_stack)
-        for lora in self.parse_lora_syntax(lora_syntax):
+        for lora in parse_lora_syntax(lora_syntax):
             lora_path, trigger_words = get_lora_info_absolute(lora["name"])
             lora_entries.append({
                 "name": lora["name"],

py/nodes/lora_stack_combiner.py

@@ -1,26 +1,102 @@
+from __future__ import annotations
+
+import inspect
+import re
+from typing import Any
+
+_STACK_INPUT_PATTERN = re.compile(r"^lora_stack(?:_([ab])|(\d+))$")
+
+
+def _is_stack_input(name: str) -> bool:
+    return bool(_STACK_INPUT_PATTERN.match(name))
+
+
+def _stack_slot_number(name: str) -> int:
+    """Numeric slot used to order stack inputs; legacy a/b map to 1/2."""
+    match = _STACK_INPUT_PATTERN.match(name)
+    if not match:
+        return -1
+    letter, digits = match.group(1), match.group(2)
+    if digits is not None:
+        return int(digits)
+    return 1 if letter == "a" else 2
+
+
+class _LoraStackOptionalInputs:
+    """Lookup that preserves explicit optional inputs and dynamic lora_stack slots."""
+
+    def __init__(self, explicit_inputs: dict[str, tuple[str, dict[str, Any]]]) -> None:
+        self._explicit_inputs = explicit_inputs
+
+    def __contains__(self, item: object) -> bool:
+        if not isinstance(item, str):
+            return False
+        return item in self._explicit_inputs or _is_stack_input(item)
+
+    def __getitem__(self, key: str) -> tuple[str, dict[str, Any]]:
+        if key in self._explicit_inputs:
+            return self._explicit_inputs[key]
+        if _is_stack_input(key):
+            return (
+                "LORA_STACK",
+                {
+                    "tooltip": "A LoRA stack to combine. Connect to add more inputs.",
+                },
+            )
+        raise KeyError(key)
+
+
 class LoraStackCombinerLM:
     NAME = "Lora Stack Combiner (LoraManager)"
     CATEGORY = "Lora Manager/stackers"
+    DESCRIPTION = (
+        "Combines multiple LoRA stacks into a single stack. "
+        "Supports dynamic inputs: connect a stack to add more inputs."
+    )
 
     @classmethod
     def INPUT_TYPES(cls):
+        optional_inputs: dict[str, tuple[str, dict[str, Any]]] = {
+            "lora_stack1": (
+                "LORA_STACK",
+                {
+                    "tooltip": "A LoRA stack to combine. Connect to add more inputs.",
+                },
+            ),
+            "lora_stack2": (
+                "LORA_STACK",
+                {
+                    "tooltip": "A LoRA stack to combine. Connect to add more inputs.",
+                },
+            ),
+        }
+
+        stack = inspect.stack()
+        if len(stack) > 2 and stack[2].function == "get_input_info":
+            optional_inputs = _LoraStackOptionalInputs(optional_inputs)  # type: ignore[assignment]
+
         return {
-            "required": {
-                "lora_stack_a": ("LORA_STACK",),
-                "lora_stack_b": ("LORA_STACK",),
-            },
+            "required": {},
+            "optional": optional_inputs,
         }
 
     RETURN_TYPES = ("LORA_STACK",)
     RETURN_NAMES = ("LORA_STACK",)
     FUNCTION = "combine_stacks"
 
-    def combine_stacks(self, lora_stack_a, lora_stack_b):
-        combined_stack = []
+    def combine_stacks(self, lora_stack1=None, lora_stack2=None, **kwargs):
+        stacks = {
+            "lora_stack1": lora_stack1,
+            "lora_stack2": lora_stack2,
+        }
+        for key, value in kwargs.items():
+            if _is_stack_input(key) and value is not None:
+                stacks[key] = value
 
-        if lora_stack_a:
-            combined_stack.extend(lora_stack_a)
-        if lora_stack_b:
-            combined_stack.extend(lora_stack_b)
+        combined_stack = []
+        for key in sorted(stacks, key=_stack_slot_number):
+            stack = stacks[key]
+            if stack:
+                combined_stack.extend(stack)
 
         return (combined_stack,)

py/nodes/lora_syntax_to_path.py

@@ -0,0 +1,62 @@
+"""Node to resolve `<lora:name:strength>` syntax to absolute file system paths.
+
+Takes the loaded_loras / active_loras STRING output from LoraLoaderLM or
+LoraStackerLM and resolves each lora name to its absolute path on disk via
+the scanner cache. Unknown names are returned as-is.
+"""
+
+import logging
+
+from ..utils.utils import get_lora_info_absolute
+from .utils import parse_lora_syntax
+
+logger = logging.getLogger(__name__)
+
+
+class LoraSyntaxToPath:
+    NAME = "LoRA Syntax → Path (LoraManager)"
+    CATEGORY = "Lora Manager/utils"
+
+    @classmethod
+    def INPUT_TYPES(cls):
+        return {
+            "required": {
+                "lora_syntax": (
+                    "STRING",
+                    {
+                        "forceInput": True,
+                        "multiline": True,
+                        "tooltip": (
+                            "<lora:name:strength> formatted text from "
+                            "loaded_loras / active_loras output"
+                        ),
+                    },
+                ),
+            },
+        }
+
+    RETURN_TYPES = ("STRING",)
+    RETURN_NAMES = ("paths",)
+    FUNCTION = "resolve"
+
+    def resolve(self, lora_syntax: str) -> tuple[str]:
+        """Parse <lora:...> syntax and resolve each name to its absolute path."""
+        if not lora_syntax or not lora_syntax.strip():
+            logger.info("Received empty lora_syntax input")
+            return ("",)
+
+        parsed = parse_lora_syntax(lora_syntax)
+        if not parsed:
+            logger.info("No valid <lora:...> entries found in input")
+            return ("",)
+
+        paths: list[str] = []
+        for entry in parsed:
+            try:
+                absolute_path, _ = get_lora_info_absolute(entry["name"])
+                paths.append(absolute_path)
+            except Exception:
+                logger.warning("Failed to resolve lora '%s', skipping", entry["name"])
+                continue
+
+        return ("\n".join(paths),)

py/nodes/metadata_overwrite.py

@@ -0,0 +1,170 @@
+"""Metadata Overwrite node — allows users to manually specify generation parameters
+that override the automatically collected/inferred metadata.
+
+Most inputs have falsy defaults (empty string / 0) which are skipped.
+clip_skip uses a sentinel default (-25) so that a wired value of 0 is
+preserved — both ComfyUI and A1111 conventions have no meaningful 0 value,
+but users may wire 0 to express "no clip skip / default".
+"""
+
+from typing import Any
+
+from ..metadata_collector.constants import (
+    CLIP_SKIP_SENTINEL as _CLIP_SKIP_SENTINEL,
+    METADATA_OVERWRITE_FIELDS,
+)
+
+
+class MetadataOverwriteLM:
+    NAME = "Metadata Overwrite (LoraManager)"
+    CATEGORY = "Lora Manager/utils"
+    DESCRIPTION = (
+        "Manually specify generation parameters to override automatically collected "
+        "metadata. Only filled/connected inputs will take effect — empty defaults "
+        "are ignored."
+    )
+
+    @classmethod
+    def INPUT_TYPES(cls) -> dict[str, Any]:
+        return {
+            "optional": {
+                "prompt": (
+                    "STRING",
+                    {
+                        "default": "",
+                        "multiline": True,
+                        "tooltip": "Positive prompt. Only overwrites when non-empty.",
+                    },
+                ),
+                "negative_prompt": (
+                    "STRING",
+                    {
+                        "default": "",
+                        "multiline": True,
+                        "tooltip": "Negative prompt. Only overwrites when non-empty.",
+                    },
+                ),
+                "seed": (
+                    "INT",
+                    {
+                        "default": 0,
+                        "min": 0,
+                        "max": 0xFFFFFFFFFFFFFFFF,
+                        "control_after_generate": False,
+                        "tooltip": "Seed value. Only overwrites when > 0.",
+                    },
+                ),
+                "steps": (
+                    "INT",
+                    {
+                        "default": 0,
+                        "min": 0,
+                        "max": 10000,
+                        "tooltip": "Number of steps. Only overwrites when > 0.",
+                    },
+                ),
+                "cfg_scale": (
+                    "FLOAT",
+                    {
+                        "default": 0.0,
+                        "min": 0.0,
+                        "max": 100.0,
+                        "tooltip": "CFG scale. Only overwrites when > 0.",
+                    },
+                ),
+                "sampler": (
+                    "STRING",
+                    {
+                        "default": "",
+                        "tooltip": "Sampler name. Only overwrites when non-empty.",
+                    },
+                ),
+                "scheduler": (
+                    "STRING",
+                    {
+                        "default": "",
+                        "tooltip": "Scheduler name. Only overwrites when non-empty.",
+                    },
+                ),
+                "model": (
+                    "STRING",
+                    {
+                        "default": "",
+                        "tooltip": (
+                            "The checkpoint or diffusion model (UNet) used "
+                            "for generation. Only overwrites when non-empty."
+                        ),
+                    },
+                ),
+                "loras": (
+                    "STRING",
+                    {
+                        "default": "",
+                        "multiline": True,
+                        "tooltip": (
+                            "LoRA syntax, e.g. <lora:name:strength> "
+                            "or <lora:name:model_strength:clip_strength>, "
+                            "separated by spaces. Only overwrites when non-empty."
+                        ),
+                    },
+                ),
+                "size": (
+                    "STRING",
+                    {
+                        "default": "",
+                        "tooltip": (
+                            "Image size in WIDTHxHEIGHT format (e.g. 512x768). "
+                            "Only overwrites when non-empty."
+                        ),
+                    },
+                ),
+                "clip_skip": (
+                    "INT",
+                    {
+                        "default": _CLIP_SKIP_SENTINEL,
+                        "min": -25,
+                        "max": 24,
+                        "tooltip": (
+                            "Clip skip (ComfyUI: -24..-1, A1111: 1+). "
+                            "Default -25 means not set — any other value "
+                            "overwrites."
+                        ),
+                    },
+                ),
+                "additional_data": (
+                    "STRING",
+                    {
+                        "default": "",
+                        "multiline": True,
+                        "tooltip": (
+                            "Additional data to embed in the image metadata. "
+                            "Inserted between Clip skip and Model hash in the "
+                            "A1111-compatible parameters string. "
+                            'Example: "Copyright": "Some license info"'
+                        ),
+                    },
+                ),
+            },
+        }
+
+    RETURN_TYPES = ("METADATA",)
+    RETURN_NAMES = ("metadata",)
+    FUNCTION = "collect_metadata"
+    OUTPUT_NODE = True
+
+    def collect_metadata(self, **kwargs: Any) -> tuple[dict[str, Any]]:
+        """Collect non-default input values into a metadata dict.
+
+        For most fields, a falsy value (empty string, 0) means "not set"
+        and is skipped.  clip_skip uses a dedicated sentinel (-25) so that
+        a wired value of 0 is preserved and reaches the metadata pipeline.
+        """
+        result: dict[str, Any] = {}
+        for key in METADATA_OVERWRITE_FIELDS:
+            value = kwargs.get(key)
+            if key == "clip_skip":
+                if value != _CLIP_SKIP_SENTINEL:
+                    result[key] = value
+            elif value:
+                result[key] = value
+        return (result,)

py/nodes/save_image.py

@@ -16,6 +16,156 @@ from PIL import Image, PngImagePlugin
 import piexif
 import logging
 
+# Civitai-compatible sampler name mapping: ComfyUI internal → A1111 display name
+CIVITAI_SAMPLER_MAP = {
+    "euler": "Euler",
+    "euler_ancestral": "Euler a",
+    "lms": "LMS",
+    "heun": "Heun",
+    "dpm_2": "DPM2",
+    "dpm_2_ancestral": "DPM2 a",
+    "dpmpp_2s_ancestral": "DPM++ 2S a",
+    "dpmpp_2m": "DPM++ 2M",
+    "dpmpp_sde": "DPM++ SDE",
+    "dpmpp_sde_gpu": "DPM++ SDE",
+    "dpmpp_2m_sde": "DPM++ 2M SDE",
+    "dpmpp_2m_sde_gpu": "DPM++ 2M SDE",
+    "dpmpp_3m_sde": "DPM++ 3M SDE",
+    "dpm_fast": "DPM fast",
+    "dpm_adaptive": "DPM adaptive",
+    "ddim": "DDIM",
+    "plms": "PLMS",
+    "uni_pc_bh2": "UniPC",
+    "uni_pc": "UniPC",
+    "lcm": "LCM",
+}
+
+# Base model display name → AIR URN slug
+# Sourced from civitai source: src/shared/constants/basemodel.constants.ts
+BASE_MODEL_AIR_SLUG = {
+    # Stable Diffusion family
+    "SD 1.4": "sd1",
+    "SD 1.5": "sd1",
+    "SD 1.5 LCM": "sd1",
+    "SD 1.5 Hyper": "sd1",
+    "SD 2.0": "sd2",
+    "SD 2.0 768": "sd2",
+    "SD 2.1": "sd2",
+    "SD 2.1 768": "sd2",
+    "SD 2.1 Unclip": "sd2",
+    "SD 3.0": "sd3",
+    "SD 3.5": "sd35",
+    "SD 3.5 Large": "sd35",
+    "SD 3.5 Large Turbo": "sd35",
+    "SD 3.5 Medium": "sd35",
+    "SDXL 0.9": "sdxl",
+    "SDXL 1.0": "sdxl",
+    "SDXL 1.0 LCM": "sdxl",
+    "SDXL Lightning": "sdxl",
+    "SDXL Hyper": "sdxl",
+    "SDXL Turbo": "sdxl",
+    "SDXL Distilled": "sdxldistilled",
+    "Stable Cascade": "scascade",
+    "Stable Video Diffusion": "svd",
+    "SVD": "svd",
+    "SVD XT": "svdxt",
+
+    # SDXL community fine-tunes
+    "Pony": "pony",
+    "Pony Diffusion": "pony",
+    "Illustrious": "illustrious",
+    "NoobAI": "noobai",
+    "Animagine": "illustrious",
+
+    # Flux family
+    "Flux.1": "flux1",
+    "Flux.1 D": "flux1",
+    "Flux.1 S": "flux1",
+    "Flux.1 Krea": "fluxkrea",
+    "Flux.1 Kontext": "flux1kontext",
+    "Flux.2": "flux2",
+    "Flux.2 D": "flux2",
+    "Flux.2 Klein 9B": "flux2klein_9b",
+    "Flux.2 Klein 9B Base": "flux2klein_9b_base",
+    "Flux.2 Klein 4B": "flux2klein_4b",
+    "Flux.2 Klein 4B Base": "flux2klein_4b_base",
+
+    # Other image models (sorted alphabetically)
+    "AuraFlow": "auraflow",
+    "Chroma": "chroma",
+    "HiDream": "hidream",
+    "HiDream-O1": "hidream-o1",
+    "Hunyuan DiT": "hydit1",
+    "Hunyuan Video": "hyv1",
+    "Kolors": "kolors",
+    "Lumina": "lumina",
+    "Mochi": "mochi",
+    "ODOR": "odor",
+    "PixArt Alpha": "pixarta",
+    "PixArt Sigma": "pixarte",
+    "Playground v2": "playgroundv2",
+    "Playground v2.5": "playgroundv2",
+    "Pony Diffusion V7": "ponyv7",
+
+    # Video models
+    "CogVideoX": "cogvideox",
+    "LTX Video": "ltxv",
+    "LTX Video 2": "ltxv2",
+    "LTX Video 2.3": "ltxv23",
+    "Wan Video": "wanvideo",
+    "Wan Video 1.3B T2V": "wanvideo_13b_t2v",
+    "Wan Video 14B T2V": "wanvideo_14b_t2v",
+    "Wan Video 14B I2V 480p": "wanvideo_14b_i2v_480p",
+    "Wan Video 14B I2V 720p": "wanvideo_14b_i2v_720p",
+
+    # Third-party / proprietary image models
+    "Boogu": "boogu",
+    "Ernie": "ernie",
+    "Grok": "grok",
+    "HappyHorse": "happyhorse",
+    "Ideogram": "ideogram",
+    "Ideogram 4.0": "ideogram",
+    "Imagen": "imagen4",
+    "Imagen 4": "imagen4",
+    "Krea": "krea2",
+    "Krea 2": "krea2",
+    "Lens": "lens",
+    "MAI": "mai",
+    "Nano Banana": "nanobanana",
+    "OpenAI": "openai",
+    "Reve": "reve",
+    "Reve 2": "reve",
+    "Reve 2.1": "reve",
+    "Seedream": "seedream",
+    "Sora": "sora2",
+    "Sora 2": "sora2",
+    "Veo": "veo3",
+    "Veo 2": "veo3",
+    "Veo 3": "veo3",
+    "ZImageTurbo": "zimageturbo",
+    "ZImageBase": "zimagebase",
+    "ZImage": "zimagebase",
+
+    # Third-party video models
+    "Hailuo by MiniMax": "minimax",
+    "Haiper": "haiper",
+    "Kling": "kling",
+    "Lightricks": "lightricks",
+    "Seedance": "seedance",
+    "Vidu": "vidu",
+
+    # Qwen family
+    "Qwen": "qwen",
+    "Qwen 2": "qwen2",
+
+    # Anima
+    "Anima": "anima",
+
+    # Special
+    "Upscaler": "upscaler",
+    "Other": "other",
+}
+
 logger = logging.getLogger(__name__)
 
 
@@ -70,11 +220,29 @@ class SaveImageLM:
                         "tooltip": "Compression quality for JPEG and lossy WebP formats (1-100). Higher values mean better quality but larger files.",
                     },
                 ),
+                "webp_method": (
+                    "INT",
+                    {
+                        "default": 6,
+                        "min": 0,
+                        "max": 6,
+                        "tooltip": "WebP compression method (0-6). 0=fastest/largest, 6=slowest/smallest. Only applies when file_format is 'webp'.",
+                    },
+                ),
+                "jpeg_subsampling": (
+                    "INT",
+                    {
+                        "default": 0,
+                        "min": 0,
+                        "max": 2,
+                        "tooltip": "JPEG chroma subsampling level. 0=4:4:4 (best quality), 1=4:2:2, 2=4:2:0 (smallest files). Only applies when file_format is 'jpeg'.",
+                    },
+                ),
                 "embed_workflow": (
                     "BOOLEAN",
                     {
                         "default": False,
-                        "tooltip": "Embeds the complete workflow data into the image metadata. Only works with PNG and WebP formats.",
+                        "tooltip": "When enabled, saved images store the complete workflow. Drag the image back into ComfyUI to restore the original node graph. PNG and WebP only.",
                     },
                 ),
                 "save_with_metadata": (
@@ -142,148 +310,194 @@ class SaveImageLM:
 
         return None
 
-    def format_metadata(self, metadata_dict):
-        """Format metadata in the requested format similar to userComment example"""
-        if not metadata_dict:
-            return ""
+    def _resolve_model_cache_entry(self, scanner_type: str, name: str):
+        """Resolve model hash, civitai metadata, and base_model from scanner cache.
+        Returns (hash_str, civitai_dict, base_model_str). All values are empty defaults when not found."""
+        scanner = ServiceRegistry.get_service_sync(scanner_type)
+        if scanner is None or not name:
+            return "", {}, ""
 
-        # Helper function to only add parameter if value is not None
-        def add_param_if_not_none(param_list, label, value):
-            if value is not None:
-                param_list.append(f"{label}: {value}")
+        entry = self._get_cached_model_by_name(scanner, name)
+        if entry is None:
+            basename = os.path.splitext(os.path.basename(name))[0]
+            hash_val = scanner.get_hash_by_filename(basename)
+            return (hash_val or "").lower(), {}, ""
+
+        hash_val = (entry.get("sha256") or "").lower()
+        civitai = entry.get("civitai") or {}
+        base_model = entry.get("base_model") or ""
+        return hash_val, civitai, base_model
+
+    @staticmethod
+    def _get_civitai_sampler_name(sampler_name: str, scheduler: str) -> str:
+        if sampler_name in CIVITAI_SAMPLER_MAP:
+            civitai_name = CIVITAI_SAMPLER_MAP[sampler_name]
+            if scheduler == "karras":
+                civitai_name += " Karras"
+            elif scheduler == "exponential":
+                civitai_name += " Exponential"
+            return civitai_name
+        else:
+            if scheduler and scheduler != "normal":
+                return f"{sampler_name}_{scheduler}"
+            return sampler_name
+
+    @staticmethod
+    def _build_air_string(base_model: str, model_type: str, model_id: int, version_id: int) -> str:
+        slug = BASE_MODEL_AIR_SLUG.get(base_model, "other")
+        type_lower = model_type.lower() if model_type else "other"
+        return f"urn:air:{slug}:{type_lower}:civitai:{model_id}@{version_id}"
+
+    def format_metadata(self, metadata_dict: dict) -> str:
+        """Format metadata as A1111-compatible parameters string with Hashes JSON and Civitai resources."""
+        if not metadata_dict: return ""
 
-        # Extract the prompt and negative prompt
         prompt = metadata_dict.get("prompt", "")
         negative_prompt = metadata_dict.get("negative_prompt", "")
-
-        # Extract loras from the prompt if present
+        steps = metadata_dict.get("steps")
+        cfg = metadata_dict.get("guidance")
+        if cfg is None:
+            cfg = metadata_dict.get("cfg_scale")
+        if cfg is None:
+            cfg = metadata_dict.get("cfg")
+        seed = metadata_dict.get("seed")
+        size = metadata_dict.get("size")
+        sampler = metadata_dict.get("sampler") or ""
+        scheduler = metadata_dict.get("scheduler") or "normal"
+        checkpoint = metadata_dict.get("checkpoint") or ""
         loras_text = metadata_dict.get("loras", "")
-        lora_hashes = {}
+        clip_skip = metadata_dict.get("clip_skip")
 
-        # If loras are found, add them on a new line after the prompt
+        # Parse LoRA entries from <lora:name:strength> format
+        lora_entries: list[tuple[str, float]] = []
         if loras_text:
-            prompt_with_loras = f"{prompt}\n{loras_text}"
+            for match in re.findall(r"<lora:([^:]+):([^>]+)>", loras_text):
+                lora_name, strength_str = match
+                try:
+                    strength = float(strength_str)
+                except (ValueError, TypeError):
+                    strength = 1.0
+                lora_entries.append((lora_name, strength))
 
-            # Extract lora names from the format <lora:name:strength>
-            lora_matches = re.findall(r"<lora:([^:]+):([^>]+)>", loras_text)
+        # Resolve checkpoint hash and Civitai data from local cache
+        ckpt_hash, ckpt_civitai, ckpt_base_model = "", {}, ""
+        ckpt_display_name = ""
+        if checkpoint:
+            ckpt_hash, ckpt_civitai, ckpt_base_model = self._resolve_model_cache_entry(
+                "checkpoint_scanner", checkpoint
+            )
+            ckpt_display_name = os.path.splitext(os.path.basename(checkpoint))[0]
 
-            # Get hash for each lora
-            for lora_name, strength in lora_matches:
-                hash_value = self.get_lora_hash(lora_name)
-                if hash_value:
-                    lora_hashes[lora_name] = hash_value
-        else:
-            prompt_with_loras = prompt
+        # Resolve LoRA hash and Civitai data from local cache
+        loras_data: list[dict] = []
+        for lora_name, strength in lora_entries:
+            lora_hash, lora_civitai, lora_base_model = self._resolve_model_cache_entry(
+                "lora_scanner", lora_name
+            )
+            loras_data.append({
+                "name": lora_name,
+                "strength": strength,
+                "hash": lora_hash,
+                "civitai": lora_civitai,
+                "base_model": lora_base_model,
+            })
 
-        # Format the first part (prompt and loras)
-        metadata_parts = [prompt_with_loras]
+        # Build Hashes JSON (A1111 / Civitai standard format)
+        hashes: dict[str, str] = {}
+        if ckpt_hash:
+            hashes["model"] = ckpt_hash[:10].upper()
+        for lora in loras_data:
+            if lora["hash"]:
+                hashes[f"LORA:{lora['name']}"] = lora["hash"][:10].upper()
 
-        # Add negative prompt
+        # Build Civitai resources JSON array
+        civitai_resources: list[dict] = []
+        if ckpt_civitai.get("id", 0) > 0:
+            ckpt_resource: dict = {}
+            ckpt_type = (ckpt_civitai.get("model") or {}).get("type", "Checkpoint")
+            model_id = ckpt_civitai.get("modelId", 0)
+            version_id = ckpt_civitai.get("id", 0)
+            if model_id and version_id:
+                ckpt_resource["air"] = self._build_air_string(
+                    ckpt_base_model, ckpt_type, int(model_id), int(version_id)
+                )
+            elif version_id:
+                ckpt_resource["modelVersionId"] = int(version_id)
+            if ckpt_civitai.get("name"):
+                ckpt_resource["versionName"] = ckpt_civitai["name"]
+            if ckpt_resource:
+                civitai_resources.append(ckpt_resource)
+
+        for lora in loras_data:
+            lora_civitai = lora["civitai"]
+            if not lora_civitai or lora_civitai.get("id", 0) <= 0:
+                continue
+            lora_resource: dict = {"weight": lora["strength"]}
+            lora_type = (lora_civitai.get("model") or {}).get("type", "LORA")
+            model_id = lora_civitai.get("modelId", 0)
+            version_id = lora_civitai.get("id", 0)
+            if model_id and version_id:
+                lora_resource["air"] = self._build_air_string(
+                    lora["base_model"], lora_type, int(model_id), int(version_id)
+                )
+            elif version_id:
+                lora_resource["modelVersionId"] = int(version_id)
+            if lora_civitai.get("name"):
+                lora_resource["versionName"] = lora_civitai["name"]
+            civitai_resources.append(lora_resource)
+
+        sampler_name = CIVITAI_SAMPLER_MAP.get(sampler, sampler) if sampler else None
+
+        scheduler_mapping = {
+            "normal": "Normal",
+            "karras": "Karras",
+            "exponential": "Exponential",
+            "sgm_uniform": "SGM Uniform",
+            "sgm_quadratic": "SGM Quadratic",
+        }
+        scheduler_name = scheduler_mapping.get(scheduler, scheduler) if scheduler else None
+
+        # Build output lines
+        lines = [prompt] if prompt else [""]
         if negative_prompt:
-            metadata_parts.append(f"Negative prompt: {negative_prompt}")
+            lines.append(f"Negative prompt: {negative_prompt}")
 
-        # Format the second part (generation parameters)
-        params = []
-
-        # Add standard parameters in the correct order
-        if "steps" in metadata_dict:
-            add_param_if_not_none(params, "Steps", metadata_dict.get("steps"))
-
-        # Combine sampler and scheduler information
-        sampler_name = None
-        scheduler_name = None
-
-        if "sampler" in metadata_dict:
-            sampler = metadata_dict.get("sampler")
-            # Convert ComfyUI sampler names to user-friendly names
-            sampler_mapping = {
-                "euler": "Euler",
-                "euler_ancestral": "Euler a",
-                "dpm_2": "DPM2",
-                "dpm_2_ancestral": "DPM2 a",
-                "heun": "Heun",
-                "dpm_fast": "DPM fast",
-                "dpm_adaptive": "DPM adaptive",
-                "lms": "LMS",
-                "dpmpp_2s_ancestral": "DPM++ 2S a",
-                "dpmpp_sde": "DPM++ SDE",
-                "dpmpp_sde_gpu": "DPM++ SDE",
-                "dpmpp_2m": "DPM++ 2M",
-                "dpmpp_2m_sde": "DPM++ 2M SDE",
-                "dpmpp_2m_sde_gpu": "DPM++ 2M SDE",
-                "ddim": "DDIM",
-            }
-            sampler_name = sampler_mapping.get(sampler, sampler)
-
-        if "scheduler" in metadata_dict:
-            scheduler = metadata_dict.get("scheduler")
-            scheduler_mapping = {
-                "normal": "Simple",
-                "karras": "Karras",
-                "exponential": "Exponential",
-                "sgm_uniform": "SGM Uniform",
-                "sgm_quadratic": "SGM Quadratic",
-            }
-            scheduler_name = scheduler_mapping.get(scheduler, scheduler)
-
-        # Add combined sampler and scheduler information
+        params: list[str] = []
+        if steps is not None:
+            params.append(f"Steps: {steps}")
         if sampler_name:
             if scheduler_name:
                 params.append(f"Sampler: {sampler_name} {scheduler_name}")
             else:
                 params.append(f"Sampler: {sampler_name}")
+        if cfg is not None:
+            params.append(f"CFG scale: {cfg}")
+        if seed is not None:
+            params.append(f"Seed: {seed}")
+        if size:
+            params.append(f"Size: {size}")
+        if clip_skip is not None:
+            try:
+                params.append(f"Clip skip: {abs(int(clip_skip))}")
+            except (ValueError, TypeError):
+                pass
+        additional_data = metadata_dict.get("additional_data", "")
+        if additional_data:
+            params.append(additional_data)
+        if ckpt_hash:
+            params.append(f"Model hash: {ckpt_hash[:10].upper()}")
+        if ckpt_display_name:
+            params.append(f"Model: {ckpt_display_name}")
+        if hashes:
+            params.append(f"Hashes: {json.dumps(hashes, separators=(',', ':'))}")
+        params.append("Version: ComfyUI")
+        if civitai_resources:
+            params.append(
+                f"Civitai resources: {json.dumps(civitai_resources, separators=(',', ':'))}"
+            )
 
-        # CFG scale (Use guidance if available, otherwise fall back to cfg_scale or cfg)
-        if "guidance" in metadata_dict:
-            add_param_if_not_none(params, "CFG scale", metadata_dict.get("guidance"))
-        elif "cfg_scale" in metadata_dict:
-            add_param_if_not_none(params, "CFG scale", metadata_dict.get("cfg_scale"))
-        elif "cfg" in metadata_dict:
-            add_param_if_not_none(params, "CFG scale", metadata_dict.get("cfg"))
-
-        # Seed
-        if "seed" in metadata_dict:
-            add_param_if_not_none(params, "Seed", metadata_dict.get("seed"))
-
-        # Size
-        if "size" in metadata_dict:
-            add_param_if_not_none(params, "Size", metadata_dict.get("size"))
-
-        # Model info
-        if "checkpoint" in metadata_dict:
-            # Ensure checkpoint is a string before processing
-            checkpoint = metadata_dict.get("checkpoint")
-            if checkpoint is not None:
-                # Get model hash
-                model_hash = self.get_checkpoint_hash(checkpoint)
-
-                # Extract basename without path
-                checkpoint_name = os.path.basename(checkpoint)
-                # Remove extension if present
-                checkpoint_name = os.path.splitext(checkpoint_name)[0]
-
-                # Add model hash if available
-                if model_hash:
-                    params.append(
-                        f"Model hash: {model_hash[:10]}, Model: {checkpoint_name}"
-                    )
-                else:
-                    params.append(f"Model: {checkpoint_name}")
-
-        # Add LoRA hashes if available
-        if lora_hashes:
-            lora_hash_parts = []
-            for lora_name, hash_value in lora_hashes.items():
-                lora_hash_parts.append(f"{lora_name}: {hash_value[:10]}")
-
-            if lora_hash_parts:
-                params.append(f'Lora hashes: "{", ".join(lora_hash_parts)}"')
-
-        # Combine all parameters with commas
-        metadata_parts.append(", ".join(params))
-
-        # Join all parts with a new line
-        return "\n".join(metadata_parts)
+        lines.append(", ".join(params))
+        return "\n".join(lines)
 
     # credit to nkchocoai
     # Add format_filename method to handle pattern substitution
@@ -298,7 +512,12 @@ class SaveImageLM:
             key = parts[0]
 
             if key == "seed" and "seed" in metadata_dict:
-                filename = filename.replace(segment, str(metadata_dict.get("seed", "")))
+                seed_value = metadata_dict.get("seed")
+                if seed_value is not None:
+                    filename = filename.replace(segment, str(seed_value))
+                else:
+                    # Fallback if seed was not captured by metadata collector
+                    filename = filename.replace(segment, "0")
             elif key == "width" and "size" in metadata_dict:
                 size = metadata_dict.get("size", "x")
                 w = size.split("x")[0] if isinstance(size, str) else size[0]
@@ -568,6 +787,8 @@ class SaveImageLM:
         extra_pnginfo=None,
         lossless_webp=True,
         quality=100,
+        webp_method=6,
+        jpeg_subsampling=0,
         embed_workflow=False,
         save_with_metadata=True,
         add_counter_to_filename=True,
@@ -603,7 +824,7 @@ class SaveImageLM:
             img = Image.fromarray(np.clip(img, 0, 255).astype(np.uint8))
 
             # Generate filename with counter if needed
-            base_filename = filename
+            base_filename = filename.replace("%batch_num%", str(i))
             if add_counter_to_filename:
                 # Use counter + i to ensure unique filenames for all images in batch
                 current_counter = counter + i
@@ -622,15 +843,14 @@ class SaveImageLM:
             elif file_format == "jpeg":
                 file = base_filename + ".jpg"
                 file_extension = ".jpg"
-                save_kwargs = {"quality": quality, "optimize": True}
+                save_kwargs = {"quality": quality, "optimize": True, "subsampling": jpeg_subsampling}
             elif file_format == "webp":
                 file = base_filename + ".webp"
                 file_extension = ".webp"
-                # Add optimization param to control performance
                 save_kwargs = {
                     "quality": quality,
                     "lossless": lossless_webp,
-                    "method": 0,
+                    "method": webp_method,
                 }
             else:
                 raise ValueError(f"Unsupported file format: {file_format}")
@@ -717,6 +937,8 @@ class SaveImageLM:
         extra_pnginfo=None,
         lossless_webp=True,
         quality=100,
+        webp_method=6,
+        jpeg_subsampling=0,
         embed_workflow=False,
         save_with_metadata=True,
         add_counter_to_filename=True,
@@ -746,6 +968,8 @@ class SaveImageLM:
             extra_pnginfo,
             lossless_webp,
             quality,
+            webp_method,
+            jpeg_subsampling,
             embed_workflow,
             save_with_metadata,
             add_counter_to_filename,

py/nodes/utils.py

@@ -36,6 +36,7 @@ any_type = AnyType("*")
 
 # Common methods extracted from lora_loader.py and lora_stacker.py
 import os
+import re
 import logging
 import copy
 import sys
@@ -69,6 +70,25 @@ def extract_lora_name(lora_path):
     return apply_lora_syntax_format(name_no_ext)
 
 
+def parse_lora_syntax(text: str) -> list[dict]:
+    """Parse <lora:name:strength> syntax from text input into a list of dicts.
+
+    Each entry contains: name, model_strength, clip_strength.
+    Supports both ``<lora:name:strength>`` and ``<lora:name:model_strength:clip_strength>``.
+    """
+    pattern = r"<lora:([^:>]+):([^:>]+)(?::([^:>]+))?>"
+    matches = re.findall(pattern, text, re.IGNORECASE)
+    loras = []
+    for match in matches:
+        model_strength = float(match[1])
+        loras.append({
+            "name": match[0],
+            "model_strength": model_strength,
+            "clip_strength": float(match[2]) if match[2] else model_strength,
+        })
+    return loras
+
+
 def get_loras_list(kwargs):
     """Helper to extract loras list from either old or new kwargs format"""
     if "loras" not in kwargs:

py/recipes/parsers/automatic.py

@@ -123,24 +123,39 @@ class AutomaticMetadataParser(RecipeMetadataParser):
                         if model_hash_from_hashes:
                             metadata["model_hash"] = model_hash_from_hashes
                     
-                    # Extract Lora hashes in alternative format
+                    # Extract Lora hashes in alternative format.
+                    # Run unconditionally (not just as fallback) so that
+                    # non-empty hashes from Lora hashes fill in the gaps left
+                    # by empty values in the Hashes JSON dict.  Some WebUI
+                    # builds write real hash values only to Lora hashes and
+                    # leave the Hashes JSON values empty.
                     lora_hashes_match = re.search(self.LORA_HASHES_REGEX, params_section)
-                    if not hashes_match and lora_hashes_match:
+                    if lora_hashes_match:
                         try:
                             lora_hashes_str = lora_hashes_match.group(1)
                             lora_hash_entries = lora_hashes_str.split(', ')
-                            
-                            # Initialize hashes dict if it doesn't exist
-                            if "hashes" not in metadata:
-                                metadata["hashes"] = {}
-                                
+
                             # Parse each lora hash entry (format: "name: hash")
                             for entry in lora_hash_entries:
                                 if ': ' in entry:
                                     lora_name, lora_hash = entry.split(': ', 1)
-                                    # Add as lora type in the same format as regular hashes
-                                    metadata["hashes"][f"lora:{lora_name}"] = lora_hash.strip()
-                            
+                                    lora_hash = lora_hash.strip()
+                                    if not lora_hash:
+                                        # Skip entries without a hash value
+                                        continue
+                                    # Initialize hashes dict if it doesn't exist
+                                    if "hashes" not in metadata:
+                                        metadata["hashes"] = {}
+                                    # Add as lora type in the same format as
+                                    # regular hashes.  Only override an
+                                    # existing entry if its value is empty
+                                    # (Lora hashes is the more reliable
+                                    # source when Hashes JSON has blanks).
+                                    key = f"lora:{lora_name}"
+                                    existing = metadata["hashes"].get(key, "")
+                                    if not existing:
+                                        metadata["hashes"][key] = lora_hash
+
                             # Remove lora hashes from params section
                             params_section = params_section.replace(lora_hashes_match.group(0), '')
                         except Exception as e:
@@ -362,6 +377,12 @@ class AutomaticMetadataParser(RecipeMetadataParser):
                         # Only process lora or hypernet types
                         if not hash_key.startswith(("lora:", "hypernet:")):
                             continue
+                        
+                        # Skip entries without a hash value — they can't be
+                        # resolved via CivitAI and would only produce a
+                        # useless "Deleted" entry in the recipe.
+                        if not lora_hash:
+                            continue
                             
                         lora_type, lora_name = hash_key.split(':', 1)
                         
@@ -387,11 +408,7 @@ class AutomaticMetadataParser(RecipeMetadataParser):
                         # Try to get info from Civitai
                         if metadata_provider:
                             try:
-                                if lora_hash:
-                                    # If we have hash, use it for lookup
-                                    civitai_info = await metadata_provider.get_model_by_hash(lora_hash)
-                                else:
-                                    civitai_info = None
+                                civitai_info = await metadata_provider.get_model_by_hash(lora_hash)
                                 
                                 populated_entry = await self.populate_lora_from_civitai(
                                     lora_entry, 

py/recipes/parsers/civitai_image.py

@@ -514,11 +514,21 @@ class CivitaiApiMetadataParser(RecipeMetadataParser):
 
                         result["loras"].append(lora_entry)
 
-            # Process modelVersionIds from Civitai image API
-            # These are model version IDs returned at root level when meta doesn't contain resources
-            if "modelVersionIds" in metadata and isinstance(
-                metadata["modelVersionIds"], list
+            # Process modelVersionIds from Civitai image API.
+            # These are version IDs returned at root level of the API response.
+            # When resources or civitaiResources are already present in metadata
+            # (which they are when ?withMeta=true is passed), those sections have
+            # complete hash/type information — modelVersionIds is a fallback for
+            # when meta is null and only the flat ID list is available. Skipping
+            # it here avoids duplicates: the same file hash often resolves to
+            # different version IDs via hash lookup (resources) vs the original
+            # version ID in modelVersionIds, and both paths would create entries.
+            if (
+                "modelVersionIds" in metadata
+                and isinstance(metadata["modelVersionIds"], list)
+                and not result.get("loras")
             ):
+
                 for version_id in metadata["modelVersionIds"]:
                     version_id_str = str(version_id)
 
@@ -526,6 +536,13 @@ class CivitaiApiMetadataParser(RecipeMetadataParser):
                     if version_id_str in added_loras:
                         continue
 
+                    # Skip if this version ID is already the recipe's checkpoint
+                    # (resolved earlier from embedded resources/Model hash,
+                    # avoiding a duplicate CivitAI API call).
+                    existing_model = result.get("model")
+                    if existing_model and str(existing_model.get("id")) == version_id_str:
+                        continue
+
                     # Initialize lora entry with version ID
                     lora_entry = {
                         "id": version_id,
@@ -559,9 +576,40 @@ class CivitaiApiMetadataParser(RecipeMetadataParser):
                             )
 
                             if populated_entry is None:
-                                continue  # Skip invalid LoRA types
+                                # Not a LoRA — try as checkpoint (only if we
+                                # don't already have one).  Reuses the same
+                                # civitai_info from the API call above so no
+                                # extra query is made.
+                                if result["model"] is None:
+                                    checkpoint_entry = {
+                                        "id": version_id,
+                                        "modelId": 0,
+                                        "name": "Unknown Model",
+                                        "version": "",
+                                        "type": "checkpoint",
+                                        "existsLocally": False,
+                                        "localPath": None,
+                                        "file_name": "",
+                                        "hash": "",
+                                        "thumbnailUrl": (
+                                            "/loras_static/images/no-preview.png"
+                                        ),
+                                        "baseModel": "",
+                                        "size": 0,
+                                        "downloadUrl": "",
+                                        "isDeleted": False,
+                                    }
+                                    cp_populated = await (
+                                        self.populate_checkpoint_from_civitai(
+                                            checkpoint_entry, civitai_info
+                                        )
+                                    )
+                                    if cp_populated.get("modelId"):
+                                        result["model"] = cp_populated
+                                continue  # Not a LoRA, don't add to loras
 
                             lora_entry = populated_entry
+
                         except Exception as e:
                             logger.error(
                                 f"Error fetching Civitai info for model version {version_id}: {e}"

py/routes/handlers/agent_handlers.py

@@ -0,0 +1,165 @@
+"""HTTP route handlers for agent skill endpoints.
+
+These handlers expose the :class:`AgentService` via HTTP, allowing the
+frontend to list available skills and execute them on selected models.
+Progress is reported via WebSocket broadcast.
+"""
+
+from __future__ import annotations
+
+import asyncio
+import logging
+from typing import Any, Dict
+
+from aiohttp import web
+
+from ...services.agent import AgentService, AgentProgressReporter
+from ...services.llm_service import LLMNotConfiguredError
+
+logger = logging.getLogger(__name__)
+
+
+class AgentHandler:
+    """HTTP handler for agent skill operations."""
+
+    def __init__(self, agent_service: AgentService | None = None) -> None:
+        self._agent_service = agent_service
+
+    async def _ensure_service(self) -> AgentService:
+        if self._agent_service is None:
+            self._agent_service = await AgentService.get_instance()
+        return self._agent_service
+
+    # ------------------------------------------------------------------
+    # GET /api/lm/agent/skills
+    # ------------------------------------------------------------------
+
+    async def get_agent_skills(self, request: web.Request) -> web.Response:
+        """Return a list of available agent skills."""
+
+        service = await self._ensure_service()
+        skills = await service.list_skills()
+        return web.json_response({"skills": skills})
+
+    # ------------------------------------------------------------------
+    # POST /api/lm/agent/execute/{skill_name}
+    # ------------------------------------------------------------------
+
+    async def execute_agent_skill(self, request: web.Request) -> web.Response:
+        """Execute an agent skill on the provided model paths.
+
+        Request body::
+
+            {"model_paths": ["/path/to/model1.safetensors", ...], "options": {}}
+
+        Returns immediately with a task ID.  Execution runs in the
+        background; progress and completion are pushed via WebSocket
+        events of type ``agent_progress``.
+        """
+
+        skill_name = request.match_info.get("skill_name", "")
+        if not skill_name:
+            return web.json_response(
+                {"error": "Skill name is required"}, status=400
+            )
+
+        try:
+            body = await request.json()
+        except Exception:
+            return web.json_response(
+                {"error": "Invalid JSON body"}, status=400
+            )
+
+        model_paths = body.get("model_paths", [])
+        if not model_paths or not isinstance(model_paths, list):
+            return web.json_response(
+                {"error": "model_paths must be a non-empty array"},
+                status=400,
+            )
+
+        service = await self._ensure_service()
+
+        # Validate LLM configuration early for skills that need it
+        # (fail fast rather than after starting background work)
+        try:
+            from ...services.llm_service import LLMService
+
+            llm = await LLMService.get_instance()
+            if not llm.is_configured():
+                return web.json_response(
+                    {
+                        "error": "LLM provider is not configured. "
+                        "Enable it in Settings → AI Provider.",
+                    },
+                    status=400,
+                )
+        except Exception as exc:
+            logger.error("Failed to check LLM configuration: %s", exc)
+
+        # Launch execution in the background
+        progress_reporter = AgentProgressReporter()
+        logger.info(
+            "LLM enrichment '%s' starting for %d model(s)",
+            skill_name, len(model_paths),
+        )
+
+        async def _run() -> None:
+            try:
+                result = await service.execute_skill(
+                    skill_name=skill_name,
+                    input_data={"model_paths": model_paths},
+                    progress_callback=progress_reporter,
+                )
+                logger.info(
+                    "LLM enrichment '%s' finished: success=%s, summary='%s', errors=%s",
+                    skill_name, result.success, result.summary, result.errors,
+                )
+            except LLMNotConfiguredError as exc:
+                logger.warning("LLM enrichment '%s' not configured: %s", skill_name, exc)
+                await progress_reporter.on_progress(
+                    {
+                        "type": "agent_progress",
+                        "skill": skill_name,
+                        "status": "error",
+                        "error": str(exc),
+                    }
+                )
+            except Exception as exc:
+                logger.error("LLM enrichment '%s' failed: %s", skill_name, exc, exc_info=True)
+                await progress_reporter.on_progress(
+                    {
+                        "type": "agent_progress",
+                        "skill": skill_name,
+                        "status": "error",
+                        "error": str(exc),
+                    }
+                )
+
+        # Fire and forget — progress comes via WebSocket
+        asyncio.create_task(_run())
+
+        return web.json_response(
+            {
+                "status": "started",
+                "skill": skill_name,
+                "model_count": len(model_paths),
+            }
+        )
+
+    # ------------------------------------------------------------------
+    # POST /api/lm/agent/cancel
+    # ------------------------------------------------------------------
+
+    async def cancel_agent_skill(self, request: web.Request) -> web.Response:
+        """Cancel a running agent skill.
+
+        NOTE: Cancellation is a stub for now — the AgentService processes
+        models sequentially and does not yet support mid-execution
+        cancellation.  This endpoint exists for API completeness.
+        """
+
+        # TODO: implement cooperative cancellation in AgentService
+        return web.json_response(
+            {"status": "acknowledged", "note": "Cancellation not yet implemented"},
+            status=200,
+        )

py/routes/handlers/hf_handlers.py

@@ -0,0 +1,508 @@
+"""Handlers for Hugging Face model listing and download.
+
+Minimal MVP implementation — uses direct HTTP to the HF API for file
+listing and the project's existing aiohttp-based Downloader for
+downloading.  No huggingface_hub dependency required.
+"""
+
+from __future__ import annotations
+
+import json
+import logging
+import os
+import re
+from typing import Any
+
+import aiohttp
+from aiohttp import web
+
+from ...config import config
+from ...services.downloader import (
+    DownloadProgress,
+    get_downloader,
+)
+from ...services.aria2_downloader import Aria2Downloader
+from ...services.settings_manager import get_settings_manager
+from ...services.service_registry import ServiceRegistry
+from ...services.websocket_manager import ws_manager
+from ...utils.constants import MODEL_FILE_EXTENSIONS
+from ...utils.metadata_manager import MetadataManager
+from ...utils.models import LoraMetadata, CheckpointMetadata, EmbeddingMetadata
+
+logger = logging.getLogger(__name__)
+
+_DEFAULT_MODEL_CLASS = LoraMetadata
+_DEFAULT_SCANNER_GETTER = "get_lora_scanner"
+
+# Shared aiohttp session for HF API calls (created on first use)
+_hf_api_session: aiohttp.ClientSession | None = None
+
+
+async def _get_hf_api_session() -> aiohttp.ClientSession:
+    """Get or create the shared aiohttp session for HF API calls."""
+    global _hf_api_session  # needed because we reassign the module-level name
+    if _hf_api_session is None or _hf_api_session.closed:
+        _hf_api_session = aiohttp.ClientSession(
+            headers={"User-Agent": "ComfyUI-LoRA-Manager/1.0"},
+            timeout=aiohttp.ClientTimeout(total=30),
+        )
+    return _hf_api_session
+
+
+async def close_hf_api_session() -> None:
+    """Close the shared HF API session, if it was ever created."""
+    global _hf_api_session
+    if _hf_api_session is not None and not _hf_api_session.closed:
+        await _hf_api_session.close()
+        _hf_api_session = None
+
+
+def _infer_model_type(model_root: str) -> tuple[Any, str]:
+    """Determine model class and scanner by matching ``model_root`` against the
+    configured root paths for each model type (from ``Config``).
+
+    The ``model_root`` value comes from the frontend's model-root dropdown,
+    which is populated from the current page's scanner roots.  By checking
+    which scanner's root list it belongs to, we avoid fragile heuristics
+    like substring-matching path names.
+    """
+    norm = os.path.normpath(model_root).replace(os.sep, "/")
+
+    # LoRA roots
+    for p in (config.loras_roots or []) + (config.extra_loras_roots or []):
+        if os.path.normpath(p).replace(os.sep, "/") == norm:
+            return LoraMetadata, "get_lora_scanner"
+
+    # Checkpoint / UNet roots
+    for p in (
+        (config.checkpoints_roots or [])
+        + (config.extra_checkpoints_roots or [])
+        + (config.unet_roots or [])
+        + (config.extra_unet_roots or [])
+    ):
+        if os.path.normpath(p).replace(os.sep, "/") == norm:
+            return CheckpointMetadata, "get_checkpoint_scanner"
+
+    # Embedding roots
+    for p in (config.embeddings_roots or []) + (config.extra_embeddings_roots or []):
+        if os.path.normpath(p).replace(os.sep, "/") == norm:
+            return EmbeddingMetadata, "get_embedding_scanner"
+
+    # Fallback — should not happen in normal use
+    logger.warning(
+        "Could not determine model type for root '%s'; defaulting to LoRA",
+        model_root,
+    )
+    return _DEFAULT_MODEL_CLASS, _DEFAULT_SCANNER_GETTER
+
+
+async def _save_hf_metadata(dest_path: str, repo: str, model_root: str) -> None:
+    """Create a proper .metadata.json and add the model to the scanner cache.
+
+    Uses ``MetadataManager.create_default_metadata()`` which computes the
+    SHA256 hash, extracts safetensors header metadata (base_model), and
+    produces a fully-populated ``LoraMetadata`` (or ``CheckpointMetadata`` /
+    ``EmbeddingMetadata``) object.  We then overlay HF-specific fields and
+    register the model in the in-memory scanner cache so it appears
+    immediately without a full filesystem walk.
+    """
+    try:
+        hf_url = f"https://huggingface.co/{repo}"
+        model_class, scanner_getter_name = _infer_model_type(model_root)
+
+        # 1. Create proper metadata (computes SHA256, reads safetensors headers)
+        metadata = await MetadataManager.create_default_metadata(
+            dest_path, model_class=model_class
+        )
+        if metadata is None:
+            logger.warning("create_default_metadata returned None for %s", dest_path)
+            return
+
+        # 2. Overlay HF-specific fields
+        metadata._unknown_fields["hf_url"] = hf_url
+        metadata.from_civitai = False  # HF models are not from CivitAI
+
+        metadata_dict = metadata.to_dict()
+        if "trainedWords" in metadata_dict and not metadata_dict["trainedWords"]:
+            del metadata_dict["trainedWords"]
+
+        # 3. Save metadata atomically
+        await MetadataManager.save_metadata(dest_path, metadata_dict)
+        logger.info("Saved HF metadata (with hf_url) for %s", dest_path)
+
+        # 4. Determine relative folder path for cache
+        #    model_root is an absolute path; dest_path is under it
+        folder = ""
+        if os.path.isabs(model_root) and dest_path.startswith(model_root):
+            rel = os.path.relpath(os.path.dirname(dest_path), model_root)
+            folder = rel.replace(os.sep, "/") if rel != "." else ""
+
+        # 5. Add to scanner cache (same as CivitAI's _execute_download does)
+        scanner_getter = getattr(ServiceRegistry, scanner_getter_name, None)
+        if scanner_getter is not None:
+            scanner = await scanner_getter()
+            if scanner is not None:
+                metadata_dict = metadata.to_dict()
+                metadata_dict["hf_url"] = hf_url
+                await scanner.add_model_to_cache(metadata_dict, folder)
+                logger.info("Added %s to scanner cache (folder=%s)", dest_path, folder)
+
+    except Exception as exc:
+        logger.warning("Failed to save HF metadata for %s: %s", dest_path, exc)
+
+
+def _find_matching_root(dest_dir: str) -> str | None:
+    """Walk up *dest_dir* to find which configured scanner root it belongs to."""
+    norm = os.path.normpath(dest_dir).replace(os.sep, "/")
+    all_roots = []
+    for root_list in (
+        config.loras_roots or [],
+        config.extra_loras_roots or [],
+        config.checkpoints_roots or [],
+        config.extra_checkpoints_roots or [],
+        config.unet_roots or [],
+        config.extra_unet_roots or [],
+        config.embeddings_roots or [],
+        config.extra_embeddings_roots or [],
+    ):
+        all_roots.extend([os.path.normpath(p).replace(os.sep, "/") for p in root_list])
+    # Find the longest matching prefix
+    match: str | None = None
+    for root in all_roots:
+        if norm.startswith(root):
+            if match is None or len(root) > len(match):
+                match = root
+    return match
+
+
+async def _add_to_scanner_cache(dest_path: str, metadata: dict[str, Any]) -> None:
+    model_dir = os.path.dirname(dest_path)
+    model_root = _find_matching_root(model_dir)
+    if not model_root:
+        raise ValueError(f"File path {dest_path} is not within any configured scanner root")
+    scanner_getter_name = _infer_model_type(model_root)[1]
+    scanner_getter = getattr(ServiceRegistry, scanner_getter_name, None)
+    if scanner_getter is None:
+        raise RuntimeError(f"Scanner getter '{scanner_getter_name}' not found in ServiceRegistry")
+    scanner = await scanner_getter()
+    if scanner is None:
+        raise RuntimeError(f"Scanner '{scanner_getter_name}' returned None")
+    await scanner.update_single_model_cache(dest_path, dest_path, metadata)
+
+
+class HfHandler:
+    """Handle Hugging Face model browsing and download."""
+
+    async def set_hf_url(self, request: web.Request) -> web.Response:
+        try:
+            payload: dict[str, Any] = await request.json()
+        except json.JSONDecodeError:
+            return web.json_response({"success": False, "error": "Invalid JSON"}, status=400)
+
+        file_path = (payload.get("file_path") or "").strip()
+        hf_url = (payload.get("hf_url") or "").strip()
+
+        if not file_path or not hf_url:
+            return web.json_response(
+                {"success": False, "error": "Missing required fields: 'file_path' and 'hf_url'"},
+                status=400,
+            )
+
+        m = re.match(r"^https?://huggingface\.co/([^/]+/[^/]+)/?$", hf_url)
+        if not m:
+            return web.json_response(
+                {
+                    "success": False,
+                    "error": "Invalid HuggingFace URL. Expected format: https://huggingface.co/user/repo",
+                },
+                status=400,
+            )
+
+        if not os.path.isfile(file_path):
+            return web.json_response(
+                {"success": False, "error": f"File not found: {file_path}"},
+                status=404,
+            )
+
+        model_root = _find_matching_root(os.path.dirname(file_path))
+        if not model_root:
+            return web.json_response(
+                {
+                    "success": False,
+                    "error": "File is not within any configured model directory. Cannot link to HuggingFace.",
+                },
+                status=400,
+            )
+
+        try:
+            existing = await MetadataManager.load_metadata_payload(file_path)
+            if existing.get("hf_url") == hf_url:
+                return web.json_response({
+                    "success": True,
+                    "message": "hf_url already set",
+                    "hf_url": hf_url,
+                })
+
+            existing["hf_url"] = hf_url
+            existing["from_civitai"] = False
+            await MetadataManager.save_metadata(file_path, existing)
+
+            await _add_to_scanner_cache(file_path, existing)
+
+            logger.info("Set hf_url=%s for %s", hf_url, file_path)
+            return web.json_response({
+                "success": True,
+                "message": f"hf_url set to {hf_url}",
+                "hf_url": hf_url,
+            })
+        except Exception as exc:
+            logger.error("Failed to set hf_url for %s: %s", file_path, exc)
+            return web.json_response(
+                {"success": False, "error": str(exc)},
+                status=500,
+            )
+
+    async def get_hf_repo_files(self, request: web.Request) -> web.Response:
+        """List model-weight files from a HF repo with real file sizes.
+
+        Uses the HF tree API endpoint which returns accurate file sizes
+        (including LFS-tracked files), unlike the model info endpoint.
+        """
+        repo = request.query.get("repo", "").strip()
+        if not repo or "/" not in repo:
+            return web.json_response(
+                {"error": "Missing or invalid 'repo' parameter (expected user/repo)"},
+                status=400,
+            )
+
+        url = f"https://huggingface.co/api/models/{repo}/tree/main"
+
+        try:
+            session = await _get_hf_api_session()
+            async with session.get(url) as resp:
+                if resp.status == 404:
+                    return web.json_response(
+                        {"error": f"Repo '{repo}' not found"}, status=404
+                    )
+                if resp.status != 200:
+                    text = await resp.text()
+                    return web.json_response(
+                        {"error": f"HF API error {resp.status}: {text[:200]}"},
+                        status=resp.status,
+                    )
+                tree: list[dict[str, Any]] = await resp.json()
+        except Exception as exc:
+            logger.error("Failed to fetch HF repo files: %s", exc)
+            return web.json_response({"error": str(exc)}, status=502)
+
+        files: list[dict[str, Any]] = []
+        for entry in tree:
+            path: str = entry.get("path", "")
+            ext = os.path.splitext(path)[1].lower()
+            if ext not in MODEL_FILE_EXTENSIONS:
+                continue
+            size = entry.get("size", 0) or 0
+            if size == 0 and "lfs" in entry:
+                size = entry["lfs"].get("size", 0) or 0
+            files.append({
+                "filename": path,
+                "size": size,
+            })
+
+        files.sort(key=lambda f: f["size"], reverse=True)
+        return web.json_response(files)
+
+    async def download_hf_model(self, request: web.Request) -> web.Response:
+        """Download a single file from Hugging Face into the model directory.
+
+        POST JSON body::
+
+            {
+              "repo": "dx8152/Flux2-Klein-9B-Consistency",
+              "filename": "Flux2-Klein-9B-consistency-V2.safetensors",
+              "revision": "main",
+              "model_root": "loras",
+              "relative_path": "",
+              "use_default_paths": false,
+              "download_id": "optional-batch-id"
+            }
+
+        If ``download_id`` is provided, real-time progress (bytes, speed,
+        percentage) is broadcast via the WebSocket progress system, matching
+        the CivitAI download experience.
+
+        Respects the ``download_backend`` setting (``aria2`` or ``default``).
+        """
+        try:
+            payload: dict[str, Any] = await request.json()
+        except json.JSONDecodeError:
+            return web.json_response({"error": "Invalid JSON"}, status=400)
+
+        repo = (payload.get("repo") or "").strip()
+        filename = (payload.get("filename") or "").strip()
+        revision = (payload.get("revision") or "main").strip()
+        model_root = (payload.get("model_root") or "").strip()
+        relative_path = (payload.get("relative_path") or "").strip()
+        use_default_paths = bool(payload.get("use_default_paths", False))
+        download_id: str | None = payload.get("download_id")
+
+        logger.info(
+            "download_hf_model: repo=%s file=%s root=%s download_id=%s",
+            repo, filename, model_root, download_id,
+        )
+
+        if not repo or not filename:
+            return web.json_response(
+                {"error": "Missing required fields: 'repo' and 'filename'"}, status=400
+            )
+
+        # Validate repo format — must be user/repo_name
+        if repo.count("/") != 1 or not re.match(r"^[a-zA-Z0-9_.-]+/[a-zA-Z0-9_.-]+$", repo):
+            return web.json_response({"error": f"Invalid repo format: {repo}"}, status=400)
+        author, repo_name = repo.split("/", 1)
+        if ".." in (author, repo_name) or "." in (author, repo_name):
+            return web.json_response({"error": f"Invalid repo format: {repo}"}, status=400)
+
+        # Validate filename — must not contain path traversal
+        if ".." in filename:
+            return web.json_response({"error": "Invalid filename"}, status=400)
+
+        # Validate relative_path — must not be absolute or escape base directory
+        if relative_path:
+            if os.path.isabs(relative_path):
+                return web.json_response({"error": "relative_path must not be absolute"}, status=400)
+            if ".." in relative_path.split("/") or "\\" in relative_path:
+                return web.json_response({"error": "Invalid relative_path"}, status=400)
+
+        # Use model_root directly as the base directory — same approach as
+        # CivitAI's download path (download_manager.py).  No realpath, no
+        # allowed-roots validation, no path-traversal check; those are
+        # unnecessary when the frontend sends the path from its own dropdown
+        # (populated from scanner roots).  Using the "business path" directly
+        # keeps dest_path consistent with scanner roots so that later folder
+        # derivation (in _save_hf_metadata) works correctly.
+        if os.path.isabs(model_root):
+            base_dir = os.path.normpath(model_root)
+        else:
+            base_dir = os.path.normpath(os.path.join(os.getcwd(), "models", model_root))
+
+        if use_default_paths:
+            target_dir = os.path.join(base_dir, "huggingface", author, repo_name)
+        elif relative_path:
+            target_dir = os.path.join(base_dir, relative_path)
+        else:
+            target_dir = base_dir
+
+        # Strip HF repo subdirectory — "diffusion_models/xxx.safetensors"
+        # is an HF repo convention, not meaningful for local storage.
+        file_base = os.path.basename(filename)
+
+        os.makedirs(target_dir, exist_ok=True)
+        dest_path = os.path.join(target_dir, file_base)
+
+        # Check if already exists (simple skip)
+        if os.path.exists(dest_path) and os.path.getsize(dest_path) > 0:
+            logger.info("download_hf_model: file already exists, skipping — %s", dest_path)
+            return web.json_response({
+                "success": True,
+                "message": f"File already exists: {dest_path}",
+                "path": dest_path,
+            })
+
+        # Build HF resolve URL
+        resolve_url = (
+            f"https://huggingface.co/{repo}/resolve/{revision}/{filename}"
+        )
+
+        # Set up progress callback if download_id is provided
+        progress_callback = None
+        if download_id:
+
+            async def _progress_callback(
+                progress: float | DownloadProgress,
+                snapshot: DownloadProgress | None = None,
+            ) -> None:
+                percent = 0.0
+                metrics = snapshot if isinstance(snapshot, DownloadProgress) else None
+
+                if isinstance(progress, DownloadProgress):
+                    percent = progress.percent_complete
+                    metrics = progress
+                elif isinstance(snapshot, DownloadProgress):
+                    percent = snapshot.percent_complete
+                else:
+                    percent = float(progress)
+
+                broadcast: dict[str, Any] = {
+                    "status": "progress",
+                    "progress": round(percent),
+                }
+                if metrics:
+                    broadcast["bytes_downloaded"] = metrics.bytes_downloaded
+                    broadcast["total_bytes"] = metrics.total_bytes
+                    broadcast["bytes_per_second"] = metrics.bytes_per_second
+
+                await ws_manager.broadcast_download_progress(download_id, broadcast)
+
+            progress_callback = _progress_callback
+
+        # Respect download backend setting (aria2 vs default)
+        download_backend = (
+            get_settings_manager().get("download_backend", "default")
+        )
+
+        if download_backend == "aria2":
+            aria2 = await Aria2Downloader.get_instance()
+            aid = download_id or f"hf_{repo}_{filename}"
+            try:
+                hf_success, hf_result = await aria2.download_file(
+                    url=resolve_url,
+                    save_path=dest_path,
+                    download_id=aid,
+                    progress_callback=progress_callback,
+                )
+                if hf_success:
+                    await _save_hf_metadata(dest_path, repo, model_root)
+                    return web.json_response({
+                        "success": True,
+                        "message": f"Downloaded to {dest_path}",
+                        "path": dest_path,
+                    })
+                else:
+                    return web.json_response(
+                        {"success": False, "error": hf_result or "aria2 download failed"},
+                        status=500,
+                    )
+            except Exception as exc:
+                logger.error("HF download (aria2) failed: %s", exc)
+                return web.json_response(
+                    {"success": False, "error": str(exc)}, status=500
+                )
+
+        # Default: use built-in aiohttp Downloader
+        downloader = await get_downloader()
+        try:
+            success, result = await downloader.download_file(
+                url=resolve_url,
+                save_path=dest_path,
+                use_auth=False,
+                allow_resume=True,
+                progress_callback=progress_callback,
+            )
+            if success:
+                await _save_hf_metadata(dest_path, repo, model_root)
+                return web.json_response({
+                    "success": True,
+                    "message": f"Downloaded to {result}",
+                    "path": result,
+                })
+            else:
+                return web.json_response(
+                    {"success": False, "error": result or "Download failed"},
+                    status=500,
+                )
+        except Exception as exc:
+            logger.error("HF download failed: %s", exc)
+            return web.json_response(
+                {"success": False, "error": str(exc)}, status=500
+            )

py/routes/handlers/misc_handlers.py

@@ -38,6 +38,12 @@ from ...services.settings_manager import get_settings_manager
 from ...services.websocket_manager import ws_manager
 from ...services.downloader import get_downloader
 from ...services.errors import ResourceNotFoundError
+from ...services.llm_service import (
+    PROVIDER_PRESETS,
+    fetch_ollama_models,
+    get_all_provider_models,
+    get_provider_model_ids,
+)
 from ...services.cache_health_monitor import CacheHealthMonitor, CacheHealthStatus
 from ...utils.models import BaseModelMetadata
 from ...utils.constants import (
@@ -48,8 +54,13 @@ from ...utils.constants import (
     SUPPORTED_MEDIA_EXTENSIONS,
     VALID_LORA_TYPES,
 )
+from .hf_handlers import HfHandler
+from .agent_handlers import AgentHandler
 from ...utils.civitai_utils import rewrite_preview_url
-from ...utils.example_images_paths import is_valid_example_images_root
+from ...utils.example_images_paths import (
+    find_non_compliant_items_in_example_images_root,
+    is_valid_example_images_root,
+)
 from ...utils.lora_metadata import extract_trained_words
 from ...utils.session_logging import get_standalone_session_log_snapshot
 from ...utils.usage_stats import UsageStats
@@ -411,9 +422,10 @@ class PromptServerProtocol(Protocol):
     """Subset of PromptServer used by the handlers."""
 
     instance: "PromptServerProtocol"
+    sockets: dict  # maps clientId (sid) → WebSocketResponse
 
     def send_sync(
-        self, event: str, payload: dict
+        self, event: str, payload: dict | None = None, sid: str | None = None
     ) -> None:  # pragma: no cover - protocol
         ...
 
@@ -468,89 +480,167 @@ class BackupServiceProtocol(Protocol):
 
 
 class NodeRegistry:
-    """Thread-safe registry for tracking LoRA nodes in active workflows."""
+    """Thread-safe registry for tracking LoRA nodes across ComfyUI tabs.
+
+    Each connected ComfyUI browser tab (identified by its ``sid`` / ``clientId``)
+    registers its own set of workflow nodes.  Queries merge all known tabs into
+    a single result so that the calling LM panel always sees *every* available
+    target node, regardless of which tab responded fastest.
+    """
 
     def __init__(self) -> None:
         self._lock = asyncio.Lock()
-        self._nodes: Dict[str, dict] = {}
-        self._registry_updated = asyncio.Event()
+        # sid → {unique_id → node_info}
+        self._tab_nodes: Dict[str, Dict[str, dict]] = {}
+        self._ready = asyncio.Event()
+        self._waiting_clients: set[str] = set()
+
+    @property
+    def pending_client_count(self) -> int:
+        """Number of clients that have not yet responded in the current refresh cycle."""
+        return len(self._waiting_clients)
+
+    # ------------------------------------------------------------------
+    # Helpers to build one node dict (extracted so it's reused for each tab)
+    # ------------------------------------------------------------------
+    @staticmethod
+    def _build_node_dict(node: dict) -> dict:
+        node_id = node["node_id"]
+        graph_id = str(node["graph_id"])
+        unique_id = f"{graph_id}:{node_id}"
+        node_type = node.get("type", "")
+        type_id = NODE_TYPES.get(node_type, 0)
+        bgcolor = node.get("bgcolor") or DEFAULT_NODE_COLOR
+
+        raw_capabilities = node.get("capabilities")
+        capabilities: dict = {}
+        if isinstance(raw_capabilities, dict):
+            capabilities = dict(raw_capabilities)
+
+        raw_widget_names: list | None = node.get("widget_names")
+        if not isinstance(raw_widget_names, list):
+            capability_widget_names = capabilities.get("widget_names")
+            raw_widget_names = (
+                capability_widget_names
+                if isinstance(capability_widget_names, list)
+                else None
+            )
+
+        widget_names: list[str] = []
+        if isinstance(raw_widget_names, list):
+            widget_names = [
+                str(widget_name)
+                for widget_name in raw_widget_names
+                if isinstance(widget_name, str) and widget_name
+            ]
+
+        if widget_names:
+            capabilities["widget_names"] = widget_names
+        else:
+            capabilities.pop("widget_names", None)
+
+        if "supports_lora" in capabilities:
+            capabilities["supports_lora"] = bool(capabilities["supports_lora"])
+
+        comfy_class = node.get("comfy_class")
+        if not isinstance(comfy_class, str) or not comfy_class:
+            comfy_class = node_type if isinstance(node_type, str) else None
+
+        return {
+            "id": node_id,
+            "graph_id": graph_id,
+            "graph_name": node.get("graph_name"),
+            "unique_id": unique_id,
+            "bgcolor": bgcolor,
+            "title": node.get("title"),
+            "type": type_id,
+            "type_name": node_type,
+            "comfy_class": comfy_class,
+            "capabilities": capabilities,
+            "widget_names": widget_names,
+            "mode": node.get("mode"),
+            "marker_role": node.get("marker_role"),
+        }
+
+    # ------------------------------------------------------------------
+    # Public API
+    # ------------------------------------------------------------------
+    async def register_nodes(self, sid: str, nodes: list[dict]) -> None:
+        """Register/replace the node list for a single ComfyUI tab (identified by *sid*)."""
+        tab_nodes: dict[str, dict] = {}
+        for node in nodes:
+            nd = self._build_node_dict(node)
+            tab_nodes[nd["unique_id"]] = nd
 
-    async def register_nodes(self, nodes: list[dict]) -> None:
         async with self._lock:
-            self._nodes.clear()
-            for node in nodes:
-                node_id = node["node_id"]
-                graph_id = str(node["graph_id"])
-                unique_id = f"{graph_id}:{node_id}"
-                node_type = node.get("type", "")
-                type_id = NODE_TYPES.get(node_type, 0)
-                bgcolor = node.get("bgcolor") or DEFAULT_NODE_COLOR
-                raw_capabilities = node.get("capabilities")
-                capabilities: dict = {}
-                if isinstance(raw_capabilities, dict):
-                    capabilities = dict(raw_capabilities)
+            prev_count = len(self._tab_nodes.get(sid, {}))
+            self._tab_nodes[sid] = tab_nodes
+            self._waiting_clients.discard(sid)
+            if not self._waiting_clients:
+                self._ready.set()
+            total_tabs = len(self._tab_nodes)
 
-                raw_widget_names: list | None = node.get("widget_names")
-                if not isinstance(raw_widget_names, list):
-                    capability_widget_names = capabilities.get("widget_names")
-                    raw_widget_names = (
-                        capability_widget_names
-                        if isinstance(capability_widget_names, list)
-                        else None
-                    )
+        if len(nodes) != prev_count or len(nodes) > 0:
+            logger.debug(
+                "[LM:Registry] stored %s nodes (was %s) for client %s (total tabs: %s)",
+                len(nodes), prev_count, sid, total_tabs,
+            )
 
-                widget_names: list[str] = []
-                if isinstance(raw_widget_names, list):
-                    widget_names = [
-                        str(widget_name)
-                        for widget_name in raw_widget_names
-                        if isinstance(widget_name, str) and widget_name
-                    ]
+    def prepare_for_refresh(self, active_sids: list[str]) -> None:
+        """Set the list of client IDs we expect to hear from during the next refresh cycle."""
+        self._ready.clear()
+        self._waiting_clients = set(active_sids)
 
-                if widget_names:
-                    capabilities["widget_names"] = widget_names
-                else:
-                    capabilities.pop("widget_names", None)
-
-                if "supports_lora" in capabilities:
-                    capabilities["supports_lora"] = bool(capabilities["supports_lora"])
-
-                comfy_class = node.get("comfy_class")
-                if not isinstance(comfy_class, str) or not comfy_class:
-                    comfy_class = node_type if isinstance(node_type, str) else None
-
-                self._nodes[unique_id] = {
-                    "id": node_id,
-                    "graph_id": graph_id,
-                    "graph_name": node.get("graph_name"),
-                    "unique_id": unique_id,
-                    "bgcolor": bgcolor,
-                    "title": node.get("title"),
-                    "type": type_id,
-                    "type_name": node_type,
-                    "comfy_class": comfy_class,
-                    "capabilities": capabilities,
-                    "widget_names": widget_names,
-                    "mode": node.get("mode"),
-                }
-            logger.debug("Registered %s nodes in registry", len(nodes))
-            self._registry_updated.set()
-
-    async def get_registry(self) -> dict:
-        async with self._lock:
-            return {
-                "nodes": dict(self._nodes),
-                "node_count": len(self._nodes),
-            }
-
-    async def wait_for_update(self, timeout: float = 1.0) -> bool:
-        self._registry_updated.clear()
+    async def wait_for_all(self, timeout: float = 2.0) -> bool:
+        """Block until every client in the current waiting set has responded
+        (or *timeout* seconds elapse).  Returns ``True`` if all responded."""
+        if not self._waiting_clients:
+            return True
         try:
-            await asyncio.wait_for(self._registry_updated.wait(), timeout=timeout)
+            await asyncio.wait_for(self._ready.wait(), timeout=timeout)
             return True
         except asyncio.TimeoutError:
             return False
 
+    async def get_merged_registry(self, active_sids: set[str] | None = None) -> dict:
+        """Return the union of all known tab nodes, pruning any tab that is no
+        longer connected."""
+        async with self._lock:
+            # Garbage-collect stale entries (disconnected tabs)
+            stale_sids = []
+            if active_sids is not None:
+                for sid in list(self._tab_nodes):
+                    if sid not in active_sids:
+                        stale_sids.append(sid)
+                        del self._tab_nodes[sid]
+            if stale_sids:
+                logger.debug(
+                    "[LM:Registry] GC pruned %s disconnected tabs: %s",
+                    len(stale_sids), stale_sids,
+                )
+
+            merged: dict[str, dict] = {}
+            tab_info: dict[str, dict] = {}
+            for sid, nodes in self._tab_nodes.items():
+                tab_info[sid] = {
+                    "node_count": len(nodes),
+                    "graph_names": list(
+                        {
+                            n.get("graph_name")
+                            for n in nodes.values()
+                            if n.get("graph_name")
+                        }
+                    ),
+                }
+                merged.update(nodes)
+
+            return {
+                "nodes": merged,
+                "node_count": len(merged),
+                "tab_count": len(self._tab_nodes),
+                "tabs": tab_info,
+            }
+
 
 class HealthCheckHandler:
     async def health_check(self, request: web.Request) -> web.Response:
@@ -1328,6 +1418,10 @@ class SettingsHandler:
             "folder_paths",
             "libraries",
             "active_library",
+            # Sensitive — never expose the actual value to the frontend;
+            # frontend receives a boolean instead (*_set).
+            "civitai_api_key",
+            "llm_api_key",
         }
     )
 
@@ -1382,6 +1476,11 @@ class SettingsHandler:
                     value = self._settings.get(key)
                     if value is not None:
                         response_data[key] = value
+            # Sensitive fields: only expose a boolean indicating whether set
+            raw_key = self._settings.get("civitai_api_key")
+            response_data["civitai_api_key_set"] = bool(raw_key)
+            raw_llm_key = self._settings.get("llm_api_key")
+            response_data["llm_api_key_set"] = bool(raw_llm_key)
             settings_file = getattr(self._settings, "settings_file", None)
             if settings_file:
                 response_data["settings_file"] = settings_file
@@ -1463,6 +1562,11 @@ class SettingsHandler:
                             {"success": False, "error": validation_error}
                         )
 
+                if key == "update_channel" and value not in ("release", "nightly"):
+                    return web.json_response(
+                        {"success": False, "error": "update_channel must be 'release' or 'nightly'"}
+                    )
+
                 if value == "__DELETE__" and key in (
                     "proxy_username",
                     "proxy_password",
@@ -1471,7 +1575,11 @@ class SettingsHandler:
                 else:
                     self._settings.set(key, value)
 
-                if key == "enable_metadata_archive_db":
+                if key in (
+                    "enable_metadata_archive_db",
+                    "enable_civarchive_api",
+                    "metadata_provider_order",
+                ):
                     await self._metadata_provider_updater()
 
                 if key in self._PROXY_KEYS:
@@ -1486,18 +1594,78 @@ class SettingsHandler:
             logger.error("Error updating settings: %s", exc, exc_info=True)
             return web.Response(status=500, text=str(exc))
 
+    async def get_llm_models(self, request: web.Request) -> web.Response:
+        """Return the model list for a provider.
+
+        For ``ollama`` the list is fetched live from the local Ollama API
+        (only models actually pulled locally are shown).  For all other
+        providers the opencode model catalog is used.
+
+        Query parameters:
+            provider (required): Internal provider id (``openai``, ``ollama``, etc.).
+
+        Returns:
+            ``{"success": true, "models": ["gpt-4o", ...]}``.
+        """
+        provider_id = request.query.get("provider", "").strip()
+        if not provider_id:
+            return web.json_response(
+                {"success": False, "error": "provider query parameter is required", "models": []},
+                status=400,
+            )
+
+        try:
+            if provider_id == "ollama":
+                api_base = request.query.get("api_base", "").strip() or self._settings.get("llm_api_base", "")
+                if not api_base:
+                    api_base = "http://localhost:11434/v1"
+                models = await fetch_ollama_models(api_base)
+            else:
+                models = await get_provider_model_ids(provider_id)
+            return web.json_response({"success": True, "models": models})
+        except Exception as exc:
+            logger.warning("get_llm_models failed for %s: %s", provider_id, exc)
+            return web.json_response(
+                {"success": False, "error": str(exc), "models": []},
+                status=500,
+            )
+
     def _validate_example_images_path(self, folder_path: str) -> str | None:
         if not os.path.exists(folder_path):
             return f"Path does not exist: {folder_path}"
         if not os.path.isdir(folder_path):
             return "Please set a dedicated folder for example images."
         if not self._is_dedicated_example_images_folder(folder_path):
+            offending = find_non_compliant_items_in_example_images_root(folder_path)
+            if offending:
+                items_str = ", ".join(repr(item) for item in offending[:5])
+                if len(offending) > 5:
+                    items_str += f" … and {len(offending) - 5} more"
+                return (
+                    f"The folder contains items that are not valid example image "
+                    f"folders: {items_str}. Please use a dedicated, empty folder "
+                    f"for example images to prevent accidental data loss."
+                )
             return "Please set a dedicated folder for example images."
         return None
 
     def _is_dedicated_example_images_folder(self, folder_path: str) -> bool:
         return is_valid_example_images_root(folder_path)
 
+    async def get_provider_models(self, request: web.Request) -> web.Response:
+        """Return the model catalog for all preset providers.
+
+        This endpoint is called asynchronously by the settings UI so that
+        page rendering never blocks on the remote model catalog fetch.
+        """
+        catalog_provider_ids = [p for p in PROVIDER_PRESETS if p != "custom"]
+        try:
+            provider_models = await get_all_provider_models(catalog_provider_ids)
+            return web.json_response({"success": True, "models": provider_models})
+        except Exception as exc:
+            logger.warning("Failed to fetch provider models: %s", exc)
+            return web.json_response({"success": False, "models": {}, "error": str(exc)})
+
 
 class UsageStatsHandler:
     def __init__(self, usage_stats_factory: UsageStatsFactory = UsageStats) -> None:
@@ -1625,6 +1793,124 @@ class LoraCodeHandler:
             logger.error("Failed to update lora code: %s", exc, exc_info=True)
             return web.json_response({"success": False, "error": str(exc)}, status=500)
 
+    async def get_update_lora_code(self, request: web.Request) -> web.Response:
+        """GET version of update_lora_code — reads parameters from query string.
+
+        Query params:
+          lora_code (required)  — the LoRA syntax to send
+          mode     (optional)   — "append" (default) or "replace"
+          node_id  (repeatable) — target node id(s), e.g. node_id=3&node_id=5
+          node_ids (optional)   — JSON-encoded array for complex references with graph_id:
+                                   [{"node_id":3,"graph_id":"g1"}, ...]
+        """
+        try:
+            node_ids_raw = request.query.get("node_ids")
+            node_id_list = request.query.getall("node_id", [])
+            lora_code = request.query.get("lora_code", "")
+            mode = request.query.get("mode", "append")
+
+            if not lora_code:
+                return web.json_response(
+                    {"success": False, "error": "Missing lora_code parameter"},
+                    status=400,
+                )
+
+            node_ids = None
+            if node_ids_raw:
+                try:
+                    node_ids = json.loads(node_ids_raw)
+                except (json.JSONDecodeError, TypeError):
+                    return web.json_response(
+                        {"success": False, "error": "node_ids must be a valid JSON array"},
+                        status=400,
+                    )
+                if not isinstance(node_ids, list) or not node_ids:
+                    return web.json_response(
+                        {"success": False, "error": "node_ids must be a non-empty JSON array"},
+                        status=400,
+                    )
+            elif node_id_list:
+                node_ids = node_id_list
+
+            results = []
+            if node_ids is None:
+                try:
+                    self._prompt_server.instance.send_sync(
+                        "lora_code_update",
+                        {"id": -1, "lora_code": lora_code, "mode": mode},
+                    )
+                    results.append({"node_id": "broadcast", "success": True})
+                except Exception as exc:  # pragma: no cover - defensive logging
+                    logger.error("Error broadcasting lora code: %s", exc)
+                    results.append(
+                        {"node_id": "broadcast", "success": False, "error": str(exc)}
+                    )
+            else:
+                for entry in node_ids:
+                    node_identifier = entry
+                    graph_identifier = None
+                    if isinstance(entry, dict):
+                        node_identifier = entry.get("node_id")
+                        graph_identifier = entry.get("graph_id")
+
+                    if node_identifier is None:
+                        results.append(
+                            {
+                                "node_id": node_identifier,
+                                "graph_id": graph_identifier,
+                                "success": False,
+                                "error": "Missing node_id parameter",
+                            }
+                        )
+                        continue
+
+                    try:
+                        parsed_node_id = int(node_identifier)
+                    except (TypeError, ValueError):
+                        parsed_node_id = node_identifier
+
+                    payload = {
+                        "id": parsed_node_id,
+                        "lora_code": lora_code,
+                        "mode": mode,
+                    }
+
+                    if graph_identifier is not None:
+                        payload["graph_id"] = str(graph_identifier)
+
+                    try:
+                        self._prompt_server.instance.send_sync(
+                            "lora_code_update",
+                            payload,
+                        )
+                        results.append(
+                            {
+                                "node_id": parsed_node_id,
+                                "graph_id": payload.get("graph_id"),
+                                "success": True,
+                            }
+                        )
+                    except Exception as exc:  # pragma: no cover - defensive logging
+                        logger.error(
+                            "Error sending lora code to node %s (graph %s): %s",
+                            parsed_node_id,
+                            graph_identifier,
+                            exc,
+                        )
+                        results.append(
+                            {
+                                "node_id": parsed_node_id,
+                                "graph_id": payload.get("graph_id"),
+                                "success": False,
+                                "error": str(exc),
+                            }
+                        )
+
+            return web.json_response({"success": True, "results": results})
+        except Exception as exc:  # pragma: no cover - defensive logging
+            logger.error("Failed to update lora code (GET): %s", exc, exc_info=True)
+            return web.json_response({"success": False, "error": str(exc)}, status=500)
+
 
 class TrainedWordsHandler:
     async def get_trained_words(self, request: web.Request) -> web.Response:
@@ -2970,15 +3256,28 @@ class NodeRegistryHandler:
         self._node_registry = node_registry
         self._prompt_server = prompt_server
         self._standalone_mode = standalone_mode
+        self._refresh_lock = asyncio.Lock()
+        self._last_slow_path_ts: float = 0.0
 
     async def register_nodes(self, request: web.Request) -> web.Response:
         try:
             data = await request.json()
             nodes = data.get("nodes", [])
+            client_id = data.get("client_id")
             if not isinstance(nodes, list):
                 return web.json_response(
                     {"success": False, "error": "nodes must be a list"}, status=400
                 )
+
+            if not isinstance(client_id, str) or not client_id:
+                return web.json_response(
+                    {
+                        "success": False,
+                        "error": "Missing client_id parameter",
+                    },
+                    status=400,
+                )
+
             for index, node in enumerate(nodes):
                 if not isinstance(node, dict):
                     return web.json_response(
@@ -3005,7 +3304,12 @@ class NodeRegistryHandler:
                     )
                 graph_name = node.get("graph_name")
                 try:
-                    node["node_id"] = int(node_id)
+                    # Handle compound node IDs from expanded group subgraphs,
+                    # e.g. "252:0" → 0 (parent scope is already in graph_id)
+                    if isinstance(node_id, str) and ":" in node_id:
+                        node["node_id"] = int(node_id.rsplit(":", 1)[-1])
+                    else:
+                        node["node_id"] = int(node_id)
                 except (TypeError, ValueError):
                     return web.json_response(
                         {
@@ -3022,7 +3326,7 @@ class NodeRegistryHandler:
                 else:
                     node["graph_name"] = str(graph_name)
 
-            await self._node_registry.register_nodes(nodes)
+            await self._node_registry.register_nodes(client_id, nodes)
             return web.json_response(
                 {
                     "success": True,
@@ -3046,33 +3350,110 @@ class NodeRegistryHandler:
                     status=503,
                 )
 
-            try:
-                self._prompt_server.instance.send_sync("lora_registry_refresh", {})
-                logger.debug("Sent registry refresh request to frontend")
-            except Exception as exc:
-                logger.error("Failed to send registry refresh message: %s", exc)
-                return web.json_response(
-                    {
-                        "success": False,
-                        "error": "Communication Error",
-                        "message": f"Failed to communicate with ComfyUI frontend: {exc}",
-                    },
-                    status=500,
-                )
+            current_sids = set(self._prompt_server.instance.sockets.keys())
 
-            registry_updated = await self._node_registry.wait_for_update(timeout=1.0)
-            if not registry_updated:
-                logger.warning("Registry refresh timeout after 1 second")
+            # Fast path: if the frontend has already pushed node data (via
+            # afterConfigureGraph / graphChanged hooks), return it immediately
+            # without triggering a WebSocket round-trip.
+            registry_info = await self._node_registry.get_merged_registry(
+                active_sids=current_sids
+            )
+            if registry_info["tab_count"] > 0:
+                logger.debug(
+                    "[LM:Registry] fast path: %s nodes across %s tabs %s",
+                    registry_info["node_count"],
+                    registry_info["tab_count"],
+                    dict(registry_info.get("tabs", {})),
+                )
+                return web.json_response({"success": True, "data": registry_info})
+
+            # Slow path: registry is empty — trigger refresh via WebSocket.
+            # Serialize with an async lock so concurrent callers don't all
+            # trigger separate WS refresh cycles.  The second caller will
+            # re-check the fast path and (usually) find populated data.
+            async with self._refresh_lock:
+                # Re-check after acquiring the lock — another concurrent call
+                # may have populated the cache while we were waiting.
+                registry_info = await self._node_registry.get_merged_registry(
+                    active_sids=current_sids
+                )
+                if registry_info["tab_count"] > 0:
+                    logger.debug(
+                        "[LM:Registry] fast path after lock wait: %s nodes across %s tabs",
+                        registry_info["node_count"],
+                        registry_info["tab_count"],
+                    )
+                    return web.json_response({"success": True, "data": registry_info})
+
+                # Cooldown: if the slow path ran recently (< 2 s) and
+                # returned empty, skip another WS round-trip.
+                elapsed = time.monotonic() - self._last_slow_path_ts
+                if elapsed < 2.0:
+                    logger.debug(
+                        "[LM:Registry] slow path cooldown (%.1fs since last refresh), returning empty",
+                        elapsed,
+                    )
+                    return web.json_response(
+                        {
+                            "success": False,
+                            "error": "Empty Registry",
+                            "message": "No workflow nodes found — ensure ComfyUI is open and the extension is loaded.",
+                        },
+                        status=408,
+                    )
+
+                logger.debug(
+                    "[LM:Registry] slow path: cache empty, triggering WS refresh (%s connected tabs: %s)",
+                    len(current_sids), list(current_sids)[:5],
+                )
+                active_sids = list(current_sids)
+                self._node_registry.prepare_for_refresh(active_sids)
+
+                try:
+                    self._prompt_server.instance.send_sync("lora_registry_refresh", {})
+                    logger.debug(
+                        "Sent registry refresh request (expecting %s clients)", len(active_sids)
+                    )
+                except Exception as exc:
+                    logger.error("Failed to send registry refresh message: %s", exc)
+                    return web.json_response(
+                        {
+                            "success": False,
+                            "error": "Communication Error",
+                            "message": f"Failed to communicate with ComfyUI frontend: {exc}",
+                        },
+                        status=500,
+                    )
+
+                if not await self._node_registry.wait_for_all(timeout=0.5):
+                    logger.warning(
+                        "Registry refresh timeout after 0.5s (%s/%s clients responded)",
+                        len(active_sids) - self._node_registry.pending_client_count,
+                        len(active_sids),
+                    )
+
+                # Re-read current sockets after the wait: a tab may have connected
+                # while we were waiting, and we don't want to garbage-collect it.
+                current_sids = set(self._prompt_server.instance.sockets.keys())
+                registry_info = await self._node_registry.get_merged_registry(
+                    active_sids=current_sids
+                )
+                self._last_slow_path_ts = time.monotonic()
+
+            if registry_info["node_count"] == 0:
+                logger.debug(
+                    "[LM:Registry] refresh OK — %s connected tab(s) but 0 compatible nodes found",
+                    registry_info["tab_count"],
+                )
                 return web.json_response(
                     {
                         "success": False,
-                        "error": "Timeout Error",
-                        "message": "Registry refresh timeout - ComfyUI frontend may not be responsive",
+                        "error": "Empty Registry",
+                        "message": "No workflow nodes found — ensure ComfyUI is open and the extension is loaded.",
                     },
                     status=408,
                 )
 
-            registry_info = await self._node_registry.get_registry()
             return web.json_response({"success": True, "data": registry_info})
         except Exception as exc:  # pragma: no cover - defensive logging
             logger.error("Failed to get registry: %s", exc, exc_info=True)
@@ -3085,17 +3466,21 @@ class NodeRegistryHandler:
         try:
             data = await request.json()
             widget_name = data.get("widget_name")
+            action = data.get("action")
             value = data.get("value")
             mode = data.get("mode", "replace")
             node_ids = data.get("node_ids")
 
-            if not isinstance(widget_name, str) or not widget_name:
+            if not action and (not isinstance(widget_name, str) or not widget_name):
                 return web.json_response(
-                    {"success": False, "error": "Missing widget_name parameter"},
+                    {
+                        "success": False,
+                        "error": "Missing parameter: provide either 'action' or 'widget_name'",
+                    },
                     status=400,
                 )
 
-            if not isinstance(value, str) or not value:
+            if value is None or (isinstance(value, str) and not value):
                 return web.json_response(
                     {"success": False, "error": "Missing value parameter"}, status=400
                 )
@@ -3130,12 +3515,15 @@ class NodeRegistryHandler:
                 except (TypeError, ValueError):
                     parsed_node_id = node_identifier
 
-                payload = {
+                payload: dict = {
                     "id": parsed_node_id,
-                    "widget_name": widget_name,
                     "value": value,
                     "mode": mode,
                 }
+                if action:
+                    payload["action"] = action
+                if widget_name:
+                    payload["widget_name"] = widget_name
 
                 if graph_identifier is not None:
                     payload["graph_id"] = str(graph_identifier)
@@ -3170,6 +3558,130 @@ class NodeRegistryHandler:
             logger.error("Failed to update node widget: %s", exc, exc_info=True)
             return web.json_response({"success": False, "error": str(exc)}, status=500)
 
+    async def get_update_node_widget(self, request: web.Request) -> web.Response:
+        """GET version of update_node_widget — reads parameters from query string.
+
+        Query params:
+          widget_name  (optional)   — the widget name to update (required unless action is set)
+          action       (optional)   — alternative action, e.g. "inject_text" (required unless widget_name is set)
+          value        (required)   — the value to set
+          mode         (optional)   — "replace" (default) or "append"
+          node_id      (repeatable) — target node id(s), e.g. node_id=3&node_id=5
+          node_ids     (optional)   — JSON-encoded array for complex references:
+                                       [{"node_id":3,"graph_id":"g1"}, ...]
+        """
+        try:
+            widget_name = request.query.get("widget_name")
+            action = request.query.get("action")
+            value = request.query.get("value")
+            mode = request.query.get("mode", "replace")
+            node_ids_raw = request.query.get("node_ids")
+            node_id_list = request.query.getall("node_id", [])
+
+            if not action and (not isinstance(widget_name, str) or not widget_name):
+                return web.json_response(
+                    {
+                        "success": False,
+                        "error": "Missing parameter: provide either 'action' or 'widget_name'",
+                    },
+                    status=400,
+                )
+
+            if value is None or (isinstance(value, str) and not value):
+                return web.json_response(
+                    {"success": False, "error": "Missing value parameter"}, status=400
+                )
+
+            node_ids = None
+            if node_ids_raw:
+                try:
+                    node_ids = json.loads(node_ids_raw)
+                except (json.JSONDecodeError, TypeError):
+                    return web.json_response(
+                        {"success": False, "error": "node_ids must be a valid JSON array"},
+                        status=400,
+                    )
+                if not isinstance(node_ids, list) or not node_ids:
+                    return web.json_response(
+                        {"success": False, "error": "node_ids must be a non-empty JSON array"},
+                        status=400,
+                    )
+            elif node_id_list:
+                node_ids = node_id_list
+
+            if not isinstance(node_ids, list) or not node_ids:
+                return web.json_response(
+                    {"success": False, "error": "node_ids must be a non-empty list"},
+                    status=400,
+                )
+
+            results = []
+            for entry in node_ids:
+                node_identifier = entry
+                graph_identifier = None
+                if isinstance(entry, dict):
+                    node_identifier = entry.get("node_id")
+                    graph_identifier = entry.get("graph_id")
+
+                if node_identifier is None:
+                    results.append(
+                        {
+                            "node_id": node_identifier,
+                            "graph_id": graph_identifier,
+                            "success": False,
+                            "error": "Missing node_id parameter",
+                        }
+                    )
+                    continue
+
+                try:
+                    parsed_node_id = int(node_identifier)
+                except (TypeError, ValueError):
+                    parsed_node_id = node_identifier
+
+                payload: dict = {
+                    "id": parsed_node_id,
+                    "value": value,
+                    "mode": mode,
+                }
+                if action:
+                    payload["action"] = action
+                if widget_name:
+                    payload["widget_name"] = widget_name
+
+                if graph_identifier is not None:
+                    payload["graph_id"] = str(graph_identifier)
+
+                try:
+                    self._prompt_server.instance.send_sync("lm_widget_update", payload)
+                    results.append(
+                        {
+                            "node_id": parsed_node_id,
+                            "graph_id": payload.get("graph_id"),
+                            "success": True,
+                        }
+                    )
+                except Exception as exc:  # pragma: no cover - defensive logging
+                    logger.error(
+                        "Error sending widget update to node %s (graph %s): %s",
+                        parsed_node_id,
+                        graph_identifier,
+                        exc,
+                    )
+                    results.append(
+                        {
+                            "node_id": parsed_node_id,
+                            "graph_id": payload.get("graph_id"),
+                            "success": False,
+                            "error": str(exc),
+                        }
+                    )
+
+            return web.json_response({"success": True, "results": results})
+        except Exception as exc:  # pragma: no cover - defensive logging
+            logger.error("Failed to update node widget (GET): %s", exc, exc_info=True)
+            return web.json_response({"success": False, "error": str(exc)}, status=500)
+
 
 class MiscHandlerSet:
     """Aggregate handlers into a lookup compatible with the registrar."""
@@ -3194,6 +3706,8 @@ class MiscHandlerSet:
         doctor: DoctorHandler,
         example_workflows: ExampleWorkflowsHandler,
         base_model: BaseModelHandlerSet,
+        hf_handler: HfHandler | None = None,
+        agent_handler: AgentHandler | None = None,
     ) -> None:
         self.health = health
         self.settings = settings
@@ -3212,6 +3726,8 @@ class MiscHandlerSet:
         self.doctor = doctor
         self.example_workflows = example_workflows
         self.base_model = base_model
+        self.hf_handler = hf_handler
+        self.agent_handler = agent_handler
 
     def to_route_mapping(
         self,
@@ -3227,13 +3743,17 @@ class MiscHandlerSet:
             "get_priority_tags": self.settings.get_priority_tags,
             "get_settings_libraries": self.settings.get_libraries,
             "activate_library": self.settings.activate_library,
+            "get_llm_models": self.settings.get_llm_models,
+            "get_provider_models": self.settings.get_provider_models,
             "update_usage_stats": self.usage_stats.update_usage_stats,
             "get_usage_stats": self.usage_stats.get_usage_stats,
             "update_lora_code": self.lora_code.update_lora_code,
+            "get_update_lora_code": self.lora_code.get_update_lora_code,
             "get_trained_words": self.trained_words.get_trained_words,
             "get_model_example_files": self.model_examples.get_model_example_files,
             "register_nodes": self.node_registry.register_nodes,
             "update_node_widget": self.node_registry.update_node_widget,
+            "get_update_node_widget": self.node_registry.get_update_node_widget,
             "get_registry": self.node_registry.get_registry,
             "check_model_exists": self.model_library.check_model_exists,
             "check_models_exist": self.model_library.check_models_exist,
@@ -3257,6 +3777,14 @@ class MiscHandlerSet:
             "get_supporters": self.supporters.get_supporters,
             "get_example_workflows": self.example_workflows.get_example_workflows,
             "get_example_workflow": self.example_workflows.get_example_workflow,
+            # Hugging Face handlers
+            "get_hf_repo_files": self.hf_handler.get_hf_repo_files,
+            "download_hf_model": self.hf_handler.download_hf_model,
+            "set_hf_url": self.hf_handler.set_hf_url,
+            # Agent skill handlers
+            "get_agent_skills": self.agent_handler.get_agent_skills,
+            "execute_agent_skill": self.agent_handler.execute_agent_skill,
+            "cancel_agent_skill": self.agent_handler.cancel_agent_skill,
             # Base model handlers
             "get_base_models": self.base_model.get_base_models,
             "refresh_base_models": self.base_model.refresh_base_models,

py/routes/handlers/model_handlers.py

@@ -154,6 +154,14 @@ class ModelPageView:
                 )
                 self._template_env._i18n_filter_added = True  # type: ignore[attr-defined]
 
+            from ...services.llm_service import PROVIDER_PRESETS
+
+            # Provider presets are embedded directly (local, no await needed).
+            # Provider model catalogs are fetched asynchronously by the
+            # frontend via GET /api/lm/llm/provider-models so page rendering
+            # never blocks on the remote model catalog (which can take up to
+            # 30s on cold cache).
+
             template_context = {
                 "is_initializing": is_initializing,
                 "settings": self._settings,
@@ -161,6 +169,8 @@ class ModelPageView:
                 "folders": [],
                 "t": self._server_i18n.get_translation,
                 "version": self._get_app_version(),
+                "provider_presets_json": json.dumps(PROVIDER_PRESETS),
+                "provider_models_json": "{}",
             }
 
             if not is_initializing:
@@ -203,11 +213,17 @@ class ModelListingHandler:
             result = await self._service.get_paginated_data(**params)
 
             format_start = time.perf_counter()
+            formatted_raw = [
+                await self._service.format_response(entry)
+                for entry in result["items"]
+            ]
+            # Filter out None entries returned for corrupted cache rows (issue #730).
+            # Note: "total" intentionally remains the pre-filter count to reflect
+            # the true number of models in the cache; corrupted entries are rare
+            # and adjusting total would cause pagination drift on every page.
+            formatted_items = [item for item in formatted_raw if item is not None]
             formatted_result = {
-                "items": [
-                    await self._service.format_response(item)
-                    for item in result["items"]
-                ],
+                "items": formatted_items,
                 "total": result["total"],
                 "page": result["page"],
                 "page_size": result["page_size"],
@@ -233,14 +249,20 @@ class ModelListingHandler:
         start_time = time.perf_counter()
         try:
             params = self._parse_common_params(request)
+            # group_by_model is meaningless for excluded view; strip it
+            params.pop("group_by_model", None)
             result = await self._service.get_excluded_paginated_data(**params)
 
             format_start = time.perf_counter()
+            formatted_raw = [
+                await self._service.format_response(entry)
+                for entry in result["items"]
+            ]
+            # Filter out None entries returned for corrupted cache rows (issue #730).
+            # "total" stays at the pre-filter count; see get_models for rationale.
+            formatted_items = [item for item in formatted_raw if item is not None]
             formatted_result = {
-                "items": [
-                    await self._service.format_response(item)
-                    for item in result["items"]
-                ],
+                "items": formatted_items,
                 "total": result["total"],
                 "page": result["page"],
                 "page_size": result["page_size"],
@@ -366,6 +388,21 @@ class ModelListingHandler:
             request.query.get("name_pattern_use_regex", "false").lower() == "true"
         )
 
+        # Group-by-model flag: deduplicate versions sharing the same civitai modelId
+        group_by_model = (
+            request.query.get("group_by_model", "false").lower() == "true"
+        )
+
+        # View-local-versions filter: show all local versions of a specific model
+        # Accepts either a CivitAI modelId (int) or a HF group key like "hf:user/repo"
+        civitai_model_id = request.query.get("civitai_model_id")
+        if civitai_model_id is not None:
+            try:
+                civitai_model_id = int(civitai_model_id)
+            except (TypeError, ValueError):
+                # Keep as string — could be an HF group key (e.g. "hf:user/repo")
+                pass
+
         return {
             "page": page,
             "page_size": page_size,
@@ -389,6 +426,8 @@ class ModelListingHandler:
             "name_pattern_include": name_pattern_include,
             "name_pattern_exclude": name_pattern_exclude,
             "name_pattern_use_regex": name_pattern_use_regex,
+            "group_by_model": group_by_model,
+            "civitai_model_id": civitai_model_id,
             **self._parse_specific_params(request),
         }
 
@@ -500,6 +539,7 @@ class ModelManagementHandler:
                     # Update model_data with new hash
                     model_data["sha256"] = sha256
                     model_data["hash_status"] = "completed"
+                    hash_status = "completed"
                 else:
                     return web.json_response(
                         {"success": False, "error": "No SHA256 hash found"}, status=400
@@ -507,6 +547,32 @@ class ModelManagementHandler:
 
             await MetadataManager.hydrate_model_data(model_data)
 
+            # hydrate_model_data replaces model_data with .metadata.json content,
+            # which may lack sha256. Restore from cache and persist the fix.
+            if not model_data.get("sha256"):
+                if sha256:
+                    model_data["sha256"] = sha256
+                    model_data["hash_status"] = model_data.get("hash_status", hash_status)
+                    data_to_save = model_data.copy()
+                    data_to_save.pop("folder", None)
+                    await MetadataManager.save_metadata(file_path, data_to_save)
+                else:
+                    sha256 = await calculate_sha256(file_path)
+                    if sha256:
+                        model_data["sha256"] = sha256.lower()
+                        model_data["hash_status"] = "completed"
+                        data_to_save = model_data.copy()
+                        data_to_save.pop("folder", None)
+                        await MetadataManager.save_metadata(file_path, data_to_save)
+                    else:
+                        return web.json_response(
+                            {
+                                "success": False,
+                                "error": "Failed to compute SHA256 hash for model",
+                            },
+                            status=500,
+                        )
+
             success, error = await self._metadata_sync.fetch_and_update_model(
                 sha256=model_data["sha256"],
                 file_path=file_path,
@@ -516,15 +582,25 @@ class ModelManagementHandler:
             if not success:
                 return web.json_response({"success": False, "error": error})
 
-            formatted_metadata = await self._service.format_response(model_data)
-            return web.json_response({"success": True, "metadata": formatted_metadata})
+            formatted = await self._service.format_response(model_data)
+            if formatted is None:
+                return web.json_response(
+                    {"success": False, "error": "Model entry is corrupted (missing file_path)"},
+                    status=500,
+                )
+            return web.json_response({"success": True, "metadata": formatted})
         except Exception as exc:
             if is_expected_offline_error(str(exc)):
                 return web.json_response(
                     {"success": False, "error": OFFLINE_FRIENDLY_MESSAGE},
                     status=503,
                 )
-            self._logger.error("Error fetching from CivitAI: %s", exc, exc_info=True)
+            self._logger.error(
+                "Error fetching from CivitAI for %s: %s",
+                locals().get("file_path", "unknown"),
+                exc,
+                exc_info=True,
+            )
             return web.json_response({"success": False, "error": str(exc)}, status=500)
 
     async def relink_civitai(self, request: web.Request) -> web.Response:
@@ -931,6 +1007,8 @@ class ModelQueryHandler:
             limit = int(request.query.get("limit", "20"))
             if limit < 0:
                 limit = 20
+            elif limit > 200:
+                limit = 20
             top_tags = await self._service.get_top_tags(limit)
             return web.json_response({"success": True, "tags": top_tags})
         except Exception as exc:
@@ -939,6 +1017,22 @@ class ModelQueryHandler:
                 {"success": False, "error": "Internal server error"}, status=500
             )
 
+    async def search_tags(self, request: web.Request) -> web.Response:
+        try:
+            query = request.query.get("q", "")
+            limit = int(request.query.get("limit", "20"))
+            if limit < 0:
+                limit = 20
+            elif limit > 200:
+                limit = 20
+            tags = await self._service.search_tags(query, limit)
+            return web.json_response({"success": True, "tags": tags})
+        except Exception as exc:
+            self._logger.error("Error searching tags: %s", exc, exc_info=True)
+            return web.json_response(
+                {"success": False, "error": "Internal server error"}, status=500
+            )
+
     async def get_base_models(self, request: web.Request) -> web.Response:
         try:
             limit = int(request.query.get("limit", "20"))
@@ -1074,10 +1168,12 @@ class ModelQueryHandler:
                 # Sort: originals first, copies last
                 sorted_models = self._sort_duplicate_group(filtered)
 
-                # Format response
+                # Format response, filtering out corrupted entries (issue #730)
                 group = {"hash": sha256, "models": []}
                 for model in sorted_models:
-                    group["models"].append(await self._service.format_response(model))
+                    formatted = await self._service.format_response(model)
+                    if formatted is not None:
+                        group["models"].append(formatted)
 
                 # Only include groups with 2+ models after filtering
                 if len(group["models"]) > 1:
@@ -1194,9 +1290,9 @@ class ModelQueryHandler:
                         (m for m in cache.raw_data if m["file_path"] == path), None
                     )
                     if model:
-                        group["models"].append(
-                            await self._service.format_response(model)
-                        )
+                        formatted = await self._service.format_response(model)
+                        if formatted is not None:
+                            group["models"].append(formatted)
                 hash_val = self._service.scanner.get_hash_by_filename(filename)
                 if hash_val:
                     main_path = self._service.get_path_by_hash(hash_val)
@@ -1206,9 +1302,9 @@ class ModelQueryHandler:
                             None,
                         )
                         if main_model:
-                            group["models"].insert(
-                                0, await self._service.format_response(main_model)
-                            )
+                            formatted = await self._service.format_response(main_model)
+                            if formatted is not None:
+                                group["models"].insert(0, formatted)
                 if group["models"]:
                     result.append(group)
             return web.json_response(
@@ -1231,9 +1327,13 @@ class ModelQueryHandler:
                     text=f"{self._service.model_type.capitalize()} file name is required",
                     status=400,
                 )
-            notes = await self._service.get_model_notes(model_name)
-            if notes is not None:
-                return web.json_response({"success": True, "notes": notes})
+            result = await self._service.get_model_notes(model_name)
+            if result is not None:
+                return web.json_response({
+                    "success": True,
+                    "notes": result["notes"],
+                    "file_path": result["file_path"],
+                })
             return web.json_response(
                 {
                     "success": False,
@@ -1269,9 +1369,28 @@ class ModelQueryHandler:
                 }
                 if include_license_flags:
                     model_data = await self._service.get_model_info_by_name(model_name)
-                    license_flags = (model_data or {}).get("license_flags")
-                    if license_flags is not None:
-                        response_payload["license_flags"] = int(license_flags)
+                    # Only return license_flags when real CivitAI model license
+                    # data exists. This mirrors ModelModal's guard
+                    # (modelData?.civitai?.model) so the preview tooltip never
+                    # shows misleading license icons for HF or other models
+                    # without actual license metadata.
+                    civitai_data = (model_data or {}).get("civitai") or {}
+                    has_license_data = (
+                        isinstance(civitai_data, dict)
+                        and isinstance(civitai_data.get("model"), dict)
+                    )
+                    if has_license_data:
+                        license_flags = (model_data or {}).get("license_flags")
+                        if license_flags is not None:
+                            response_payload["license_flags"] = int(license_flags)
+                    # Include the user's license icon style preference so the
+                    # ComfyUI tooltip can pick the right set without a separate
+                    # API call.
+                    try:
+                        settings = get_settings_manager()
+                        response_payload["use_new_license_icons"] = settings.get("use_new_license_icons", True)
+                    except Exception:
+                        pass
                 return web.json_response(response_payload)
             return web.json_response(
                 {
@@ -1720,14 +1839,20 @@ class ModelDownloadHandler:
 
     async def delete_download_history_item(self, request: web.Request) -> web.Response:
         try:
-            item_id = int(request.query.get("id", "0"))
-            if not item_id:
+            download_id = request.query.get("download_id")
+            id_str = request.query.get("id")
+            item_id = int(id_str) if id_str else None
+
+            if not download_id and not item_id:
                 return web.json_response(
-                    {"success": False, "error": "id is required"}, status=400
+                    {"success": False, "error": "id or download_id is required"},
+                    status=400,
                 )
 
             service = await DownloadQueueService.get_instance()
-            deleted = await service.delete_history_item(item_id)
+            deleted = await service.delete_history_item(
+                id=item_id, download_id=download_id
+            )
             return web.json_response({"success": deleted})
         except Exception as exc:
             self._logger.error(
@@ -1737,14 +1862,20 @@ class ModelDownloadHandler:
 
     async def retry_download_from_history(self, request: web.Request) -> web.Response:
         try:
-            item_id = int(request.query.get("id", "0"))
-            if not item_id:
+            download_id = request.query.get("download_id")
+            id_str = request.query.get("id")
+            item_id = int(id_str) if id_str else None
+
+            if not download_id and not item_id:
                 return web.json_response(
-                    {"success": False, "error": "id is required"}, status=400
+                    {"success": False, "error": "id or download_id is required"},
+                    status=400,
                 )
 
             service = await DownloadQueueService.get_instance()
-            item = await service.retry_from_history(item_id)
+            item = await service.retry_from_history(
+                item_id=item_id, download_id=download_id
+            )
             if item is None:
                 return web.json_response(
                     {"success": False, "error": "History item not found or not retryable"},
@@ -1820,6 +1951,39 @@ class ModelDownloadHandler:
             )
             return web.json_response({"success": False, "error": str(exc)}, status=500)
 
+    async def update_download_queue_status(self, request: web.Request) -> web.Response:
+        """Update the status of a queue item (non-terminal transitions).
+
+        Supported transitions include ``queued → downloading``,
+        ``downloading → paused``, ``paused → downloading``, etc.
+        Terminal transitions (``completed``, ``failed``, ``canceled``)
+        should use ``complete_download_in_queue`` instead.
+        """
+        try:
+            download_id = request.query.get("download_id")
+            status = request.query.get("status")
+            if not download_id or not status:
+                return web.json_response(
+                    {
+                        "success": False,
+                        "error": "download_id and status are required",
+                    },
+                    status=400,
+                )
+            service = await DownloadQueueService.get_instance()
+            updated = await service.update_status(download_id, status)
+            if not updated:
+                return web.json_response(
+                    {"success": False, "error": "Download not found in queue"},
+                    status=404,
+                )
+            return web.json_response({"success": True})
+        except Exception as exc:
+            self._logger.error(
+                "Error updating download queue status: %s", exc, exc_info=True
+            )
+            return web.json_response({"success": False, "error": str(exc)}, status=500)
+
 
 class ModelCivitaiHandler:
     """CivitAI integration endpoints."""
@@ -1861,7 +2025,9 @@ class ModelCivitaiHandler:
             return web.json_response(result)
         except Exception as exc:
             self._logger.error(
-                "Error in fetch_all_civitai for %ss: %s", self._service.model_type, exc
+                "Error in fetch_all_civitai for %ss: %s",
+                self._service.model_type, exc,
+                exc_info=True,
             )
             return web.Response(text=str(exc), status=500)
 
@@ -2833,6 +2999,7 @@ class ModelHandlerSet:
             "bulk_delete_models": self.management.bulk_delete_models,
             "verify_duplicates": self.management.verify_duplicates,
             "get_top_tags": self.query.get_top_tags,
+            "search_tags": self.query.search_tags,
             "get_base_models": self.query.get_base_models,
             "get_model_types": self.query.get_model_types,
             "scan_models": self.query.scan_models,
@@ -2862,6 +3029,7 @@ class ModelHandlerSet:
             "retry_all_failed_downloads": self.download.retry_all_failed_downloads,
             "complete_download_in_queue": self.download.complete_download_in_queue,
             "get_download_stats": self.download.get_download_stats,
+            "update_download_queue_status": self.download.update_download_queue_status,
             "get_civitai_versions": self.civitai.get_civitai_versions,
             "get_civitai_model_by_version": self.civitai.get_civitai_model_by_version,
             "get_civitai_model_by_hash": self.civitai.get_civitai_model_by_hash,

py/routes/handlers/preview_handlers.py

@@ -2,6 +2,7 @@
 
 from __future__ import annotations
 
+import asyncio
 import logging
 import mimetypes
 import urllib.parse
@@ -53,6 +54,7 @@ class PreviewHandler:
 
         if not resolved.is_file():
             logger.debug("Preview file not found at %s", str(resolved))
+            asyncio.create_task(self._cleanup_stale_preview_url(normalized))
             raise web.HTTPNotFound(text="Preview file not found")
 
         # aiohttp's FileResponse handles range requests, content headers, and
@@ -69,6 +71,35 @@ class PreviewHandler:
         resp.headers["Cache-Control"] = "public, max-age=86400"
         return resp
 
+    async def _cleanup_stale_preview_url(self, normalized_preview_path: str) -> None:
+        """Fire-and-forget: clear stale preview_url from all model caches.
+
+        When a preview file is no longer on disk, remove its reference from
+        every cached entry so subsequent list API responses return an empty
+        ``preview_url``, letting the frontend show the no-preview placeholder.
+        """
+        try:
+            from ...services.service_registry import ServiceRegistry
+
+            for service_name in ("lora_scanner", "checkpoint_scanner", "embedding_scanner"):
+                scanner = ServiceRegistry.get_service_sync(service_name)
+                if scanner is None or not hasattr(scanner, "_cache"):
+                    continue
+                cache = getattr(scanner, "_cache", None)
+                if cache is None or not hasattr(cache, "clear_preview_by_path"):
+                    continue
+                cleared = await cache.clear_preview_by_path(normalized_preview_path)
+                if cleared and hasattr(scanner, "_persist_current_cache"):
+                    await scanner._persist_current_cache()
+                    logger.info(
+                        "Cleared stale preview_url for %d %s entries (%s)",
+                        cleared,
+                        service_name,
+                        normalized_preview_path,
+                    )
+        except Exception as exc:
+            logger.debug("Failed to clean up stale preview_url: %s", exc)
+
     async def _stream_file(
         self, request: web.Request, path: Path
     ) -> web.StreamResponse:

py/routes/handlers/recipe_handlers.py

@@ -32,6 +32,7 @@ from ...utils.civitai_utils import (
     extract_civitai_image_id_from_cdn_url,
     rewrite_preview_url,
 )
+from ...utils.constants import NSFW_LEVELS
 from ...utils.exif_utils import ExifUtils
 from ...recipes.merger import GenParamsMerger
 from ...recipes.enrichment import RecipeEnricher
@@ -71,6 +72,7 @@ class RecipeHandlerSet:
             "save_recipe": self.management.save_recipe,
             "delete_recipe": self.management.delete_recipe,
             "get_top_tags": self.query.get_top_tags,
+            "search_tags": self.query.search_tags,
             "get_base_models": self.query.get_base_models,
             "get_roots": self.query.get_roots,
             "get_folders": self.query.get_folders,
@@ -316,12 +318,11 @@ class RecipeQueryHandler:
                 raise RuntimeError("Recipe scanner unavailable")
 
             limit = int(request.query.get("limit", "20"))
-            cache = await recipe_scanner.get_cached_data()
-
-            tag_counts: Dict[str, int] = {}
-            for recipe in getattr(cache, "raw_data", []):
-                for tag in recipe.get("tags", []) or []:
-                    tag_counts[tag] = tag_counts.get(tag, 0) + 1
+            if limit < 0:
+                limit = 20
+            elif limit > 200:
+                limit = 20
+            tag_counts = await self._get_recipe_tag_counts(recipe_scanner)
 
             sorted_tags = [
                 {"tag": tag, "count": count} for tag, count in tag_counts.items()
@@ -332,6 +333,55 @@ class RecipeQueryHandler:
             self._logger.error("Error retrieving top tags: %s", exc, exc_info=True)
             return web.json_response({"success": False, "error": str(exc)}, status=500)
 
+    async def search_tags(self, request: web.Request) -> web.Response:
+        try:
+            await self._ensure_dependencies_ready()
+            recipe_scanner = self._recipe_scanner_getter()
+            if recipe_scanner is None:
+                raise RuntimeError("Recipe scanner unavailable")
+
+            query = request.query.get("q", "")
+            limit = int(request.query.get("limit", "20"))
+            if limit < 0:
+                limit = 20
+            elif limit > 200:
+                limit = 20
+
+            tag_counts = await self._get_recipe_tag_counts(recipe_scanner)
+            normalized_query = (query or "").strip().lower()
+            if not normalized_query:
+                sorted_tags = [
+                    {"tag": tag, "count": count} for tag, count in tag_counts.items()
+                ]
+                sorted_tags.sort(key=lambda entry: entry["count"], reverse=True)
+                return web.json_response(
+                    {"success": True, "tags": sorted_tags[: (limit if limit > 0 else 20)]}
+                )
+
+            matched = [
+                {"tag": tag, "count": count}
+                for tag, count in tag_counts.items()
+                if normalized_query in tag.lower()
+            ]
+            matched.sort(key=lambda entry: entry["count"], reverse=True)
+            if limit == 0:
+                result = matched
+            else:
+                result = matched[:limit]
+            return web.json_response({"success": True, "tags": result})
+        except Exception as exc:
+            self._logger.error("Error searching recipe tags: %s", exc, exc_info=True)
+            return web.json_response({"success": False, "error": str(exc)}, status=500)
+
+    async def _get_recipe_tag_counts(self, recipe_scanner) -> Dict[str, int]:
+        """Compute tag->count mapping from cached recipe data."""
+        cache = await recipe_scanner.get_cached_data()
+        tag_counts: Dict[str, int] = {}
+        for recipe in getattr(cache, "raw_data", []):
+            for tag in recipe.get("tags", []) or []:
+                tag_counts[tag] = tag_counts.get(tag, 0) + 1
+        return tag_counts
+
     async def get_base_models(self, request: web.Request) -> web.Response:
         try:
             await self._ensure_dependencies_ready()
@@ -1120,6 +1170,13 @@ class RecipeManagementHandler:
             if parsed_embedded.get("base_model") and not metadata.get("base_model"):
                 metadata["base_model"] = parsed_embedded["base_model"]
 
+        # Extract preview_nsfw_level from the CivitAI API response
+        # (injected into civitai_meta_raw by _download_remote_media).
+        if isinstance(civitai_meta_raw, dict):
+            bl = civitai_meta_raw.get("browsingLevel")
+            if isinstance(bl, int) and bl > 0:
+                metadata["preview_nsfw_level"] = bl
+
         civitai_client = self._civitai_client_getter()
         await RecipeEnricher.enrich_recipe(
             recipe=metadata,
@@ -1515,8 +1572,31 @@ class RecipeManagementHandler:
                     # CivitAI API returns modelVersionIds at the root level of
                     # the image response, NOT inside the meta object.
                     mvids = image_info.get("modelVersionIds")
-                    if mvids and isinstance(civitai_meta_raw, dict):
-                        civitai_meta_raw["modelVersionIds"] = mvids
+                    if mvids:
+                        if isinstance(civitai_meta_raw, dict):
+                            civitai_meta_raw["modelVersionIds"] = mvids
+                        else:
+                            # meta is null but modelVersionIds exists — create a
+                            # minimal dict so downstream parsers can discover
+                            # LoRAs and checkpoints from the API response.
+                            civitai_meta_raw = {"modelVersionIds": mvids}
+
+                    # Inject browsingLevel (canonical integer) so the recipe's
+                    # preview_nsfw_level can be set, enabling proper NSFW blur
+                    # of the preview image.  Fall back to nsfwLevel (string)
+                    # when browsingLevel is absent.
+                    if isinstance(civitai_meta_raw, dict):
+                        browsing_level = image_info.get("browsingLevel")
+                        nsfw_level_str = image_info.get("nsfwLevel")
+                        if isinstance(browsing_level, int) and browsing_level > 0:
+                            civitai_meta_raw["browsingLevel"] = browsing_level
+                        elif (
+                            isinstance(nsfw_level_str, str)
+                            and nsfw_level_str in NSFW_LEVELS
+                        ):
+                            civitai_meta_raw["browsingLevel"] = NSFW_LEVELS[
+                                nsfw_level_str
+                            ]
 
                 original_url = (
                     image_info.get("url") if civitai_image_id and image_info else None
@@ -1796,6 +1876,13 @@ class RecipeManagementHandler:
             "source_path": image_url,
         }
 
+        # Extract preview_nsfw_level from the CivitAI API response
+        # (injected into civitai_meta_raw by _download_remote_media).
+        if isinstance(civitai_meta_raw, dict):
+            bl = civitai_meta_raw.get("browsingLevel")
+            if isinstance(bl, int) and bl > 0:
+                metadata["preview_nsfw_level"] = bl
+
         if civitai_parsed:
             civitai_loras = civitai_parsed.get("loras", [])
             if civitai_loras and not metadata.get("loras"):
@@ -2180,6 +2267,31 @@ class RecipeManagementHandler:
                         "Failed to download image for recipe: %s", exc
                     )
 
+            # Fallback: try to locate a custom image on disk using model_hash + image id
+            if image_bytes is None:
+                image_id = image_data.get("id") or ""
+                if image_id and model_hash:
+                    from ...utils.example_images_paths import get_model_folder
+                    model_folder = get_model_folder(model_hash)
+                    if model_folder and os.path.exists(model_folder):
+                        for fname in os.listdir(model_folder):
+                            if f"custom_{image_id}" in fname:
+                                ext = os.path.splitext(fname)[1].lower()
+                                if ext not in (".jpg", ".jpeg", ".png", ".webp", ".gif"):
+                                    continue
+                                fpath = os.path.join(model_folder, fname)
+                                if os.path.isfile(fpath):
+                                    try:
+                                        with open(fpath, "rb") as f:
+                                            image_bytes = f.read()
+                                        extension = ext
+                                    except Exception as exc:
+                                        self._logger.warning(
+                                            "Failed to read custom image file %s: %s",
+                                            fpath, exc,
+                                        )
+                                break
+
             prompt = (
                 (parsed.get("gen_params") or {}).get("prompt") or ""
             )

py/routes/misc_route_registrar.py

@@ -22,6 +22,8 @@ class RouteDefinition:
 MISC_ROUTE_DEFINITIONS: tuple[RouteDefinition, ...] = (
     RouteDefinition("GET", "/api/lm/settings", "get_settings"),
     RouteDefinition("POST", "/api/lm/settings", "update_settings"),
+    RouteDefinition("GET", "/api/lm/llm/models", "get_llm_models"),
+    RouteDefinition("GET", "/api/lm/llm/provider-models", "get_provider_models"),
     RouteDefinition("GET", "/api/lm/doctor/diagnostics", "get_doctor_diagnostics"),
     RouteDefinition("POST", "/api/lm/doctor/repair-cache", "repair_doctor_cache"),
     RouteDefinition("POST", "/api/lm/doctor/resolve-filename-conflicts", "resolve_doctor_filename_conflicts"),
@@ -37,10 +39,12 @@ MISC_ROUTE_DEFINITIONS: tuple[RouteDefinition, ...] = (
     RouteDefinition("POST", "/api/lm/update-usage-stats", "update_usage_stats"),
     RouteDefinition("GET", "/api/lm/get-usage-stats", "get_usage_stats"),
     RouteDefinition("POST", "/api/lm/update-lora-code", "update_lora_code"),
+    RouteDefinition("GET", "/api/lm/update-lora-code", "get_update_lora_code"),
     RouteDefinition("GET", "/api/lm/trained-words", "get_trained_words"),
     RouteDefinition("GET", "/api/lm/model-example-files", "get_model_example_files"),
     RouteDefinition("POST", "/api/lm/register-nodes", "register_nodes"),
     RouteDefinition("POST", "/api/lm/update-node-widget", "update_node_widget"),
+    RouteDefinition("GET", "/api/lm/update-node-widget", "get_update_node_widget"),
     RouteDefinition("GET", "/api/lm/get-registry", "get_registry"),
     RouteDefinition("GET", "/api/lm/check-model-exists", "check_model_exists"),
     RouteDefinition("GET", "/api/lm/check-models-exist", "check_models_exist"),
@@ -94,6 +98,26 @@ MISC_ROUTE_DEFINITIONS: tuple[RouteDefinition, ...] = (
     RouteDefinition(
         "GET", "/api/lm/delete-model-version", "delete_model_version"
     ),
+    # Hugging Face model endpoints
+    RouteDefinition(
+        "GET", "/api/lm/hf-repo-files", "get_hf_repo_files"
+    ),
+    RouteDefinition(
+        "POST", "/api/lm/download-hf-model", "download_hf_model"
+    ),
+    RouteDefinition(
+        "POST", "/api/lm/set-hf-url", "set_hf_url"
+    ),
+    # Agent skill endpoints
+    RouteDefinition(
+        "GET", "/api/lm/agent/skills", "get_agent_skills"
+    ),
+    RouteDefinition(
+        "POST", "/api/lm/agent/execute/{skill_name}", "execute_agent_skill"
+    ),
+    RouteDefinition(
+        "POST", "/api/lm/agent/cancel", "cancel_agent_skill"
+    ),
 )
 
 

py/routes/misc_routes.py

@@ -39,6 +39,8 @@ from .handlers.misc_handlers import (
     build_service_registry_adapter,
 )
 from .handlers.base_model_handlers import BaseModelHandlerSet
+from .handlers.hf_handlers import HfHandler
+from .handlers.agent_handlers import AgentHandler
 from .misc_route_registrar import MiscRouteRegistrar
 
 logger = logging.getLogger(__name__)
@@ -136,6 +138,8 @@ class MiscRoutes:
         doctor = DoctorHandler(settings_service=self._settings)
         example_workflows = ExampleWorkflowsHandler()
         base_model = BaseModelHandlerSet()
+        hf_handler = HfHandler()
+        agent_handler = AgentHandler()
 
         return self._handler_set_factory(
             health=health,
@@ -155,6 +159,8 @@ class MiscRoutes:
             doctor=doctor,
             example_workflows=example_workflows,
             base_model=base_model,
+            hf_handler=hf_handler,
+            agent_handler=agent_handler,
         )
 
 

py/routes/model_route_registrar.py

@@ -46,6 +46,7 @@ COMMON_ROUTE_DEFINITIONS: tuple[RouteDefinition, ...] = (
         "GET", "/api/lm/{prefix}/auto-organize-progress", "get_auto_organize_progress"
     ),
     RouteDefinition("GET", "/api/lm/{prefix}/top-tags", "get_top_tags"),
+    RouteDefinition("GET", "/api/lm/{prefix}/search-tags", "search_tags"),
     RouteDefinition("GET", "/api/lm/{prefix}/base-models", "get_base_models"),
     RouteDefinition("GET", "/api/lm/{prefix}/model-types", "get_model_types"),
     RouteDefinition("GET", "/api/lm/{prefix}/scan", "scan_models"),
@@ -138,6 +139,9 @@ COMMON_ROUTE_DEFINITIONS: tuple[RouteDefinition, ...] = (
     RouteDefinition(
         "GET", "/api/lm/downloads/queue/complete", "complete_download_in_queue"
     ),
+    RouteDefinition(
+        "GET", "/api/lm/downloads/queue/status", "update_download_queue_status"
+    ),
     RouteDefinition("POST", "/api/lm/{prefix}/cancel-task", "cancel_task"),
     RouteDefinition("GET", "/{prefix}", "handle_models_page"),
 )

py/routes/recipe_route_registrar.py

@@ -29,6 +29,7 @@ ROUTE_DEFINITIONS: tuple[RouteDefinition, ...] = (
     RouteDefinition("POST", "/api/lm/recipes/save", "save_recipe"),
     RouteDefinition("DELETE", "/api/lm/recipe/{recipe_id}", "delete_recipe"),
     RouteDefinition("GET", "/api/lm/recipes/top-tags", "get_top_tags"),
+    RouteDefinition("GET", "/api/lm/recipes/search-tags", "search_tags"),
     RouteDefinition("GET", "/api/lm/recipes/base-models", "get_base_models"),
     RouteDefinition("GET", "/api/lm/recipes/roots", "get_roots"),
     RouteDefinition("GET", "/api/lm/recipes/folders", "get_folders"),

py/routes/stats_routes.py

@@ -11,6 +11,8 @@ from ..config import config
 from ..services.settings_manager import get_settings_manager
 from ..services.server_i18n import server_i18n
 from ..services.service_registry import ServiceRegistry
+from ..services.model_query import normalize_sub_type, resolve_sub_type
+from ..utils.constants import VALID_LORA_SUB_TYPES, VALID_CHECKPOINT_SUB_TYPES
 from ..utils.usage_stats import UsageStats
 
 logger = logging.getLogger(__name__)
@@ -140,6 +142,21 @@ class StatsRoutes:
             # Get usage statistics
             usage_data = await self.usage_stats.get_stats()
             
+            # CivitAI model type distribution across all model types
+            # Use the same logic as the filter panel: normalize_sub_type(resolve_sub_type(entry))
+            # with sub-type validation per model type
+            model_types_counter: Counter[str] = Counter()
+            for entry in lora_cache.raw_data:
+                ntype = normalize_sub_type(resolve_sub_type(entry))
+                if ntype and ntype in VALID_LORA_SUB_TYPES:
+                    model_types_counter[ntype] += 1
+            for entry in checkpoint_cache.raw_data:
+                ntype = normalize_sub_type(resolve_sub_type(entry))
+                if ntype and ntype in VALID_CHECKPOINT_SUB_TYPES:
+                    model_types_counter[ntype] += 1
+            # Embeddings: always count as "embedding" regardless of CivitAI sub-type
+            model_types_counter['embedding'] = len(embedding_cache.raw_data)
+            
             return web.json_response({
                 'success': True,
                 'data': {
@@ -154,7 +171,8 @@ class StatsRoutes:
                     'total_generations': usage_data.get('total_executions', 0),
                     'unused_loras': self._count_unused_models(lora_cache.raw_data, usage_data.get('loras', {})),
                     'unused_checkpoints': self._count_unused_models(checkpoint_cache.raw_data, usage_data.get('checkpoints', {})),
-                    'unused_embeddings': self._count_unused_models(embedding_cache.raw_data, usage_data.get('embeddings', {}))
+                    'unused_embeddings': self._count_unused_models(embedding_cache.raw_data, usage_data.get('embeddings', {})),
+                    'model_types_distribution': dict(model_types_counter.most_common())
                 }
             })
             
@@ -459,9 +477,12 @@ class StatsRoutes:
                 if unused_lora_percent > 50:
                     insights.append({
                         'type': 'warning',
-                        'title': 'High Number of Unused LoRAs',
-                        'description': f'{unused_lora_percent:.1f}% of your LoRAs ({unused_loras}/{total_loras}) have never been used.',
-                        'suggestion': 'Consider organizing or archiving unused models to free up storage space.'
+                        'key': 'insights.unusedLoras.high',
+                        'params': {
+                            'percent': f'{unused_lora_percent:.1f}',
+                            'count': str(unused_loras),
+                            'total': str(total_loras)
+                        }
                     })
             
             if total_checkpoints > 0:
@@ -469,9 +490,12 @@ class StatsRoutes:
                 if unused_checkpoint_percent > 30:
                     insights.append({
                         'type': 'warning',
-                        'title': 'Unused Checkpoints Detected',
-                        'description': f'{unused_checkpoint_percent:.1f}% of your checkpoints ({unused_checkpoints}/{total_checkpoints}) have never been used.',
-                        'suggestion': 'Review and consider removing checkpoints you no longer need.'
+                        'key': 'insights.unusedCheckpoints.detected',
+                        'params': {
+                            'percent': f'{unused_checkpoint_percent:.1f}',
+                            'count': str(unused_checkpoints),
+                            'total': str(total_checkpoints)
+                        }
                     })
             
             if total_embeddings > 0:
@@ -479,9 +503,12 @@ class StatsRoutes:
                 if unused_embedding_percent > 50:
                     insights.append({
                         'type': 'warning',
-                        'title': 'High Number of Unused Embeddings',
-                        'description': f'{unused_embedding_percent:.1f}% of your embeddings ({unused_embeddings}/{total_embeddings}) have never been used.',
-                        'suggestion': 'Consider organizing or archiving unused embeddings to optimize your collection.'
+                        'key': 'insights.unusedEmbeddings.high',
+                        'params': {
+                            'percent': f'{unused_embedding_percent:.1f}',
+                            'count': str(unused_embeddings),
+                            'total': str(total_embeddings)
+                        }
                     })
             
             # Storage insights
@@ -492,18 +519,20 @@ class StatsRoutes:
             if total_size > 100 * 1024 * 1024 * 1024:  # 100GB
                 insights.append({
                     'type': 'info',
-                    'title': 'Large Collection Detected',
-                    'description': f'Your model collection is using {self._format_size(total_size)} of storage.',
-                    'suggestion': 'Consider using external storage or cloud solutions for better organization.'
+                    'key': 'insights.collection.large',
+                    'params': {
+                        'size': self._format_size(total_size)
+                    }
                 })
             
             # Recent activity insight
             if usage_data.get('total_executions', 0) > 100:
                 insights.append({
                     'type': 'success',
-                    'title': 'Active User',
-                    'description': f'You\'ve completed {usage_data["total_executions"]} generations so far!',
-                    'suggestion': 'Keep exploring and creating amazing content with your models.'
+                    'key': 'insights.activity.active',
+                    'params': {
+                        'count': str(usage_data['total_executions'])
+                    }
                 })
             
             return web.json_response({

py/routes/update_routes.py

@@ -16,6 +16,105 @@ logger = logging.getLogger(__name__)
 
 NETWORK_EXCEPTIONS = (ClientError, OSError, asyncio.TimeoutError)
 
+# User-managed directories that live inside the plugin folder (portable
+# mode) and must survive a Git-based update. ``git clean -fd`` would
+# otherwise delete them because they are untracked and, in released tags,
+# not listed in ``.gitignore``. ``-e`` excludes a path from cleaning
+# regardless of whether it is ignored.
+_PRESERVE_DIRS = ('settings.json', 'civitai', 'wildcards', 'backups', 'stats', 'logs', 'cache', 'model_cache')
+
+
+def _clean_excludes() -> List[str]:
+    """Build the ``-e`` arguments for ``git clean`` from :data:`_PRESERVE_DIRS`."""
+    excludes: List[str] = []
+    for name in _PRESERVE_DIRS:
+        excludes.append('-e')
+        excludes.append(name)
+        # For directories, also exclude nested matches explicitly
+        # (``-e dir`` alone matches the dir entry; ``-e dir/**`` guards
+        # contents under all git versions as defense-in-depth).
+        excludes.append('-e')
+        excludes.append(f'{name}/**')
+    return excludes
+
+
+def _stage_preserved_items(plugin_root: str) -> tuple[str, list[str]]:
+    """Move preserved user-data items to a temp directory outside *plugin_root*.
+
+    This ensures that ``git reset --hard``, ``git clean -fd``, and ZIP-based
+    replacement cannot touch these files even when ``-e`` exclusion patterns
+    are mishandled (e.g. on Windows where forward-slash patterns may not
+    match backslash-prefixed paths in some Git builds, or where file locks
+    prevent deletion/recreation).
+
+    Returns:
+        ``(backup_root, staged_names)``: the temp directory path and the
+        list of item names that were successfully moved.
+    """
+    backup_root = tempfile.mkdtemp(prefix='lora_manager_update_')
+    staged: list[str] = []
+    for name in _PRESERVE_DIRS:
+        src = os.path.join(plugin_root, name)
+        if not os.path.lexists(src):
+            continue
+        dst = os.path.join(backup_root, name)
+        try:
+            shutil.move(src, dst)
+            staged.append(name)
+            logger.debug("Staged '%s' for update safety", name)
+        except OSError:
+            # ``shutil.move`` may fail on Windows if a file handle inside
+            # the directory is still open (e.g. a SQLite WAL file). Fall
+            # back to copy-then-remove.
+            logger.debug("Move failed for '%s', falling back to copy", name)
+            try:
+                if os.path.isdir(src) and not os.path.islink(src):
+                    shutil.copytree(src, dst, symlinks=True)
+                    shutil.rmtree(src, ignore_errors=True)
+                else:
+                    shutil.copy2(src, dst)
+                    os.remove(src)
+                staged.append(name)
+                logger.info("Copied (then removed) '%s' for update safety", name)
+            except Exception as exc:
+                logger.warning(
+                    "Could not stage '%s': %s (will rely on git -e / skip lists)", name, exc
+                )
+    return backup_root, staged
+
+
+def _restore_preserved_items(plugin_root: str, backup_root: str, staged: list[str]) -> None:
+    """Move staged items back from *backup_root* into *plugin_root*.
+
+    Any leftover placeholder at the destination (created by git checkout or
+    ZIP extraction) is removed before the move.
+    """
+    for name in staged:
+        src = os.path.join(backup_root, name)
+        dst = os.path.join(plugin_root, name)
+        try:
+            if os.path.lexists(dst):
+                if os.path.isdir(dst) and not os.path.islink(dst):
+                    shutil.rmtree(dst, ignore_errors=True)
+                else:
+                    os.remove(dst)
+            shutil.move(src, dst)
+            logger.debug("Restored '%s' after update", name)
+        except OSError:
+            logger.debug("Move failed restoring '%s', falling back to copy", name)
+            try:
+                if os.path.isdir(src) and not os.path.islink(src):
+                    shutil.copytree(src, dst, symlinks=True, dirs_exist_ok=True)
+                    shutil.rmtree(src, ignore_errors=True)
+                else:
+                    shutil.copy2(src, dst)
+                    os.remove(src)
+                logger.info("Copied '%s' back after update", name)
+            except Exception as exc:
+                logger.error("Failed to restore '%s': %s", name, exc)
+    shutil.rmtree(backup_root, ignore_errors=True)
+
+
 
 class UpdateRoutes:
     """Routes for handling plugin update checks"""
@@ -26,6 +125,7 @@ class UpdateRoutes:
         app.router.add_get('/api/lm/check-updates', UpdateRoutes.check_updates)
         app.router.add_get('/api/lm/version-info', UpdateRoutes.get_version_info)
         app.router.add_post('/api/lm/perform-update', UpdateRoutes.perform_update)
+        app.router.add_post('/api/lm/switch-channel', UpdateRoutes.switch_channel)
     
     @staticmethod
     async def check_updates(request):
@@ -44,10 +144,17 @@ class UpdateRoutes:
 
             # Fetch remote version from GitHub
             if nightly:
-                remote_version, changelog = await UpdateRoutes._get_nightly_version()
-                releases = None
+                local_hash = git_info.get('short_hash', '')
+                nightly_version, releases_result = await asyncio.gather(
+                    UpdateRoutes._get_nightly_version(local_hash),
+                    UpdateRoutes._get_remote_version()
+                )
+                remote_version, _, behind_by, commit_date = nightly_version
+                _, changelog, releases = releases_result
             else:
                 remote_version, changelog, releases = await UpdateRoutes._get_remote_version()
+                behind_by = 0
+                commit_date = ''
             
             # Compare versions
             if nightly:
@@ -60,6 +167,10 @@ class UpdateRoutes:
                     remote_version.replace('v', '')
                 )
             
+            current_dir = os.path.dirname(os.path.abspath(__file__))
+            plugin_root = os.path.dirname(os.path.dirname(current_dir))
+            has_git = os.path.exists(os.path.join(plugin_root, '.git'))
+
             response_data = {
                 'success': True,
                 'current_version': local_version,
@@ -67,13 +178,13 @@ class UpdateRoutes:
                 'update_available': update_available,
                 'changelog': changelog,
                 'git_info': git_info,
-                'nightly': nightly
+                'nightly': nightly,
+                'has_git': has_git,
+                'releases': releases,
+                'behind_by': behind_by,
+                'commit_date': commit_date
             }
             
-            # Include releases list for stable mode
-            if releases is not None:
-                response_data['releases'] = releases
-            
             return web.json_response(response_data)
             
         except NETWORK_EXCEPTIONS as e:
@@ -105,9 +216,14 @@ class UpdateRoutes:
             # Format: version-short_hash
             version_string = f"{local_version}-{short_hash}"
             
+            current_dir = os.path.dirname(os.path.abspath(__file__))
+            plugin_root = os.path.dirname(os.path.dirname(current_dir))
+            has_git = os.path.exists(os.path.join(plugin_root, '.git'))
+
             return web.json_response({
                 'success': True,
-                'version': version_string
+                'version': version_string,
+                'has_git': has_git
             })
             
         except Exception as e:
@@ -135,20 +251,22 @@ class UpdateRoutes:
             if os.path.exists(settings_path):
                 with open(settings_path, 'r', encoding='utf-8') as f:
                     settings_backup = f.read()
-                logger.info("Backed up settings.json")
+                logger.debug("Backed up settings.json (%d bytes)", len(settings_backup))
 
-            git_folder = os.path.join(plugin_root, '.git')
-            if os.path.exists(git_folder):
-                # Git update
-                success, new_version = await UpdateRoutes._perform_git_update(plugin_root, nightly)
-            else:
-                # Fallback: Download ZIP and replace files
-                success, new_version = await UpdateRoutes._download_and_replace_zip(plugin_root)
+            staged_backup_dir, staged_items = _stage_preserved_items(plugin_root)
+            try:
+                git_folder = os.path.join(plugin_root, '.git')
+                if os.path.exists(git_folder):
+                    success, new_version = await UpdateRoutes._perform_git_update(plugin_root, nightly)
+                else:
+                    success, new_version = await UpdateRoutes._download_and_replace_zip(plugin_root)
+            finally:
+                _restore_preserved_items(plugin_root, staged_backup_dir, staged_items)
 
             if settings_backup and success:
                 with open(settings_path, 'w', encoding='utf-8') as f:
                     f.write(settings_backup)
-                logger.info("Restored settings.json")
+                logger.debug("Restored settings.json content (%d bytes)", len(settings_backup))
 
             if success:
                 return web.json_response({
@@ -169,6 +287,164 @@ class UpdateRoutes:
                 'error': str(e)
             })
 
+    @staticmethod
+    async def switch_channel(request):
+        """
+        Switch between release and nightly update channels.
+
+        ZIP/CNR install → Nightly: git init + checkout main (one-way upgrade)
+        Git install → Release:   git checkout latest tag (.git preserved)
+        ZIP/CNR install → Release: ZIP download (no .git, stays in ZIP mode)
+        Git install → Nightly:   git checkout main + pull
+        """
+        try:
+            body = await request.json() if request.has_body else {}
+            channel = body.get('channel', '')
+
+            if channel not in ('release', 'nightly'):
+                return web.json_response({
+                    'success': False,
+                    'error': f'Invalid channel: {channel}. Must be "release" or "nightly".'
+                })
+
+            current_dir = os.path.dirname(os.path.abspath(__file__))
+            plugin_root = os.path.dirname(os.path.dirname(current_dir))
+
+            settings_path = ensure_settings_file(logger)
+            settings_backup = None
+            if os.path.exists(settings_path):
+                with open(settings_path, 'r', encoding='utf-8') as f:
+                    settings_backup = f.read()
+                logger.debug("Backed up settings.json before channel switch (%d bytes)", len(settings_backup))
+
+            staged_backup_dir, staged_items = _stage_preserved_items(plugin_root)
+            try:
+                git_folder = os.path.join(plugin_root, '.git')
+
+                if channel == 'nightly':
+                    git_backup = None
+                    if os.path.exists(git_folder):
+                        git_backup = UpdateRoutes._backup_git(git_folder, 'nightly')
+
+                    success = False
+                    new_version = ''
+                    try:
+                        if os.path.exists(git_folder):
+                            success, new_version = await UpdateRoutes._perform_git_update(
+                                plugin_root, nightly=True
+                            )
+                        else:
+                            success, new_version = UpdateRoutes._init_git_repo(plugin_root)
+                    finally:
+                        UpdateRoutes._restore_git(git_backup, git_folder, success, 'nightly')
+                else:
+                    success = False
+                    new_version = ''
+                    if os.path.exists(git_folder):
+                        success, new_version = await UpdateRoutes._perform_git_update(
+                            plugin_root, nightly=False
+                        )
+                    else:
+                        tracking_file = os.path.join(plugin_root, '.tracking')
+                        if os.path.exists(tracking_file):
+                            os.remove(tracking_file)
+                        success, new_version = await UpdateRoutes._download_and_replace_zip(plugin_root)
+            finally:
+                _restore_preserved_items(plugin_root, staged_backup_dir, staged_items)
+
+            if settings_backup and success:
+                with open(settings_path, 'w', encoding='utf-8') as f:
+                    f.write(settings_backup)
+                logger.debug("Restored settings.json content after channel switch (%d bytes)", len(settings_backup))
+
+            if success:
+                return web.json_response({
+                    'success': True,
+                    'channel': channel,
+                    'new_version': new_version,
+                    'message': f'Switched to {channel} channel'
+                })
+            else:
+                return web.json_response({
+                    'success': False,
+                    'error': f'Failed to switch to {channel} channel'
+                })
+
+        except Exception as e:
+            logger.error("Failed to switch channel: %s", e, exc_info=True)
+            return web.json_response({
+                'success': False,
+                'error': str(e)
+            })
+
+    @staticmethod
+    def _init_git_repo(plugin_root: str) -> tuple[bool, str]:
+        """
+        Initialize a Git repository in a ZIP-installed plugin folder.
+        Clones the remote history and checks out main branch.
+        """
+        try:
+            import git
+        except ImportError:
+            logger.error(
+                "GitPython is not available: cannot initialize git repo. "
+                "Install git or set $GIT_PYTHON_GIT_EXECUTABLE to the git binary path."
+            )
+            return False, ""
+
+        clean_excludes = _clean_excludes()
+
+        try:
+            repo = git.Repo.init(plugin_root)
+            origin = repo.create_remote(
+                'origin',
+                'https://github.com/willmiao/ComfyUI-Lora-Manager.git'
+            )
+            origin.fetch()
+
+            repo.create_head('main', origin.refs.main)
+            repo.git.checkout('main', '--force')
+            repo.git.reset('--hard')
+            repo.git.clean('-fd', *clean_excludes)
+
+            tracking_file = os.path.join(plugin_root, '.tracking')
+            if os.path.exists(tracking_file):
+                os.remove(tracking_file)
+                logger.info("Removed .tracking file (now in git mode)")
+
+            new_version = f"main-{repo.head.commit.hexsha[:7]}"
+            logger.info("Initialized git repo on main branch: %s", new_version)
+            return True, new_version
+
+        except Exception as e:
+            logger.error("Failed to initialize git repo: %s", e, exc_info=True)
+            return False, ""
+
+    @staticmethod
+    def _backup_git(git_folder, label):
+        try:
+            backup_dir = tempfile.mkdtemp()
+            backup = os.path.join(backup_dir, '.git')
+            shutil.copytree(git_folder, backup)
+            logger.info("Backed up .git before switching to %s", label)
+            return backup
+        except Exception as e:
+            logger.error("Failed to backup .git before %s switch: %s", label, e)
+            return None
+
+    @staticmethod
+    def _restore_git(git_backup, git_folder, success, label):
+        if git_backup and not success:
+            try:
+                if os.path.exists(git_folder):
+                    shutil.rmtree(git_folder)
+                shutil.copytree(git_backup, git_folder)
+                logger.info("Restored .git after failed %s switch", label)
+            except Exception as e:
+                logger.error("Failed to restore .git after %s switch: %s", label, e)
+        if git_backup:
+            shutil.rmtree(os.path.dirname(git_backup), ignore_errors=True)
+
     @staticmethod
     async def _download_and_replace_zip(plugin_root: str) -> tuple[bool, str]:
         """
@@ -223,8 +499,7 @@ class UpdateRoutes:
             except Exception:
                 logger.debug("Could not close downloaded-version history database", exc_info=True)
 
-            # Skip settings.json, civitai, model cache and runtime cache folders
-            UpdateRoutes._clean_plugin_folder(plugin_root, skip_files=['settings.json', 'civitai', 'model_cache', 'cache', 'wildcards', 'backups', 'stats'])
+            UpdateRoutes._clean_plugin_folder(plugin_root, skip_files=list(_PRESERVE_DIRS))
 
             # Extract ZIP to temp dir
             with tempfile.TemporaryDirectory() as tmp_dir:
@@ -234,7 +509,7 @@ class UpdateRoutes:
                     extracted_root = next(os.scandir(tmp_dir)).path
 
                     # Copy files, skipping user data that should be preserved
-                    skip_items = {'settings.json', 'civitai', 'wildcards', 'backups', 'stats'}
+                    skip_items = set(_PRESERVE_DIRS)
                     for item in os.listdir(extracted_root):
                         if item in skip_items:
                             continue
@@ -251,7 +526,7 @@ class UpdateRoutes:
                     # for ComfyUI Manager to work properly
                     tracking_info_file = os.path.join(plugin_root, '.tracking')
                     tracking_files = []
-                    skip_tracked = {'civitai', 'wildcards', 'backups', 'stats'}
+                    skip_tracked = set(_PRESERVE_DIRS) - {'settings.json'}
                     for root, dirs, files in os.walk(extracted_root):
                         # Skip user data directories and their contents
                         rel_root = os.path.relpath(root, extracted_root)
@@ -274,7 +549,8 @@ class UpdateRoutes:
         except Exception as e:
             logger.error(f"ZIP update failed: {e}", exc_info=True)
             return False, ""
-        
+
+    @staticmethod
     def _clean_plugin_folder(plugin_root, skip_files=None):
         skip_files = skip_files or []
         for item in os.listdir(plugin_root):
@@ -287,41 +563,54 @@ class UpdateRoutes:
                 os.remove(path)
     
     @staticmethod
-    async def _get_nightly_version() -> tuple[str, List[str]]:
-        """
-        Fetch latest commit from main branch
-        """
+    async def _get_nightly_version(local_hash: str = "") -> tuple[str, List[str], int, str]:
         repo_owner = "willmiao"
         repo_name = "ComfyUI-Lora-Manager"
-        
-        # Use GitHub API to fetch the latest commit from main branch
+
         github_url = f"https://api.github.com/repos/{repo_owner}/{repo_name}/commits/main"
-        
+
         try:
             downloader = await get_downloader()
-            success, data = await downloader.make_request('GET', github_url, custom_headers={'Accept': 'application/vnd.github+json'})
-            
+            success, data = await downloader.make_request(
+                'GET', github_url,
+                custom_headers={'Accept': 'application/vnd.github+json'}
+            )
+
             if not success:
-                logger.warning(f"Failed to fetch GitHub commit: {data}")
-                return "main", []
-            
-            commit_sha = data.get('sha', '')[:7]  # Short hash
+                logger.warning("Failed to fetch GitHub commit: %s", data)
+                return "main", [], 0, ""
+
+            commit_sha = data.get('sha', '')[:7]
             commit_message = data.get('commit', {}).get('message', '')
-            
-            # Format as "main-{short_hash}"
+            commit_date = data.get('commit', {}).get('committer', {}).get('date', '')[:10]
+
             version = f"main-{commit_sha}"
-            
-            # Use commit message as changelog
             changelog = [commit_message] if commit_message else []
-            
-            return version, changelog
-        
+
+            behind_by = 0
+            if local_hash and local_hash not in ('unknown', 'stable'):
+                compare_url = (
+                    f"https://api.github.com/repos/{repo_owner}/{repo_name}"
+                    f"/compare/{local_hash}...main"
+                )
+                c_ok, c_data = await downloader.make_request(
+                    'GET', compare_url,
+                    custom_headers={'Accept': 'application/vnd.github+json'}
+                )
+                if c_ok:
+                    if c_data.get('status') in ('ahead', 'diverged'):
+                        behind_by = c_data.get('ahead_by', 0)
+                    else:
+                        behind_by = c_data.get('behind_by', 0)
+
+            return version, changelog, behind_by, commit_date
+
         except NETWORK_EXCEPTIONS as e:
             logger.warning("Unable to reach GitHub for nightly version: %s", e)
-            return "main", []
+            return "main", [], 0, ""
         except Exception as e:
-            logger.error(f"Error fetching nightly version: {e}", exc_info=True)
-            return "main", []
+            logger.error("Error fetching nightly version: %s", e, exc_info=True)
+            return "main", [], 0, ""
     
     @staticmethod
     def _compare_nightly_versions(local_git_info: Dict[str, str], remote_version: str) -> bool:
@@ -365,6 +654,8 @@ class UpdateRoutes:
             )
             return False, ""
 
+        clean_excludes = _clean_excludes()
+
         try:
             # Open the Git repository
             repo = git.Repo(plugin_root)
@@ -376,8 +667,9 @@ class UpdateRoutes:
             if nightly:
                 # Reset to discard any local changes
                 repo.git.reset('--hard')
-                # Clean untracked files
-                repo.git.clean('-fd')
+                # Clean untracked files, but preserve user-managed directories
+                # (wildcards, backups, stats, civitai, caches, settings.json).
+                repo.git.clean('-fd', *clean_excludes)
                 
                 # Switch to main branch and pull latest
                 main_branch = 'main'
@@ -394,8 +686,9 @@ class UpdateRoutes:
             else:
                 # Reset to discard any local changes
                 repo.git.reset('--hard')
-                # Clean untracked files
-                repo.git.clean('-fd')
+                # Clean untracked files, but preserve user-managed directories
+                # (wildcards, backups, stats, civitai, caches, settings.json).
+                repo.git.clean('-fd', *clean_excludes)
                 
                 # Get latest release tag
                 tags = sorted(repo.tags, key=lambda t: t.commit.committed_datetime, reverse=True)

py/services/agent/__init__.py

@@ -0,0 +1,27 @@
+"""LLM-powered metadata enrichment pipeline infrastructure.
+
+This package provides the orchestration layer for LLM-powered features.
+Skills define *what* to do (prompt template).  The :class:`AgentService`
+handles *how* (LLM calls, context gathering, validation, progress).
+
+NOTE: The current implementation is a code-driven pipeline, not a true
+agent loop.  Future agent orchestration (LLM-driven tool selection) will
+live alongside this package with its own namespace.
+"""
+
+from __future__ import annotations
+
+from .skill_definition import SkillDefinition, SkillPermissions
+from .skill_registry import SkillRegistry
+from .agent_service import AgentService, AgentProgressReporter, SkillResult
+from .post_processor import PostProcessor
+
+__all__ = [
+    "AgentProgressReporter",
+    "AgentService",
+    "PostProcessor",
+    "SkillDefinition",
+    "SkillPermissions",
+    "SkillRegistry",
+    "SkillResult",
+]

py/services/agent/agent_service.py

@@ -0,0 +1,489 @@
+"""Pipeline orchestration service.
+
+The :class:`AgentService` coordinates LLM-powered pipeline execution:
+
+1. Look up the pipeline definition in :class:`SkillRegistry`
+2. Validate input against its ``input_schema``
+3. Prepare context via :mod:`~py.metadata_ops` (read metadata, list base models, fetch HF README)
+4. If ``llm_required``: call :class:`LLMService` with the rendered prompt
+5. Post-process via :class:`PostProcessor` (delegates I/O to :mod:`~py.metadata_ops`)
+6. Broadcast progress and completion via :class:`WebSocketManager`
+
+Pipeline definitions (*skills*) describe *what* to do (prompt template).
+The AgentService handles *how* (LLM calls, context gathering, validation,
+progress).
+"""
+
+from __future__ import annotations
+
+import asyncio
+import json
+import logging
+import re
+from dataclasses import dataclass, field
+from typing import Any, Dict, List, Optional
+
+import aiohttp
+
+import os
+
+from ...config import config
+from ..llm_service import LLMService
+from ..websocket_manager import ws_manager
+from .post_processor import PostProcessor
+from .skill_registry import SkillRegistry
+from .skills.enrich_hf_metadata.readme_processor import (
+    clean_readme_for_llm,
+    extract_relevant_section,
+)
+
+logger = logging.getLogger(__name__)
+
+
+class AgentProgressReporter:
+    """Protocol-compatible progress reporter backed by WebSocket broadcast."""
+
+    async def on_progress(self, payload: Dict[str, Any]) -> None:
+        await ws_manager.broadcast(payload)
+
+
+@dataclass
+class SkillResult:
+    """Outcome of a skill execution."""
+
+    success: bool
+    updated_models: List[Dict[str, Any]] = field(default_factory=list)
+    errors: List[str] = field(default_factory=list)
+    summary: str = ""
+
+
+def _validate_schema(data: Any, schema: Dict[str, Any], path: str = "") -> List[str]:
+    """Minimal JSON schema validator.
+
+    Supports a subset of JSON Schema: ``type``, ``properties``, ``required``,
+    ``items``, ``enum``.  Returns a list of error messages (empty = valid).
+    """
+
+    errors: List[str] = []
+    if not schema:
+        return errors
+
+    expected_type = schema.get("type")
+    if expected_type:
+        type_map = {
+            "string": str,
+            "number": (int, float),
+            "integer": int,
+            "boolean": bool,
+            "array": list,
+            "object": dict,
+            "null": type(None),
+        }
+        expected_py = type_map.get(expected_type)
+        if expected_py is not None and not isinstance(data, expected_py):
+            errors.append(f"{path or 'root'}: expected {expected_type}, got {type(data).__name__}")
+            return errors
+
+    if expected_type == "object" and isinstance(data, dict):
+        properties = schema.get("properties", {})
+        required = schema.get("required", [])
+        for req_key in required:
+            if req_key not in data:
+                errors.append(f"{path or 'root'}: missing required property '{req_key}'")
+        for key, value in data.items():
+            if key in properties:
+                errors.extend(_validate_schema(value, properties[key], f"{path}.{key}"))
+
+    if expected_type == "array" and isinstance(data, list):
+        items_schema = schema.get("items")
+        if items_schema:
+            for i, item in enumerate(data):
+                errors.extend(_validate_schema(item, items_schema, f"{path}[{i}]"))
+
+    if "enum" in schema and data not in schema["enum"]:
+        errors.append(f"{path or 'root'}: value '{data}' not in enum {schema['enum']}")
+
+    return errors
+
+
+# ------------------------------------------------------------------
+# Prompt template rendering
+# ------------------------------------------------------------------
+
+
+def _render_prompt(template: str, variables: Dict[str, Any]) -> str:
+    """Render a prompt template with ``{{variable}}`` placeholders.
+
+    Uses simple regex substitution — no Jinja2 dependency needed.
+    """
+
+    def replace(match: re.Match) -> str:
+        key = match.group(1).strip()
+        value = variables.get(key, "")
+        if isinstance(value, (dict, list)):
+            return json.dumps(value, ensure_ascii=False, indent=2)
+        return str(value)
+
+    return re.sub(r"\{\{(\w+)\}\}", replace, template)
+
+
+class AgentService:
+    """Orchestrate agent skill execution.
+
+    Usage::
+
+        service = await AgentService.get_instance()
+        result = await service.execute_skill(
+            skill_name="enrich_hf_metadata",
+            input_data={"model_paths": ["/path/to/model.safetensors"]},
+            progress_callback=AgentProgressReporter(),
+        )
+    """
+
+    _instance: Optional["AgentService"] = None
+    _lock: asyncio.Lock = asyncio.Lock()
+
+    def __init__(
+        self,
+        *,
+        skill_registry: Optional[SkillRegistry] = None,
+        llm_service: Optional[LLMService] = None,
+    ) -> None:
+        self._registry = skill_registry
+        self._llm_service = llm_service
+
+    @classmethod
+    async def get_instance(cls) -> "AgentService":
+        """Return the lazily-initialised global ``AgentService``."""
+
+        if cls._instance is None:
+            async with cls._lock:
+                if cls._instance is None:
+                    cls._instance = cls(
+                        skill_registry=await SkillRegistry.get_instance(),
+                        llm_service=await LLMService.get_instance(),
+                    )
+        return cls._instance
+
+    @classmethod
+    def reset_instance(cls) -> None:
+        """Reset the cached singleton — primarily for tests."""
+
+        cls._instance = None
+
+    async def _ensure_registry(self) -> SkillRegistry:
+        if self._registry is None:
+            self._registry = await SkillRegistry.get_instance()
+        return self._registry
+
+    async def _ensure_llm(self) -> LLMService:
+        if self._llm_service is None:
+            self._llm_service = await LLMService.get_instance()
+        return self._llm_service
+
+    async def list_skills(self) -> List[Dict[str, Any]]:
+        """Return a JSON-serialisable list of available skills."""
+
+        registry = await self._ensure_registry()
+        return [
+            {
+                "name": s.name,
+                "title": s.title,
+                "description": s.description,
+                "llm_required": s.llm_required,
+                "model_type_filter": s.model_type_filter,
+            }
+            for s in registry.list_skills()
+        ]
+
+    async def execute_skill(
+        self,
+        *,
+        skill_name: str,
+        input_data: Dict[str, Any],
+        progress_callback: Optional[AgentProgressReporter] = None,
+    ) -> SkillResult:
+        """Execute a pipeline (skill) on the given models.
+
+        Args:
+            skill_name: Name of the pipeline to execute
+            input_data: Input validated against the pipeline's ``input_schema``
+            progress_callback: Optional WebSocket progress reporter
+
+        Returns:
+            :class:`SkillResult` with success status and updated model info
+        """
+
+        registry = await self._ensure_registry()
+        skill = registry.get_skill(skill_name)
+        if skill is None:
+            return SkillResult(
+                success=False,
+                errors=[f"Skill not found: {skill_name}"],
+                summary=f"Skill '{skill_name}' does not exist",
+            )
+
+        input_errors = _validate_schema(input_data, skill.input_schema)
+        if input_errors:
+            return SkillResult(
+                success=False,
+                errors=input_errors,
+                summary=f"Invalid input: {'; '.join(input_errors)}",
+            )
+
+        model_paths = input_data.get("model_paths", [])
+        if not model_paths:
+            return SkillResult(
+                success=False,
+                errors=["No model_paths provided"],
+                summary="No models to process",
+            )
+
+        total = len(model_paths)
+        processed = 0
+        success_count = 0
+        skipped_count = 0
+        updated_models: List[Dict[str, Any]] = []
+        errors: List[str] = []
+        post_processor = PostProcessor()
+
+        await self._emit_progress(
+            progress_callback, skill_name, status="started",
+            total=total, processed=0, success=0,
+        )
+
+        llm = await self._ensure_llm()
+        llm_configured = llm.is_configured() if skill.llm_required else True
+
+        for model_path in model_paths:
+            model_filename = os.path.basename(model_path)
+            logger.info(
+                "[%s] [%d/%d] %s",
+                skill_name, processed + 1, total, model_filename,
+            )
+            updated_data: Dict[str, Any] = {}
+            skip_model = False
+            try:
+                from ...metadata_ops import read_metadata
+                metadata = await read_metadata(model_path)
+
+                # Fast-fail: enrich_hf_metadata requires hf_url to have HF README context
+                if skill_name == "enrich_hf_metadata" and not metadata.get("hf_url", ""):
+                    logger.info(
+                        "[%s] SKIP %s — no hf_url in metadata",
+                        skill_name, model_filename,
+                    )
+                    skipped_count += 1
+                    skip_model = True
+
+                if not skip_model:
+                    prompt_vars: Dict[str, Any] = {"model_path": model_path}
+                    if skill.llm_required and llm_configured:
+                        prompt_vars = await self._build_prompt_context(
+                            skill_name, model_path, metadata, registry, llm,
+                        )
+
+                    llm_response: Optional[Dict[str, Any]] = None
+                    if skill.llm_required and llm_configured:
+                        prompt_template = registry.load_prompt(skill_name)
+                        rendered = _render_prompt(prompt_template, prompt_vars)
+                        llm_response = await llm.chat_completion_json(
+                            system_prompt=prompt_vars.get(
+                                "system_prompt",
+                                "You are a helpful assistant that extracts structured metadata.",
+                            ),
+                            user_prompt=rendered,
+                        )
+                        if llm_response:
+                            logger.info(
+                                "[%s] [%d/%d] %s → base_model=%s confidence=%s",
+                                skill_name, processed + 1, total, model_filename,
+                                (llm_response.get("base_model") or "?")[:50],
+                                llm_response.get("confidence", "?"),
+                            )
+
+                    model_result = await post_processor.process(
+                        skill_name=skill_name,
+                        model_path=model_path,
+                        llm_output=llm_response or {},
+                        metadata=metadata,
+                        readme_content=prompt_vars.get("readme_content_full", ""),
+                    )
+
+                    if model_result.get("success", True):
+                        success_count += 1
+                        uf = model_result.get("updated_fields", [])
+                        if uf:
+                            updated_models.append({"path": model_path, "updated_fields": uf})
+                        updated_data = model_result.get("updates", {})
+                        if "preview_url" in updated_data and updated_data["preview_url"]:
+                            updated_data["preview_url"] = config.get_preview_static_url(
+                                updated_data["preview_url"]
+                            )
+                    else:
+                        errors.extend(
+                            model_result.get("errors", [model_result.get("error", "Unknown error")])
+                        )
+
+            except Exception as exc:
+                logger.error("Skill %s failed for %s: %s", skill_name, model_path, exc)
+                errors.append(f"{model_path}: {exc}")
+
+            processed += 1
+            await self._emit_progress(
+                progress_callback, skill_name, status="processing",
+                total=total, processed=processed, success=success_count,
+                skipped=skipped_count,
+                current_path=model_path,
+                updated_data=updated_data,
+            )
+
+        result = SkillResult(
+            success=success_count > 0,
+            updated_models=updated_models,
+            errors=errors,
+            summary=f"Processed {processed}/{total} models, {success_count} succeeded, {skipped_count} skipped",
+        )
+
+        await self._emit_progress(
+            progress_callback, skill_name, status="completed",
+            total=total, processed=processed, success=success_count,
+            skipped=skipped_count,
+            updated_models=updated_models, errors=errors, summary=result.summary,
+        )
+
+        return result
+
+    # ------------------------------------------------------------------
+    # Base model grouping (keeps the prompt compact)
+    # ------------------------------------------------------------------
+
+    @staticmethod
+    def _format_base_models(models: List[str]) -> str:
+        """Format the base model list as a flat, one-per-line list.
+
+        Attempts to group by family consistently degraded LLM extraction
+        accuracy — the LLM finds individual model names harder to spot
+        in comma-separated groups than in a simple ``- Name`` list.
+        """
+        return "\n".join(f"- {m}" for m in models)
+
+    async def _build_prompt_context(
+        self,
+        skill_name: str,
+        model_path: str,
+        metadata: Dict[str, Any],
+        registry: SkillRegistry,
+        llm: Any,
+    ) -> Dict[str, Any]:
+        """Gather variables for the skill's prompt template.
+
+        Reads metadata, fetches the HF README (if applicable), lists available
+        base models, loads user priority tags, and returns a dict that maps to
+        ``{{variable}}`` placeholders in ``prompt.md``.
+        """
+        from ...metadata_ops import identify_model_type, list_base_models
+        from ..settings_manager import SettingsManager
+
+        context: Dict[str, Any] = {
+            "model_path": model_path,
+            "model_basename": "",
+            "hf_url": "",
+            "repo": "",
+            "readme_content": "",
+            "readme_content_full": "",
+            "current_metadata": {},
+            "base_models": [],
+            "priority_tags": "",
+        }
+
+        # Extract model basename (filename without extension) for the LLM
+        # to use when locating the matching section in collection repos.
+        raw_basename = os.path.splitext(os.path.basename(model_path))[0]
+        context["model_basename"] = raw_basename or ""
+
+        context["current_metadata"] = {
+            "file_name": metadata.get("file_name", ""),
+            "base_model": metadata.get("base_model", ""),
+            "tags": metadata.get("tags", []),
+            "modelDescription": metadata.get("modelDescription", ""),
+            "trainedWords": metadata.get("trainedWords", []),
+            "sha256": (metadata.get("sha256") or "")[:16] + "..." if metadata.get("sha256") else "",
+            "size": metadata.get("size", 0),
+        }
+
+        hf_url = metadata.get("hf_url", "")
+        context["hf_url"] = hf_url
+        repo = self._extract_repo_from_url(hf_url) if hf_url else ""
+        context["repo"] = repo or ""
+        if repo:
+            readme = await self._fetch_readme(repo)
+            # Trim README to the section relevant to this model file
+            # (collection repos often have multiple models in one README).
+            if readme and raw_basename:
+                trimmed = extract_relevant_section(readme, raw_basename)
+                cleaned = clean_readme_for_llm(trimmed) if trimmed else ""
+            else:
+                cleaned = clean_readme_for_llm(readme) if readme else ""
+            context["readme_content"] = cleaned if cleaned else "(README not available)"
+            context["readme_content_full"] = readme or ""
+
+        try:
+            raw_models = await list_base_models()
+            context["base_models"] = self._format_base_models(raw_models)
+        except Exception as exc:
+            logger.debug("Failed to list base models: %s", exc)
+            context["base_models"] = "</not available>"
+
+        # Determine model type and load the corresponding priority_tags
+        try:
+            model_type = await identify_model_type(model_path)
+            context["model_type"] = model_type
+            settings = SettingsManager()
+            priority_config = settings.get_priority_tag_config()
+            context["priority_tags"] = priority_config.get(model_type, "")
+        except Exception as exc:
+            logger.debug("Failed to load priority tags: %s", exc)
+            context["model_type"] = "lora"
+            context["priority_tags"] = ""
+
+        return context
+
+    @staticmethod
+    def _extract_repo_from_url(hf_url: str) -> Optional[str]:
+        """Extract ``user/repo`` from a HuggingFace URL."""
+        if not hf_url:
+            return None
+        m = re.match(r"https?://huggingface\.co/([^/]+/[^/]+)", hf_url)
+        return m.group(1) if m else None
+
+    @staticmethod
+    async def _fetch_readme(repo: str) -> str:
+        """Fetch README.md from HuggingFace (tries ``main``, then ``master``)."""
+        async with aiohttp.ClientSession(
+            headers={"User-Agent": "ComfyUI-LoRA-Manager/1.0"},
+            timeout=aiohttp.ClientTimeout(total=30),
+        ) as session:
+            for branch in ("main", "master"):
+                url = f"https://huggingface.co/{repo}/raw/{branch}/README.md"
+                try:
+                    async with session.get(url) as resp:
+                        if resp.status == 200:
+                            return await resp.text()
+                except Exception as exc:
+                    logger.debug("Failed to fetch README from %s: %s", url, exc)
+        return ""
+
+    async def _emit_progress(
+        self,
+        callback: Optional[AgentProgressReporter],
+        skill_name: str,
+        *,
+        status: str,
+        **extra: Any,
+    ) -> None:
+        """Send a progress update via WebSocket (if callback is set)."""
+        payload: Dict[str, Any] = {"type": "agent_progress", "skill": skill_name, "status": status}
+        payload.update(extra)
+        if callback is not None:
+            await callback.on_progress(payload)

py/services/agent/post_processor.py

@@ -0,0 +1,336 @@
+"""Post-processing engine for skill pipeline outputs.
+
+The :class:`PostProcessor` takes the LLM's structured JSON output and applies
+it to a model's on-disk metadata via the :mod:`~py.metadata_ops` functions.
+
+It handles all the skill-specific business logic — conditions, transformations,
+and orchestration of multiple side-effects (write metadata, download preview,
+refresh cache).  All actual I/O is delegated to :mod:`~py.metadata_ops`.
+"""
+
+from __future__ import annotations
+
+import json
+import logging
+import os
+import re
+from datetime import datetime, timezone
+from typing import Any, Dict, List, Optional
+
+logger = logging.getLogger(__name__)
+
+
+class PostProcessor:
+    """Deterministic post-processor for skill pipeline outputs.
+
+    Usage (called by :class:`~py.services.agent.agent_service.AgentService`)::
+
+        processor = PostProcessor()
+        result = await processor.process(
+            skill_name="enrich_hf_metadata",
+            model_path="/path/to/model.safetensors",
+            llm_output={...},
+            metadata={...},    # from metadata_ops.read_metadata()
+        )
+    """
+
+    async def process(
+        self,
+        *,
+        skill_name: str,
+        model_path: str,
+        llm_output: Dict[str, Any],
+        metadata: Dict[str, Any],
+        readme_content: str = "",
+    ) -> Dict[str, Any]:
+        """Route *llm_output* to the correct skill post-processor.
+
+        *readme_content* is optional raw markdown content (e.g. HF README)
+        that is converted to HTML and stored as ``modelDescription`` for
+        the description tab.
+
+        Returns a dict with keys ``success`` (bool), ``updated_fields`` (list),
+        ``preview_downloaded`` (bool), and ``errors`` (list).
+        """
+        if skill_name == "enrich_hf_metadata":
+            return await self._process_enrich_hf_metadata(
+                model_path, llm_output, metadata, readme_content,
+            )
+        return {
+            "success": False,
+            "updated_fields": [],
+            "errors": [f"No post-processor registered for skill: {skill_name}"],
+        }
+
+    # ------------------------------------------------------------------
+    # enrich_hf_metadata
+    # ------------------------------------------------------------------
+
+    async def _process_enrich_hf_metadata(
+        self,
+        model_path: str,
+        llm_output: Dict[str, Any],
+        metadata: Dict[str, Any],
+        readme_content: str = "",
+    ) -> Dict[str, Any]:
+        from ...metadata_ops import (
+            apply_metadata_updates,
+            download_preview,
+            refresh_cache,
+        )
+        from .skills.enrich_hf_metadata.readme_processor import (
+            convert_readme_to_html,
+            extract_gallery_images,
+            extract_gallery_table_images,
+            extract_relevant_section,
+            extract_simple_markdown_images,
+            extract_html_img_tags,
+            extract_repo_from_hf_url,
+        )
+
+        updated_fields: List[str] = []
+        preview_downloaded = False
+
+        # -- Determine whether this is an HF-sourced model -----------------
+        is_hf_model = not metadata.get("from_civitai", True)
+
+        # -- Collect updates -----------------------------------------------
+        updates: Dict[str, Any] = {}
+
+        # base_model
+        new_base = (llm_output.get("base_model") or "").strip()
+        current_base = metadata.get("base_model", "") or ""
+        if new_base and self._should_overwrite(current_base, is_hf_model):
+            updates["base_model"] = new_base
+
+        # trigger words → civitai.trainedWords
+        new_triggers = llm_output.get("trigger_words", [])
+        trigger_words_empty = True
+        if isinstance(new_triggers, list):
+            cleaned = [t.strip() for t in new_triggers if t.strip()]
+            cleaned = [t for t in cleaned if t.lower() not in ("none", "null", "n/a")]
+            trigger_words_empty = not cleaned
+            current_civitai = metadata.get("civitai") or {}
+            current_triggers = current_civitai.get("trainedWords") or []
+            if self._should_overwrite_list(current_triggers, is_hf_model):
+                trig_civitai = dict(current_civitai)
+                if "civitai" in updates and isinstance(updates["civitai"], dict):
+                    trig_civitai.update(updates["civitai"])
+                trig_civitai["trainedWords"] = cleaned
+                updates["civitai"] = trig_civitai
+
+        # modelDescription — from raw README content (converted to HTML)
+        if readme_content and is_hf_model:
+            converted = convert_readme_to_html(readme_content)
+            if converted:
+                updates["modelDescription"] = converted
+
+        # short_description → civitai.description (for "About this version")
+        short_desc = (llm_output.get("short_description") or "").strip()
+        if short_desc and is_hf_model:
+            current_civitai = metadata.get("civitai") or {}
+            desc_civitai = dict(current_civitai)
+            if "civitai" in updates and isinstance(updates["civitai"], dict):
+                desc_civitai.update(updates["civitai"])
+            desc_civitai["description"] = short_desc
+            updates["civitai"] = desc_civitai
+
+        # gallery images → civitai.images (from YAML frontmatter widget entries
+        # and Sample Gallery markdown tables in the README body)
+        gallery_images: List[Dict[str, Any]] = []
+        if readme_content and is_hf_model:
+            hf_url = metadata.get("hf_url", "") or ""
+            repo = extract_repo_from_hf_url(hf_url)
+            if repo:
+                rec_w = llm_output.get("recommended_width") or 0
+                rec_h = llm_output.get("recommended_height") or 0
+
+                # 1. Widget images (YAML frontmatter)
+                gallery = extract_gallery_images(
+                    readme_content, repo,
+                    default_width=rec_w, default_height=rec_h,
+                )
+
+                # 2. Sample Gallery table images (markdown body), deduplicated
+                existing_urls = {img["url"] for img in gallery if img.get("url")}
+                table_images = extract_gallery_table_images(
+                    readme_content, repo,
+                    existing_urls=existing_urls,
+                    default_width=rec_w, default_height=rec_h,
+                )
+                existing_urls.update(img["url"] for img in table_images if img.get("url"))
+
+                # 3. Simple markdown images `![alt](url)` in the body
+                simple_images = extract_simple_markdown_images(
+                    readme_content, repo,
+                    existing_urls=existing_urls,
+                    default_width=rec_w, default_height=rec_h,
+                )
+                existing_urls.update(img["url"] for img in simple_images if img.get("url"))
+
+                # 4. HTML `<img>` tags (used by many collection repos)
+                html_images = extract_html_img_tags(
+                    readme_content, repo,
+                    existing_urls=existing_urls,
+                    default_width=rec_w, default_height=rec_h,
+                )
+
+                all_images = gallery + table_images + simple_images + html_images
+                if all_images:
+                    gallery_images = all_images
+                    current_civitai = metadata.get("civitai") or {}
+                    gallery_civitai = dict(current_civitai)
+                    if "civitai" in updates and isinstance(updates["civitai"], dict):
+                        gallery_civitai.update(updates["civitai"])
+                    gallery_civitai["images"] = all_images
+                    updates["civitai"] = gallery_civitai
+
+        # tags
+        new_tags = llm_output.get("tags", [])
+        if isinstance(new_tags, list) and new_tags:
+            existing_tags = metadata.get("tags") or []
+            merged = self._merge_tags(existing_tags, new_tags)
+            if len(merged) > len(existing_tags) or is_hf_model:
+                updates["tags"] = merged
+
+        # metadata_source & llm_enriched_at (always set)
+        updates["metadata_source"] = "agent:enrich_hf_metadata"
+        updates["llm_enriched_at"] = datetime.now(timezone.utc).isoformat()
+
+        # Store LLM confidence in metadata so it's accessible for evaluation
+        raw_confidence = (llm_output.get("confidence") or "").strip()
+        if raw_confidence:
+            updates["_llm_confidence"] = raw_confidence
+
+        # Fallback: extract instance_prompt from YAML frontmatter when the LLM
+        # returned empty trigger words but the README has instance_prompt.
+        if trigger_words_empty:
+            instance_prompt = _extract_yaml_instance_prompt(readme_content)
+            if instance_prompt:
+                current_civitai = metadata.get("civitai") or {}
+                trig_civitai = dict(current_civitai)
+                if "civitai" in updates and isinstance(updates["civitai"], dict):
+                    trig_civitai.update(updates["civitai"])
+                trig_civitai["trainedWords"] = [instance_prompt]
+                updates["civitai"] = trig_civitai
+
+        preview_remote_url = (llm_output.get("preview_url") or "").strip()
+        # Fallback: if the LLM couldn't find a preview image in the cleaned
+        # README, find the first gallery image from the *model-specific
+        # section* of the README (not the repo-wide first image, which
+        # belongs to a different model in collection repos).
+        if not preview_remote_url and readme_content and is_hf_model:
+            model_basename = os.path.splitext(os.path.basename(model_path))[0]
+            relevant_section = extract_relevant_section(
+                readme_content, model_basename,
+            )
+            if relevant_section and relevant_section != readme_content:
+                for img in gallery_images:
+                    img_url = img.get("url", "")
+                    if img_url and img_url in relevant_section:
+                        preview_remote_url = img_url
+                        break
+        # Last resort: use the first gallery image from the full README.
+        if not preview_remote_url and gallery_images:
+            preview_remote_url = gallery_images[0].get("url", "")
+        current_preview = metadata.get("preview_url") or ""
+        if preview_remote_url and not (current_preview and os.path.exists(current_preview)):
+            local_path = await download_preview(model_path, preview_remote_url)
+            if local_path:
+                preview_downloaded = True
+                updates["preview_url"] = local_path
+
+        # notes — plain-text summary of usage info from the LLM
+        new_notes = (llm_output.get("notes") or "").strip()
+        if new_notes:
+            updates["notes"] = new_notes
+
+        # usage_tips — JSON string (e.g. {"strength_min":0.85,"strength_max":1.4})
+        raw_tips = (llm_output.get("usage_tips") or "").strip()
+        if raw_tips and raw_tips != "{}":
+            try:
+                json.loads(raw_tips)
+                updates["usage_tips"] = raw_tips
+            except (json.JSONDecodeError, TypeError):
+                logger.warning(
+                    "LLM returned invalid usage_tips JSON: %s", raw_tips[:200]
+                )
+
+        if updates:
+            updated_fields = await apply_metadata_updates(model_path, updates)
+
+        # -- Refresh scanner cache ------------------------------------------
+        if updated_fields or preview_downloaded:
+            await refresh_cache(model_path)
+
+        return {
+            "success": True,
+            "updated_fields": updated_fields,
+            "preview_downloaded": preview_downloaded,
+            "updates": updates,
+            "errors": [],
+        }
+
+    # ------------------------------------------------------------------
+    # Helpers
+    # ------------------------------------------------------------------
+
+    @staticmethod
+    def _should_overwrite(current_value: str, is_hf_model: bool) -> bool:
+        """Return ``True`` when a scalar field should be overwritten."""
+        return is_hf_model or not current_value or current_value.lower() in (
+            "", "unknown",
+        )
+
+    @staticmethod
+    def _should_overwrite_list(current_list: List[str], is_hf_model: bool) -> bool:
+        """Return ``True`` when a list field should be overwritten."""
+        return is_hf_model or not current_list
+
+    @staticmethod
+    def _merge_tags(existing: List[str], new: List[str]) -> List[str]:
+        """Merge *new* tags into *existing*, all lowercased.
+
+        This matches the behaviour of :class:`TagUpdateService` which
+        normalises every tag to lowercase for case-insensitive dedup.
+        """
+        merged: List[str] = []
+        seen: set = set()
+        for tag in list(existing) + list(new):
+            t = tag.strip().lower()
+            if t and t not in seen:
+                merged.append(t)
+                seen.add(t)
+        return merged
+
+
+# ------------------------------------------------------------------
+# Module-level helpers
+# ------------------------------------------------------------------
+
+
+def _extract_yaml_instance_prompt(readme_content: str) -> str:
+    """Extract ``instance_prompt`` from the YAML frontmatter of a HF README.
+
+    Returns the prompt text, or empty string if not found.  Handles
+    ``null`` / ``~`` YAML null values by returning empty string.
+    """
+    if not readme_content or not readme_content.startswith("---"):
+        return ""
+
+    # Find end of frontmatter
+    end = readme_content.find("---", 3)
+    if end == -1:
+        return ""
+    frontmatter = readme_content[3:end]
+
+    for line in frontmatter.split("\n"):
+        line = line.strip()
+        m = re.match(r"^instance_prompt:\s*(.*)", line)
+        if m:
+            val = m.group(1).strip().strip('"').strip("'")
+            if val.lower() in ("null", "~", "none", ""):
+                return ""
+            return val
+
+    return ""

py/services/agent/skill_definition.py

@@ -0,0 +1,45 @@
+"""Skill definition data structures.
+
+Each skill is described by a :class:`SkillDefinition` that declares its
+input/output schemas, whether it needs an LLM call, and what permissions
+its post-processor has.
+"""
+
+from __future__ import annotations
+
+from dataclasses import dataclass, field
+from typing import Any, Dict, List, Optional, Tuple
+
+
+@dataclass(frozen=True)
+class SkillPermissions:
+    """Declarative permission scope for a skill's post-processor.
+
+    These are auditable constraints — the :class:`AgentService` checks them
+    before invoking the handler.  They are defense-in-depth, not a sandbox.
+    """
+
+    write_metadata: bool = True
+    write_previews: bool = True
+    network_domains: Tuple[str, ...] = ()
+
+
+@dataclass(frozen=True)
+class SkillDefinition:
+    """Immutable description of an agent skill."""
+
+    name: str
+    title: str
+    description: str
+    llm_required: bool
+    input_schema: Dict[str, Any] = field(default_factory=dict)
+    output_schema: Dict[str, Any] = field(default_factory=dict)
+    model_type_filter: Optional[List[str]] = None
+    permissions: SkillPermissions = field(default_factory=SkillPermissions)
+
+    def applies_to_model_type(self, model_type: str) -> bool:
+        """Return ``True`` if this skill can run on the given model type."""
+
+        if self.model_type_filter is None:
+            return True
+        return model_type in self.model_type_filter

py/services/agent/skill_registry.py

@@ -0,0 +1,210 @@
+"""Discovery and loading of prompt-based skills.
+
+Skills live in ``py/services/agent/skills/<name>/`` directories.  Each
+directory must contain a ``prompt.md`` file with YAML frontmatter::
+
+    ---
+    name: my_skill
+    title: "My Skill"
+    description: "What this skill does"
+    llm_required: true
+    ---
+
+    Prompt template with ``{{variable}}`` placeholders.
+
+Legacy ``SKILL.md`` files are also supported for backward compatibility.
+
+The registry scans the skills directory on first access and caches results.
+"""
+
+from __future__ import annotations
+
+import asyncio
+import logging
+import re
+from pathlib import Path
+from typing import Any, Dict, List, Optional
+
+import yaml
+
+from .skill_definition import SkillDefinition, SkillPermissions
+
+logger = logging.getLogger(__name__)
+
+# Directory where built-in skills are stored
+_SKILLS_DIR = Path(__file__).parent / "skills"
+
+#: Preferred file names for prompt definition files (tried in order).
+#: ``prompt.md`` is the current convention; ``SKILL.md`` is the legacy name
+#: kept for backward compatibility.
+_PROMPT_FILE_NAMES: tuple[str, ...] = ("prompt.md", "SKILL.md")
+
+
+# ---------------------------------------------------------------------------
+# Frontmatter parser
+# ---------------------------------------------------------------------------
+
+_FRONTMATTER_RE = re.compile(
+    r"^---\s*\n(.*?\n)---\s*\n?(.*)", re.DOTALL
+)
+
+
+def _parse_skill_file(path: Path) -> tuple[dict, str]:
+    """Read a prompt definition file (``prompt.md`` or legacy ``SKILL.md``) and
+    return (frontmatter_dict, body_text).
+
+    Raises ``ValueError`` if the file lacks valid YAML frontmatter.
+    """
+    text = path.read_text(encoding="utf-8")
+    m = _FRONTMATTER_RE.match(text)
+    if not m:
+        raise ValueError(f"Missing or invalid YAML frontmatter in {path}")
+    frontmatter = yaml.safe_load(m.group(1))
+    if not isinstance(frontmatter, dict):
+        raise ValueError(f"Frontmatter in {path} is not a mapping")
+    body = m.group(2).strip()
+    return frontmatter, body
+
+
+class SkillRegistry:
+    """Discover and load agent skills from the filesystem."""
+
+    _instance: Optional["SkillRegistry"] = None
+    _lock: asyncio.Lock = asyncio.Lock()
+
+    def __init__(self, skills_dir: Path = _SKILLS_DIR) -> None:
+        self._skills_dir = skills_dir
+        self._skills: Dict[str, SkillDefinition] = {}
+        self._loaded: bool = False
+
+    # ------------------------------------------------------------------
+    # Singleton access
+    # ------------------------------------------------------------------
+
+    @classmethod
+    async def get_instance(cls) -> "SkillRegistry":
+        """Return the lazily-initialised global ``SkillRegistry``."""
+
+        if cls._instance is None:
+            async with cls._lock:
+                if cls._instance is None:
+                    registry = cls()
+                    registry._discover()
+                    cls._instance = registry
+        return cls._instance
+
+    @classmethod
+    def reset_instance(cls) -> None:
+        """Reset the cached singleton — primarily for tests."""
+
+        cls._instance = None
+
+    # ------------------------------------------------------------------
+    # Discovery
+    # ------------------------------------------------------------------
+
+    @staticmethod
+    def _find_prompt_file(skill_dir: Path) -> Path | None:
+        """Return the first prompt definition file that exists in *skill_dir*.
+
+        Tries ``_PROMPT_FILE_NAMES`` in order so that new conventions
+        (``prompt.md``) take precedence while legacy ``SKILL.md`` files
+        still load without changes.
+        """
+        for name in _PROMPT_FILE_NAMES:
+            candidate = skill_dir / name
+            if candidate.exists():
+                return candidate
+        return None
+
+    def _discover(self) -> None:
+        """Scan the skills directory and load all valid skill definitions."""
+
+        self._skills.clear()
+        if not self._skills_dir.is_dir():
+            logger.warning("Skills directory does not exist: %s", self._skills_dir)
+            self._loaded = True
+            return
+
+        for entry in sorted(self._skills_dir.iterdir()):
+            if not entry.is_dir():
+                continue
+            prompt_file = self._find_prompt_file(entry)
+            if prompt_file is None:
+                continue
+            try:
+                definition = self._load_skill_definition(prompt_file)
+                if definition is not None:
+                    self._skills[definition.name] = definition
+                    logger.debug("Loaded skill: %s", definition.name)
+            except Exception as exc:
+                logger.warning("Failed to load skill from %s: %s", prompt_file, exc)
+
+        self._loaded = True
+        logger.info("Discovered %d prompt-based skills", len(self._skills))
+
+    def _load_skill_definition(self, path: Path) -> Optional[SkillDefinition]:
+        """Parse a prompt definition file's frontmatter into a
+        :class:`SkillDefinition`."""
+
+        try:
+            data, _body = _parse_skill_file(path)
+        except (ValueError, yaml.YAMLError) as exc:
+            logger.warning("Failed to parse prompt file %s: %s", path, exc)
+            return None
+
+        if "name" not in data:
+            logger.warning("Prompt file %s missing required 'name' field", path)
+            return None
+
+        perm_data = data.get("permissions", {})
+        permissions = SkillPermissions(
+            write_metadata=perm_data.get("write_metadata", True),
+            write_previews=perm_data.get("write_previews", True),
+            network_domains=tuple(perm_data.get("network_domains", [])),
+        )
+
+        return SkillDefinition(
+            name=data["name"],
+            title=data.get("title", data["name"]),
+            description=data.get("description", ""),
+            llm_required=data.get("llm_required", False),
+            input_schema=data.get("input_schema", {}),
+            output_schema=data.get("output_schema", {}),
+            model_type_filter=data.get("model_type_filter"),
+            permissions=permissions,
+        )
+
+    # ------------------------------------------------------------------
+    # Public API
+    # ------------------------------------------------------------------
+
+    def list_skills(self) -> List[SkillDefinition]:
+        """Return all discovered skill definitions."""
+
+        if not self._loaded:
+            self._discover()
+        return list(self._skills.values())
+
+    def get_skill(self, name: str) -> Optional[SkillDefinition]:
+        """Return the skill definition for ``name``, or ``None`` if not found."""
+
+        if not self._loaded:
+            self._discover()
+        return self._skills.get(name)
+
+    def load_prompt(self, name: str) -> str:
+        """Load and return the prompt template body for the named skill."""
+
+        skill_dir = self._skills_dir / name
+        skill_path = self._find_prompt_file(skill_dir)
+        if skill_path is None:
+            raise FileNotFoundError(
+                f"Prompt file not found for skill '{name}' in {skill_dir} "
+                f"(tried {list(_PROMPT_FILE_NAMES)})"
+            )
+        try:
+            _frontmatter, body = _parse_skill_file(skill_path)
+            return body
+        except (ValueError, yaml.YAMLError) as exc:
+            raise ValueError(f"Failed to parse prompt from {skill_path}: {exc}") from exc

py/services/agent/skills/enrich_hf_metadata/prompt.md

@@ -0,0 +1,165 @@
+---
+name: enrich_hf_metadata
+title: "Enrich Metadata from HuggingFace"
+description: >
+  Parse the HuggingFace model card via LLM to extract description, trigger
+  words, base model, tags, and preview image URL.
+llm_required: true
+---
+
+You are an expert assistant for AI image generation models. Your task is to extract structured metadata from a HuggingFace model card (README.md).
+
+## Model Information
+
+- **Repository**: {{hf_url}}
+- **Model file path**: {{model_path}}
+- **Model filename**: {{model_basename}}
+- **Repository ID**: {{repo}}
+
+## Current Metadata (may be incomplete)
+
+```json
+{{current_metadata}}
+```
+
+## User Priority Tags Reference
+
+The user has configured the following list of **meaningful tag categories** for this model type (`{{model_type}}`):
+
+```
+{{priority_tags}}
+```
+
+These are the subjects, styles, and concepts the user considers useful for categorization. Use this list as a **reference** when evaluating tags (see the **tags** section below).
+
+## Available Base Models
+
+The following base models are currently valid in this system.  Use the EXACT
+name listed — do not invent aliases or modify variant suffixes.
+
+{{base_models}}
+
+## HuggingFace README Content
+
+```
+{{readme_content}}
+```
+
+## Extraction Instructions
+
+Extract the following information from the README content above:
+
+### base_model
+The base model this model was trained on. Use EXACTLY one of the names from the **Available Base Models** list above. Do not invent new names or use aliases.
+
+Check the YAML frontmatter for ``base_model:`` first. If the frontmatter has no ``base_model:``, look at the **model filename** (``{{model_basename}}``), YAML ``tags:``, README title and first paragraph for clues — the base model family is often embedded in the name
+
+### trigger_words
+The trigger words or activation prompts needed to use this LoRA. Look for:
+- `instance_prompt:` in the YAML frontmatter
+- Phrases like "trigger word:", "trigger:", "use this prompt:", "activation prompt:"
+- In collection repos: the trigger section **specific to this model file** (look near matching download links or anchor IDs)
+- Example prompts at the start (usually the first word or phrase before any description)
+Return as an array of strings. If none found, return an empty array `[]`. **Never** return `["None"]` or any placeholder value — a truly empty list means no trigger words exist.
+
+### short_description
+A concise 1-2 sentence summary of what this model does. Extract from the "Model description" section or the first paragraph.  For collection repos, focus on the **specific model version** matching `{{model_basename}}`, not the repo as a whole.  Return empty string if the README is too minimal.
+
+### tags
+3-8 relevant tags for categorizing this model. **Quality over quantity.**
+
+Sources to consider:
+- The YAML frontmatter `tags:` list (filter out technical ones — see below)
+- The subject, style, character, or concept the model represents
+- The model filename itself may give clues (e.g. "pokemon", "anime", "pixelart")
+
+**Critical filtering rules — apply them strictly:**
+
+1. **Exclude technical/generic tags.** Reject any tag that describes the model's **training methodology, framework, architecture, or modality** rather than its content. Examples to exclude: `text-to-image`, `diffusers`, `lora`, `dreambooth`, `diffusers-training`, `flux`, `sdxl`, `checkpoint`, `pytorch`, `safetensors`, `fine-tuning`, `stable-diffusion`, and any variant of these.
+
+2. **Cross-reference against the priority_tags reference.** Only include a tag if it meaningfully describes what the model actually creates (subject, style, character type) and is semantically close to one of the priority_tags. If none of the README's tags match meaningful categories, prefer returning a smaller set or an empty array over including low-value tags.
+
+3. **All lowercase, no spaces, no hyphens** (use single words like `"photorealistic"`, `"anime"`, `"character"`).
+
+Return empty array if no meaningful content tags remain after filtering.
+
+### recommended_width, recommended_height
+The recommended image generation resolution for this model, in pixels. Look for sections like "Best Dimensions", "Recommended size", "Suggested resolution", or similar phrasing in the README. Prefer the explicitly marked "Best" or default resolution. If the table/list has multiple entries (e.g. "768 x 1024 (Best)" and "1024 x 1024 (Default)"), use the one marked "Best". Return integers. If no resolution can be determined, return 0 for both.
+
+### preview_url
+The URL of the most suitable preview image from the README. Look for:
+- Image tags near the section matching the model filename (`{{model_basename}}`)
+- The YAML frontmatter `widget:` section (which often has `output.url` fields)
+- In collection repos: the sample images listed **under the section** for this specific model version
+- Generic `![alt](url)` in the body
+Choose the first image that appears to be a generation example (not a logo or diagram). Construct the absolute URL as `https://huggingface.co/{{repo}}/resolve/main/{filename}`. If no suitable image is found, return an empty string.
+
+### notes
+A plain-text summary of the model card's key practical usage information. Combine trigger words, style modifiers, recommended parameters (steps, CFG, resolution, sampler), and any setup tips into a readable paragraph.  For collection repos, focus on the **specific model version** matching `{{model_basename}}`.  Return empty string if the README has no useful usage info.
+
+### usage_tips
+A JSON string with structured usage recommendations. Extract from the README any explicit ranges or recommended values (e.g. "Set LoRA strength: **0.85 - 1.4**", "CLIP strength: 0.5"). Possible fields (include only those you can determine):
+
+```json
+{
+  "strength_min": 0.85,
+  "strength_max": 1.4,
+  "strength_range": "0.85-1.4",
+  "strength": 0.6,
+  "clip_strength": 0.5,
+  "clip_skip": 2
+}
+```
+
+Return the JSON string (e.g. `'{"strength_min":0.85,"strength_max":1.4}'`). Return `"{}"` if nothing useful is found.
+
+### confidence
+Your confidence level in the extracted data:
+- "high" — most fields were explicitly stated in the README
+- "medium" — some fields were inferred from context
+- "low" — most fields are guesses based on limited information
+
+## Important: Handling Collection Repos (multiple model files)
+
+Many HuggingFace repos contain **multiple model files** in a single repository
+(e.g. a "LoRA collection" with different styles/characters in separate files).
+
+The model file currently being enriched is: **`{{model_basename}}`**
+
+To find the correct section in the README:
+
+1. **Search for download links** containing the filename — the surrounding paragraph is your section.
+2. **Search for anchor IDs** (`<a id="...">`) or section headings whose text matches words from the filename.
+3. **Search for HTML headings** (`<h1>`, `<h2>`, `<span>`) containing parts of the filename.
+4. If no match is found, use the full README as usual — the model may be the only one in the repo.
+
+When a matching section IS found, prefer metadata from that section.
+When no section matches (e.g. single-model repos or repos without per-file sections),
+extract metadata from the full README normally.  Do not return empty data just
+because the filename doesn't appear in the README.
+
+## Output Format
+
+Return ONLY a JSON object with exactly these fields (no markdown fences, no extra text):
+
+```json
+{
+  "model_path": "{{model_path}}",
+  "base_model": "<canonical name or empty string>",
+  "trigger_words": ["<word1>", "<word2>"],
+  "short_description": "<1-2 sentence summary>",
+  "tags": ["<tag1>", "<tag2>"],
+  "recommended_width": 768,
+  "recommended_height": 1024,
+  "preview_url": "<image URL or empty string>",
+  "notes": "<plain-text usage summary or empty string>",
+  "usage_tips": "<JSON string like '{\"strength_min\":0.85,\"strength_max\":1.4}' or '{}'>",
+  "confidence": "<high|medium|low>"
+}
+```
+
+Important:
+- Only include the JSON object, no other text
+- If a field cannot be determined, use an empty string or empty array
+- Do not fabricate information not supported by the README
+- Never use placeholder values like `"None"` or `"unknown"` for missing data — use empty string or empty array

py/services/aria2_downloader.py

@@ -84,6 +84,7 @@ class Aria2Downloader:
         self._transfers: Dict[str, Aria2Transfer] = {}
         self._poll_interval = 0.5
         self._state_store = Aria2TransferStateStore()
+        self._stderr_reader_task: Optional[asyncio.Task] = None
 
     @property
     def is_running(self) -> bool:
@@ -115,7 +116,7 @@ class Aria2Downloader:
 
         try:
             while True:
-                status = await self.get_status(download_id)
+                status = await self._get_status_with_retry(download_id)
                 if status is None:
                     return False, "aria2 download not found"
 
@@ -136,6 +137,35 @@ class Aria2Downloader:
         finally:
             self._transfers.pop(download_id, None)
 
+    async def _get_status_with_retry(
+        self, download_id: str, *, max_retries: int = 4, retry_delay: float = 3.0
+    ) -> Optional[Dict[str, Any]]:
+        """Call get_status with retry for transient RPC failures.
+
+        Only retries on :exc:`Aria2Error` (RPC-level failure).  Returns
+        ``None`` immediately when the download_id is not tracked (a missing
+        transfer is not a transient condition, so retrying is pointless).
+
+        A single failed RPC call should not immediately fail the download,
+        because aria2 may be temporarily busy (e.g. finalizing multiple
+        concurrent downloads) and a retry will often succeed.
+        """
+        last_exc: Optional[Exception] = None
+        for attempt in range(max_retries):
+            try:
+                return await self.get_status(download_id)
+            except Aria2Error as exc:
+                last_exc = exc
+                if attempt < max_retries - 1:
+                    logger.warning(
+                        "aria2 get_status transient failure (attempt %d/%d) for %s: %s",
+                        attempt + 1, max_retries, download_id, exc,
+                    )
+                    await asyncio.sleep(retry_delay)
+        raise Aria2Error(
+            f"Failed to query aria2 download status after {max_retries} attempts: {last_exc}"
+        ) from last_exc
+
     async def _schedule_download(
         self,
         url: str,
@@ -171,6 +201,13 @@ class Aria2Downloader:
             "auto-file-renaming": "false",
             "file-allocation": "none",
         }
+
+        # Pass proxy to aria2 so the actual file transfer goes through the
+        # same proxy used by the aiohttp-based URL resolution step above.
+        downloader = await get_downloader()
+        if downloader.proxy_url:
+            options["all-proxy"] = downloader.proxy_url
+
         if request_headers:
             options["header"] = [
                 f"{key}: {value}" for key, value in request_headers.items()
@@ -312,6 +349,16 @@ class Aria2Downloader:
     async def close(self) -> None:
         """Shut down the RPC process and session."""
 
+        # Cancel the background stderr reader first so it stops reading
+        # from the pipe before the subprocess is terminated.
+        if self._stderr_reader_task is not None:
+            self._stderr_reader_task.cancel()
+            try:
+                await asyncio.wait_for(self._stderr_reader_task, timeout=2.0)
+            except (asyncio.CancelledError, asyncio.TimeoutError):
+                pass
+            self._stderr_reader_task = None
+
         if self._rpc_session is not None:
             await self._rpc_session.close()
             self._rpc_session = None
@@ -331,6 +378,23 @@ class Aria2Downloader:
                 process.kill()
                 await process.wait()
 
+    async def _drain_stderr(self) -> None:
+        """Continuously drain aria2's stderr pipe so it never blocks.
+
+        When the 64 KB pipe buffer fills up, aria2's ``write()`` to stderr
+        blocks, which freezes the entire ``aria2c`` process — including its
+        RPC handler.  This background task reads lines from stderr as they
+        arrive and forwards them to Python's logger.
+        """
+        try:
+            assert self._process is not None and self._process.stderr is not None
+            async for line in self._process.stderr:
+                text = line.decode("utf-8", errors="replace").rstrip()
+                if text:
+                    logger.debug("aria2 stderr: %s", text)
+        except Exception:
+            pass
+
     async def _dispatch_progress(self, callback, snapshot: DownloadProgress) -> None:
         try:
             result = callback(snapshot, snapshot)
@@ -465,6 +529,17 @@ class Aria2Downloader:
 
             await self._wait_until_ready()
 
+            # Drain aria2's stderr in a background task so the pipe buffer
+            # never fills up.  If the pipe blocks, aria2 itself freezes and
+            # cannot respond to RPC — this was the root cause of the
+            # "Failed to query aria2 download status" timeout bug.
+            # Must start AFTER _wait_until_ready to avoid a race where the
+            # drain task consumes aria2's early-exit error message before
+            # _wait_until_ready can read it.
+            self._stderr_reader_task = asyncio.create_task(
+                self._drain_stderr()
+            )
+
     def _resolve_executable(self) -> str:
         settings = get_settings_manager()
         configured_path = (settings.get("aria2c_path") or "").strip()
@@ -584,7 +659,9 @@ class Aria2Downloader:
         if self._rpc_session is None or self._rpc_session.closed:
             async with self._rpc_session_lock:
                 if self._rpc_session is None or self._rpc_session.closed:
-                    timeout = aiohttp.ClientTimeout(total=30)
+                    timeout = aiohttp.ClientTimeout(
+                        total=None, sock_connect=10, sock_read=60
+                    )
                     self._rpc_session = aiohttp.ClientSession(timeout=timeout)
         return self._rpc_session
 

py/services/base_model_service.py

@@ -1,7 +1,7 @@
 from abc import ABC, abstractmethod
 import asyncio
 import re
-from typing import Any, Dict, List, Optional, Type, TYPE_CHECKING
+from typing import Any, Dict, List, Optional, Type, Union, TYPE_CHECKING
 import logging
 import os
 import time
@@ -104,6 +104,109 @@ class BaseModelService(ABC):
         fetch_duration = time.perf_counter() - t0
         initial_count = len(sorted_data)
 
+        # Optionally filter by civitai model ID (shows all local versions of a specific model)
+        civitai_model_id = kwargs.get("civitai_model_id")
+        if civitai_model_id is not None:
+            sorted_data = [
+                item for item in sorted_data
+                if self._extract_group_key(item) == civitai_model_id
+            ]
+            # VLM mode: always sort by version ID descending (newest version first),
+            # regardless of the current sort_by preference.
+            # Fall back to modified timestamp for non-CivitAI sources.
+            sorted_data.sort(
+                key=lambda x: self._extract_version_id(x)
+                or x.get("modified", 0)
+                or 0,
+                reverse=True,
+            )
+
+        # Optionally group by civitai modelId, showing only the latest version per model
+        dedup_lost = 0
+        if kwargs.get("group_by_model") and civitai_model_id is None:
+            # Determine whether to further sub-group by base model
+            # When version_grouping is "same_base", versions with different
+            # base models are effectively different groups — the dedup key
+            # needs to include base_model so the version count and VLM flow
+            # stay consistent (card shows correct count for its base model).
+            ufs = self.settings.get("version_grouping", "same_base")
+            group_by_base = ufs == "same_base"
+
+            dedup_map = {}  # (modelId [,base_model]) -> (item, version_or_modified)
+            version_counter = {}  # same-key -> count
+            standalone = []
+            for item in sorted_data:
+                mid = self._extract_group_key(item)
+                if mid is None:
+                    standalone.append(item)
+                    continue
+                key = (mid, item.get("base_model") or "") if group_by_base else mid
+                # Count all versions per key
+                version_counter[key] = version_counter.get(key, 0) + 1
+                # Prefer CivitAI version_id; fall back to modified timestamp
+                vid = self._extract_version_id(item)
+                if vid is None:
+                    vid = item.get("modified", 0) or 0
+                if key not in dedup_map or vid > dedup_map[key][1]:
+                    dedup_map[key] = (item, vid)
+            # Attach version_count to each surviving grouped item (shallow copy
+            # to avoid mutating cached dicts — the cache is shared across requests)
+            for key, (item, vid) in dedup_map.items():
+                item = dict(item)
+                item["version_count"] = version_counter[key]
+                dedup_map[key] = (item, vid)
+            dedup_lost = len(sorted_data) - (len(dedup_map) + len(standalone))
+            sorted_data = [entry[0] for entry in dedup_map.values()] + standalone
+
+        # Re-sort by version_count (grouped: after dedup; non-grouped: group internally, sort, expand)
+        if sort_params.key == "versions_count" and civitai_model_id is None:
+            reverse = sort_params.order == "desc"
+            if kwargs.get("group_by_model"):
+                # Grouped mode: items are already dedup'd with version_count attached
+                sorted_data.sort(
+                    key=lambda x: (
+                        x.get("version_count", 0),
+                        (x.get("model_name") or x.get("file_name") or "").lower(),
+                        x.get("file_path", "").lower(),
+                    ),
+                    reverse=reverse,
+                )
+            else:
+                # Non-grouped mode: group internally, sort groups by count, expand
+                # Respect the version_grouping setting (same logic as grouped dedup)
+                ufs = self.settings.get("version_grouping", "same_base")
+                group_by_base = ufs == "same_base"
+
+                model_groups: Dict[Any, List[Dict]] = {}
+                ungrouped_standalone: List[Dict] = []
+                for item in sorted_data:
+                    mid = self._extract_group_key(item)
+                    if mid is None:
+                        ungrouped_standalone.append(item)
+                        continue
+                    key = (mid, item.get("base_model") or "") if group_by_base else mid
+                    model_groups.setdefault(key, []).append(item)
+                # Sort versions within each group by version id (descending);
+                # fall back to modified timestamp for non-CivitAI sources.
+                for items in model_groups.values():
+                    items.sort(
+                        key=lambda x: self._extract_version_id(x)
+                        or x.get("modified", 0)
+                        or 0,
+                        reverse=True,
+                    )
+                # Sort groups by version count
+                sorted_groups = sorted(
+                    model_groups.values(),
+                    key=lambda items: len(items),
+                    reverse=reverse,
+                )
+                # Flatten: grouped items first, standalone items last
+                sorted_data = []
+                for items in sorted_groups:
+                    sorted_data.extend(items)
+                sorted_data.extend(ungrouped_standalone)
+
         t1 = time.perf_counter()
         if hash_filters:
             filtered_data = await self._apply_hash_filters(sorted_data, hash_filters)
@@ -172,7 +275,7 @@ class BaseModelService(ABC):
         overall_duration = time.perf_counter() - overall_start
         logger.debug(
             "%s.get_paginated_data took %.3fs (fetch: %.3fs, filter: %.3fs, update_filter: %.3fs, pagination: %.3fs, annotate: %.3fs). "
-            "Counts: initial=%d, post_filter=%d, final=%d",
+            "Counts: initial=%d, dedup=%d, post_filter=%d, final=%d",
             self.__class__.__name__,
             overall_duration,
             fetch_duration,
@@ -181,6 +284,7 @@ class BaseModelService(ABC):
             pagination_duration,
             annotate_duration,
             initial_count,
+            dedup_lost,
             post_filter_count,
             final_count,
         )
@@ -495,7 +599,7 @@ class BaseModelService(ABC):
         if not ordered_ids:
             return annotated
 
-        strategy_value = self.settings.get("update_flag_strategy")
+        strategy_value = self.settings.get("version_grouping")
         if isinstance(strategy_value, str) and strategy_value.strip():
             strategy = strategy_value.strip().lower()
         else:
@@ -602,6 +706,33 @@ class BaseModelService(ABC):
 
         return annotated
 
+    @staticmethod
+    def _extract_hf_group_key(item: Dict) -> Optional[str]:
+        """Extract `hf:{owner}/{repo}` from item's ``hf_url``, or None."""
+        hf_url = item.get("hf_url") if isinstance(item, dict) else None
+        if not hf_url or not isinstance(hf_url, str):
+            return None
+        m = re.match(
+            r"https?://huggingface\.co/([^/]+/[^/]+)", hf_url.strip()
+        )
+        if not m:
+            return None
+        return f"hf:{m.group(1)}"
+
+    @staticmethod
+    def _extract_group_key(item: Dict) -> Union[int, str, None]:
+        """Return the group identity key: CivitAI modelId (int) or HF repo (str).
+
+        Preference order:
+        1. CivitAI ``modelId`` (int)
+        2. HF repo identity ``hf:{owner}/{repo}`` (str)
+        3. ``None`` (no known grouping source)
+        """
+        mid = BaseModelService._extract_model_id(item)
+        if mid is not None:
+            return mid
+        return BaseModelService._extract_hf_group_key(item)
+
     @staticmethod
     def _extract_model_id(item: Dict) -> Optional[int]:
         civitai = item.get("civitai") if isinstance(item, dict) else None
@@ -696,8 +827,12 @@ class BaseModelService(ABC):
         }
 
     @abstractmethod
-    async def format_response(self, model_data: Dict) -> Dict:
-        """Format model data for API response - must be implemented by subclasses"""
+    async def format_response(self, model_data: Dict) -> Optional[Dict]:
+        """Format model data for API response - must be implemented by subclasses.
+
+        Subclasses should return None for corrupted entries so the handler
+        layer can filter them out. See issue #730.
+        """
         pass
 
     # Common service methods that delegate to scanner
@@ -705,6 +840,12 @@ class BaseModelService(ABC):
         """Get top tags sorted by frequency"""
         return await self.scanner.get_top_tags(limit)
 
+    async def search_tags(
+        self, query: str, limit: int = 50
+    ) -> List[Dict]:
+        """Search tags by substring, sorted by frequency"""
+        return await self.scanner.search_tags(query, limit)
+
     async def get_base_models(self, limit: int = 20) -> List[Dict]:
         """Get base models sorted by frequency"""
         return await self.scanner.get_base_models(limit)
@@ -856,13 +997,21 @@ class BaseModelService(ABC):
 
         return unified_tree
 
-    async def get_model_notes(self, model_name: str) -> Optional[str]:
-        """Get notes for a specific model file"""
+    async def get_model_notes(self, model_name: str) -> Optional[dict]:
+        """Get notes and file_path for a specific model file.
+
+        Supports both simple names (``OWSMianne_ANIMA_V1``) and full-path
+        syntax (``Anima/character/OWSMianne_ANIMA_V1``).
+        """
         cache = await self.scanner.get_cached_data()
 
         for model in cache.raw_data:
-            if model["file_name"] == model_name:
-                return model.get("notes", "")
+            file_name = model.get("file_name", "")
+            if file_name == model_name or model_name.endswith("/" + file_name) or model_name.endswith("\\" + file_name):
+                return {
+                    "notes": model.get("notes", ""),
+                    "file_path": model.get("file_path", ""),
+                }
 
         return None
 
@@ -985,6 +1134,11 @@ class BaseModelService(ABC):
 
         Listing/search endpoints return lightweight cache entries; this method performs
         a lazy read of the on-disk metadata snapshot when callers need full detail.
+
+        As a beneficial side effect, the in-memory and persistent caches are
+        opportunistically synchronised with the on-disk metadata — this keeps the
+        caches fresh even when a ``.metadata.json`` file was edited outside of the
+        normal save path (e.g. manually or by an external script).
         """
         metadata, should_skip = await MetadataManager.load_metadata(
             file_path, self.metadata_class
@@ -1002,6 +1156,19 @@ class BaseModelService(ABC):
                 MetadataManager.save_metadata(file_path, metadata)
             )
 
+        # Opportunistically sync the in-memory + persistent caches.
+        # The .metadata.json disk read is already paid for; the sync only
+        # performs work when the cache is actually stale, and uses targeted,
+        # in-place operations to minimise overhead even with large model sets.
+        #
+        # Fire-and-forget by design: the task is intentionally untracked.
+        # sync_cache_from_metadata handles its own errors internally.
+        asyncio.create_task(
+            self.scanner.sync_cache_from_metadata(
+                file_path, metadata.to_dict()
+            )
+        )
+
         return self.filter_civitai_data(metadata.to_dict().get("civitai", {}))
 
     async def get_model_description(self, file_path: str) -> Optional[str]:

py/services/batch_import_service.py

@@ -523,6 +523,10 @@ class BatchImportService:
                 if payload.get("checkpoint"):
                     metadata["checkpoint"] = payload["checkpoint"]
 
+                nsfw = payload.get("preview_nsfw_level")
+                if isinstance(nsfw, int) and nsfw > 0:
+                    metadata["preview_nsfw_level"] = nsfw
+
                 image_bytes = None
                 image_base64 = payload.get("image_base64")
 

py/services/checkpoint_scanner.py

@@ -114,6 +114,13 @@ class CheckpointScanner(ModelScanner):
                 and metadata.hash_status == "completed"
                 and metadata.sha256
             ):
+                # Ensure the in-memory hash index is populated even when
+                # the hash was already computed and persisted to the metadata
+                # file.  Without this, usage tracking (and any other caller
+                # that queries get_hash_by_filename first) will miss on every
+                # lookup and keep calling back into this method, creating a
+                # tight loop that never populates the index.
+                self._hash_index.add_entry(metadata.sha256.lower(), file_path)
                 return metadata.sha256
 
             async with self._hash_calculation_lock:
@@ -125,6 +132,7 @@ class CheckpointScanner(ModelScanner):
                     and metadata.hash_status == "completed"
                     and metadata.sha256
                 ):
+                    self._hash_index.add_entry(metadata.sha256.lower(), file_path)
                     return metadata.sha256
 
                 task = self._hash_calculation_tasks.get(real_path)
@@ -175,6 +183,9 @@ class CheckpointScanner(ModelScanner):
 
             # Check if hash is already calculated
             if metadata.hash_status == "completed" and metadata.sha256:
+                # Populate the in-memory hash index even for pre-computed
+                # hashes, mirroring the fix in calculate_hash_for_model.
+                self._hash_index.add_entry(metadata.sha256.lower(), file_path)
                 return metadata.sha256
 
             # Update status to calculating
@@ -193,6 +204,20 @@ class CheckpointScanner(ModelScanner):
             # Update hash index
             self._hash_index.add_entry(sha256.lower(), file_path)
 
+            # Update the in-memory cache entry so that subsequent
+            # _persist_current_cache / _save_persistent_cache calls
+            # write the hash back to the SQLite models table.  Without
+            # this the hash only lives in the metadata file and the
+            # in-memory hash index, both of which are lost across
+            # restarts, causing the same re-computation loop on the
+            # next session.
+            if self._cache is not None and self._cache.raw_data:
+                for entry in self._cache.raw_data:
+                    if entry.get("file_path") == file_path:
+                        entry["sha256"] = sha256.lower()
+                        entry["hash_status"] = "completed"
+                        break
+
             logger.info(f"Hash calculated for checkpoint: {file_path}")
             return sha256
 

py/services/checkpoint_service.py

@@ -1,6 +1,6 @@
 import os
 import logging
-from typing import Dict
+from typing import Dict, Optional
 
 from .base_model_service import BaseModelService
 from .auto_tag_service import extract_auto_tags
@@ -21,20 +21,37 @@ class CheckpointService(BaseModelService):
         """
         super().__init__("checkpoint", scanner, CheckpointMetadata, update_service=update_service)
     
-    async def format_response(self, checkpoint_data: Dict) -> Dict:
-        """Format Checkpoint data for API response"""
+    async def format_response(self, checkpoint_data: Dict) -> Optional[Dict]:
+        """Format Checkpoint data for API response.
+
+        Returns None when the entry is missing critical fields (corrupted cache
+        row), so the handler layer can filter it out. See issue #730.
+        """
+        # Guard against corrupted cache entries missing critical fields
+        file_path = checkpoint_data.get("file_path")
+        if not file_path or not isinstance(file_path, str):
+            logger.warning(
+                "Skipping corrupted checkpoint entry (missing file_path): %s",
+                checkpoint_data.get("file_name", "<unknown>"),
+            )
+            return None
+
         # Get sub_type from cache entry (new canonical field)
         sub_type = checkpoint_data.get("sub_type", "checkpoint")
-        
+
+        file_name = checkpoint_data.get("file_name") or ""
+        model_name = checkpoint_data.get("model_name") or file_name
+        folder = checkpoint_data.get("folder") or ""
+
         return {
-            "model_name": checkpoint_data["model_name"],
-            "file_name": checkpoint_data["file_name"],
+            "model_name": model_name,
+            "file_name": file_name,
             "preview_url": config.get_preview_static_url(checkpoint_data.get("preview_url", "")),
             "preview_nsfw_level": checkpoint_data.get("preview_nsfw_level", 0),
             "base_model": checkpoint_data.get("base_model", ""),
-            "folder": checkpoint_data["folder"],
+            "folder": folder,
             "sha256": checkpoint_data.get("sha256", ""),
-            "file_path": checkpoint_data["file_path"].replace(os.sep, "/"),
+            "file_path": file_path.replace(os.sep, "/"),
             "file_size": checkpoint_data.get("size", 0),
             "modified": checkpoint_data.get("modified", ""),
             "tags": checkpoint_data.get("tags", []),
@@ -48,6 +65,8 @@ class CheckpointService(BaseModelService):
             "skip_metadata_refresh": bool(checkpoint_data.get("skip_metadata_refresh", False)),
             "civitai": self.filter_civitai_data(checkpoint_data.get("civitai", {}), minimal=True),
             "auto_tags": checkpoint_data.get("auto_tags") or extract_auto_tags(checkpoint_data),
+            "version_count": checkpoint_data.get("version_count"),
+            "hf_url": checkpoint_data.get("hf_url", ""),
         }
     
     def find_duplicate_hashes(self) -> Dict:

py/services/civarchive_client.py

@@ -304,6 +304,20 @@ class CivArchiveClient:
             version_id = file_data.get("model_version_id") or file_data.get("modelVersionId")
             if model_id is None or version_id is None:
                 continue
+            # CivitAI / CivArchive model IDs are small integers (typically ≤ 7
+            # digits).  Reject suspiciously large values that indicate the API
+            # returned a malformed payload (e.g. a hash reinterpreted as an ID)
+            # to avoid pointless HTTP 500 errors from CivArchive.
+            _MAX_VALID_CIVITAI_ID = 100_000_000
+            try:
+                if int(model_id) >= _MAX_VALID_CIVITAI_ID or int(version_id) >= _MAX_VALID_CIVITAI_ID:
+                    logger.debug(
+                        "Skipping implausible CivArchive model_id=%s / version_id=%s",
+                        model_id, version_id,
+                    )
+                    continue
+            except (TypeError, ValueError):
+                continue
             resolved = await self.get_model_version(model_id, version_id)
             if resolved:
                 return resolved
@@ -327,7 +341,7 @@ class CivArchiveClient:
             if resolved:
                 return resolved, None
 
-            logger.error("Error fetching version of CivArchive model by hash %s", model_hash[:10])
+            logger.debug("Error fetching version of CivArchive model by hash %s", model_hash[:10])
             return None, "No version data found"
 
         except RateLimitError:
@@ -417,7 +431,7 @@ class CivArchiveClient:
 
             if version_id is not None:
                 raw_id = version_data.get("id")
-                if raw_id != version_id:
+                if raw_id is not None and str(raw_id) != str(version_id):
                     logger.warning(
                         "Requested version %s doesn't match default version %s for model %s",
                         version_id,

py/services/civitai_base_model_service.py

@@ -196,6 +196,7 @@ class CivitaiBaseModelService:
             "ernie": "ERNI",
             "ernie turbo": "ETRB",
             "nucleus": "NUCL",
+            "krea 2": "KR2",
             "svd": "SVD",
             "ltxv": "LTXV",
             "ltxv2": "LTV2",
@@ -212,6 +213,18 @@ class CivitaiBaseModelService:
             "wan video 2.2 i2v-a14b": "WAN",
             "wan video 2.5 t2v": "WAN",
             "wan video 2.5 i2v": "WAN",
+            "wan video 2.7": "WAN",
+            "wan image 2.7": "WI27",
+            "ace audio": "ACE",
+            "boogu": "BOOG",
+            "grok": "GROK",
+            "happyhorse": "HAPP",
+            "hidream-o1": "HIO1",
+            "lens": "LENS",
+            "mai": "MAI",
+            "upscaler": "UPSC",
+            "ideogram 4.0": "ID40",
+            "qwen 2": "QWN2",
         }
 
         if lower_name in special_cases:
@@ -391,6 +404,7 @@ class CivitaiBaseModelService:
                 "LTXV2",
                 "LTXV 2.3",
                 "CogVideoX",
+                "HappyHorse",
                 "Mochi",
                 "Hunyuan Video",
                 "Wan Video",
@@ -403,15 +417,25 @@ class CivitaiBaseModelService:
                 "Wan Video 2.2 I2V-A14B",
                 "Wan Video 2.5 T2V",
                 "Wan Video 2.5 I2V",
+                "Wan Image 2.7",
+                "Wan Video 2.7",
             ],
             "Other Models": [
+                "ACE Audio",
                 "Illustrious",
                 "Pony",
                 "Pony V7",
+                "Boogu",
                 "HiDream",
+                "HiDream-O1",
+                "Ideogram 4.0",
                 "Qwen",
+                "Qwen 2",
                 "AuraFlow",
                 "Chroma",
+                "Grok",
+                "Lens",
+                "MAI",
                 "ZImageTurbo",
                 "ZImageBase",
                 "PixArt a",
@@ -424,6 +448,8 @@ class CivitaiBaseModelService:
                 "Ernie",
                 "Ernie Turbo",
                 "Nucleus",
+                "Krea 2",
+                "Upscaler",
             ],
         }
 

py/services/civitai_client.py

@@ -56,7 +56,7 @@ class CivitaiClient:
         self._MAX_CACHE_ENTRIES = 500
 
     def _build_image_info_url(self, image_id: str) -> str:
-        return f"{self.base_url}/images?imageId={image_id}&nsfw=X"
+        return f"{self.base_url}/images?imageId={image_id}&nsfw=X&withMeta=true"
 
     async def _make_request(
         self,

py/services/download_manager.py

@@ -29,6 +29,7 @@ from .metadata_service import get_default_metadata_provider, get_metadata_provid
 from .downloader import get_downloader, DownloadProgress, DownloadStreamControl
 from .aria2_downloader import Aria2Error, get_aria2_downloader
 from .aria2_transfer_state import Aria2TransferStateStore
+from .download_queue_service import DownloadQueueService
 
 # Download to temporary file first
 import tempfile
@@ -229,6 +230,12 @@ class DownloadManager:
         Returns:
             Dict with download result
         """
+        logger.debug(
+            "[download] download_from_civitai called: model_id=%s, model_version_id=%s, "
+            "source=%s, file_params=%s",
+            model_id, model_version_id, source, file_params,
+        )
+
         # Validate that at least one identifier is provided
         if not model_id and not model_version_id:
             return {
@@ -249,6 +256,7 @@ class DownloadManager:
             "source": source,
             "file_params": copy.deepcopy(file_params) if file_params is not None else None,
             "progress": 0,
+
             "status": "queued",
             "transfer_backend": self._get_model_download_backend(),
             "bytes_downloaded": 0,
@@ -288,8 +296,8 @@ class DownloadManager:
             return result
         except asyncio.CancelledError:
             return {
-                "success": False,
-                "error": "Download was cancelled",
+                "success": True,
+                "cancelled": True,
                 "download_id": task_id,
             }
         finally:
@@ -360,6 +368,15 @@ class DownloadManager:
                     if self._active_downloads[task_id].get("transfer_backend") == "aria2":
                         await self._persist_aria2_state(task_id)
 
+                # Update SQLite queue status to 'downloading'
+                try:
+                    queue_service = await DownloadQueueService.get_instance()
+                    await queue_service.update_status(task_id, "downloading")
+                except Exception:
+                    logger.warning(
+                        "Failed to update queue status for %s", task_id, exc_info=True
+                    )
+
                 # Use original download implementation
                 try:
                     # Check for cancellation before starting
@@ -396,6 +413,22 @@ class DownloadManager:
                         if self._active_downloads[task_id].get("transfer_backend") == "aria2":
                             await self._persist_aria2_state(task_id)
 
+                    # Move queue item to history on completion
+                    try:
+                        queue_service = await DownloadQueueService.get_instance()
+                        await queue_service.complete_download(
+                            download_id=task_id,
+                            status=result.get("status", "completed") if result.get("success") else "failed",
+                            error=result.get("error") if not result.get("success") else None,
+                            file_path=result.get("file_path"),
+                            bytes_downloaded=self._active_downloads.get(task_id, {}).get("bytes_downloaded", 0),
+                            total_bytes=self._active_downloads.get(task_id, {}).get("total_bytes"),
+                        )
+                    except Exception:
+                        logger.warning(
+                            "Failed to complete queue item for %s", task_id, exc_info=True
+                        )
+
                     return result
                 except asyncio.CancelledError:
                     # Handle cancellation
@@ -404,6 +437,19 @@ class DownloadManager:
                         self._active_downloads[task_id]["bytes_per_second"] = 0.0
                         if self._active_downloads[task_id].get("transfer_backend") == "aria2":
                             await self._persist_aria2_state(task_id)
+
+                    # Move queue item to history as canceled
+                    try:
+                        queue_service = await DownloadQueueService.get_instance()
+                        await queue_service.complete_download(
+                            download_id=task_id,
+                            status="canceled",
+                        )
+                    except Exception:
+                        logger.warning(
+                            "Failed to cancel queue item for %s", task_id, exc_info=True
+                        )
+
                     logger.info(f"Download cancelled for task {task_id}")
                     raise
                 except Exception as e:
@@ -417,6 +463,22 @@ class DownloadManager:
                         self._active_downloads[task_id]["bytes_per_second"] = 0.0
                         if self._active_downloads[task_id].get("transfer_backend") == "aria2":
                             await self._persist_aria2_state(task_id)
+
+                    # Move queue item to history as failed
+                    try:
+                        queue_service = await DownloadQueueService.get_instance()
+                        await queue_service.complete_download(
+                            download_id=task_id,
+                            status="failed",
+                            error=str(e),
+                            bytes_downloaded=self._active_downloads.get(task_id, {}).get("bytes_downloaded", 0),
+                            total_bytes=self._active_downloads.get(task_id, {}).get("total_bytes"),
+                        )
+                    except Exception:
+                        logger.warning(
+                            "Failed to complete queue item for %s", task_id, exc_info=True
+                        )
+
                     return {"success": False, "error": str(e)}
         finally:
             # Schedule cleanup of download record after delay
@@ -620,7 +682,10 @@ class DownloadManager:
                     u for u in download_urls if not u.startswith(CIVITAI_DOWNLOAD_URL_PREFIXES)
                 ]
                 download_urls = non_civitai_urls + civitai_urls
-        else:
+
+        # Fallback: when mirrors is empty or all mirrors have been deleted,
+        # use the file's downloadUrl directly (e.g. CivitAI download endpoint).
+        if not download_urls:
             download_url = file_info.get("downloadUrl")
             if download_url:
                 download_urls.append(normalize_civitai_download_url(download_url))
@@ -1233,10 +1298,24 @@ class DownloadManager:
                     "download_id": download_id,
                 }
 
-            # Check if this checkpoint should be treated as a diffusion model based on baseModel
+            # Check if this checkpoint should be treated as a diffusion model
+            # Priority: (1) any file has type "UNet" or "Diffusion Model",
+            #            (2) baseModel is in DIFFUSION_MODEL_BASE_MODELS
             is_diffusion_model = False
             if model_type == "checkpoint":
-                if base_model_value in DIFFUSION_MODEL_BASE_MODELS:
+                # Check file types first (more direct signal from CivitAI)
+                version_files = version_info.get("files", [])
+                for f in version_files:
+                    f_type = f.get("type", "")
+                    if f_type in ("UNet", "Diffusion Model"):
+                        is_diffusion_model = True
+                        logger.info(
+                            f"File type '{f_type}' detected, routing checkpoint to unet folder"
+                        )
+                        break
+
+                # Fallback to baseModel name check
+                if not is_diffusion_model and base_model_value in DIFFUSION_MODEL_BASE_MODELS:
                     is_diffusion_model = True
                     logger.info(
                         f"baseModel '{base_model_value}' is a known diffusion model, routing to unet folder"
@@ -1310,7 +1389,17 @@ class DownloadManager:
 
             # Update save directory with relative path if provided
             if relative_path:
+                base_save_dir = save_dir
                 save_dir = os.path.join(save_dir, relative_path)
+                # Security: validate path containment after joining
+                resolved_dir = os.path.abspath(os.path.normpath(save_dir))
+                base_dir = os.path.abspath(os.path.normpath(base_save_dir))
+                if not resolved_dir.startswith(base_dir + os.sep) and resolved_dir != base_dir:
+                    logger.warning(
+                        "Path traversal detected: %s escapes %s",
+                        resolved_dir, base_dir,
+                    )
+                    return {"success": False, "error": "Download path is outside allowed directory"}
                 # Create directory if it doesn't exist
                 os.makedirs(save_dir, exist_ok=True)
 
@@ -1352,86 +1441,100 @@ class DownloadManager:
 
             # If file_params is provided, try to find matching file
             if file_params and model_version_id:
+                target_file_id = file_params.get("id")
                 target_type = file_params.get("type", "Model")
-                target_format = file_params.get("format", "SafeTensor")
-                target_size = file_params.get("size", "full")
+                target_format = file_params.get("format")
+                target_size = file_params.get("size")
                 target_fp = file_params.get("fp")
                 is_primary = file_params.get("isPrimary", False)
 
-                if is_primary:
-                    # Find primary file
+                logger.debug(
+                    "[download] file_params received: id=%s, type=%s, format=%s, size=%s, fp=%s, isPrimary=%s, "
+                    "model_version_id=%s, total_files=%d",
+                    target_file_id, target_type, target_format, target_size, target_fp, is_primary,
+                    model_version_id, len(files),
+                )
+
+                if target_file_id:
+                    target_id_str = str(target_file_id)
+                    for f in files:
+                        f_id = f.get("id")
+                        if str(f_id) == target_id_str:
+                            file_info = f
+                            logger.debug(
+                                "[download] MATCH by ID: id=%s name='%s'",
+                                f_id, f.get("name"),
+                            )
+                            break
+                    if not file_info:
+                        logger.debug("[download] No file found with id=%s", target_file_id)
+
+                elif is_primary:
                     file_info = next(
                         (
                             f
                             for f in files
                             if f.get("primary")
-                            and f.get("type") in ("Model", "Negative", "Diffusion Model")
+                            and f.get("type") in ("Model", "Negative", "Diffusion Model", "UNet")
                         ),
                         None,
                     )
                 else:
-                    # Match by metadata
+                    # Lenient metadata match: only compare fields present on both sides
                     for f in files:
                         f_type = f.get("type", "")
-                        f_meta = f.get("metadata", {})
-
-                        # Check type match
                         if f_type != target_type:
                             continue
 
-                        # Check metadata match
-                        if f_meta.get("format") != target_format:
+                        f_meta = f.get("metadata", {})
+                        f_format = f_meta.get("format") or f.get("format")
+                        f_size = f_meta.get("size") or f.get("size")
+                        f_fp = f_meta.get("fp") or f.get("fp")
+
+                        if target_format and f_format != target_format:
                             continue
-                        if f_meta.get("size") != target_size:
+                        if target_size and f_size and f_size != target_size:
                             continue
-                        if target_fp and f_meta.get("fp") != target_fp:
+                        if target_fp and f_fp and f_fp != target_fp:
                             continue
 
                         file_info = f
                         break
 
+                if not file_info:
+                    logger.debug(
+                        "[download] No match found via file_params — falling back to primary file lookup",
+                    )
+            elif not file_params:
+                logger.debug(
+                    "[download] No file_params provided (null/None) — will use primary file lookup. "
+                    "model_version_id=%s, total_files=%d",
+                    model_version_id, len(files),
+                )
+
             # Fallback to primary file if no match found
             if not file_info:
+                logger.debug("[download] Looking for primary file as fallback")
                 file_info = next(
                     (
                         f
                         for f in files
-                        if f.get("primary") and f.get("type") in ("Model", "Negative", "Diffusion Model")
+                        if f.get("primary") and f.get("type") in ("Model", "Negative", "Diffusion Model", "UNet")
                     ),
                     None,
                 )
+                if file_info:
+                    logger.debug(
+                        "[download] Fallback primary file selected: id=%s, name=%s",
+                        file_info.get("id"), file_info.get("name"),
+                    )
+                else:
+                    logger.debug("[download] No primary file found in fallback lookup")
 
             if not file_info:
                 return {"success": False, "error": "No suitable file found in metadata"}
-            mirrors = file_info.get("mirrors") or []
-            download_urls = []
-            if mirrors:
-                for mirror in mirrors:
-                    if mirror.get("deletedAt") is None and mirror.get("url"):
-                        download_urls.append(
-                            normalize_civitai_download_url(mirror["url"])
-                        )
 
-                # When source is 'civarchive', prioritize non-Civitai URLs
-                # This avoids failed downloads from deleted Civitai models
-                if source == "civarchive" and len(download_urls) > 1:
-                    civitai_urls = [
-                        u
-                        for u in download_urls
-                        if u.startswith(CIVITAI_DOWNLOAD_URL_PREFIXES)
-                    ]
-                    non_civitai_urls = [
-                        u
-                        for u in download_urls
-                        if not u.startswith(CIVITAI_DOWNLOAD_URL_PREFIXES)
-                    ]
-                    download_urls = non_civitai_urls + civitai_urls
-            else:
-                download_url = file_info.get("downloadUrl")
-                if download_url:
-                    download_urls.append(
-                        normalize_civitai_download_url(download_url)
-                    )
+            download_urls = self._build_download_urls_from_file_info(file_info, source=source)
 
             if not download_urls:
                 return {"success": False, "error": "No mirror URL found"}
@@ -1734,6 +1837,9 @@ class DownloadManager:
             model_tags, model_type
         )
 
+        if not first_tag:
+            first_tag = "no tags"  # Default if no tags available
+
         # Format the template with available data
         formatted_path = path_template
         formatted_path = formatted_path.replace("{base_model}", mapped_base_model)
@@ -1749,6 +1855,15 @@ class DownloadManager:
         if model_type == "embedding":
             formatted_path = formatted_path.replace(" ", "_")
 
+        # Sanitize the resolved path to prevent path traversal:
+        # - Strip leading slashes (prevents os.path.join from treating path as absolute)
+        # - Collapse double slashes from empty placeholder substitutions
+        # - Strip trailing slashes for cleanliness
+        formatted_path = formatted_path.lstrip("/")
+        while "//" in formatted_path:
+            formatted_path = formatted_path.replace("//", "/")
+        formatted_path = formatted_path.rstrip("/")
+
         return formatted_path
 
     async def _execute_download(
@@ -1974,7 +2089,21 @@ class DownloadManager:
                     break
 
                 last_error = result
-                if os.path.exists(save_path):
+                # For aria2: if the .aria2 control file is missing, aria2 considers
+                # the download complete.  A transient RPC failure may have made us
+                # think the download failed even though the file is fully on disk.
+                # Keep the file so a retry can find it already complete.
+                if (
+                    transfer_backend == "aria2"
+                    and os.path.exists(save_path)
+                    and not os.path.exists(f"{save_path}.aria2")
+                ):
+                    logger.warning(
+                        "aria2 download reported failure but .aria2 file is absent "
+                        "for %s — the file is likely complete.  Preserving it for retry.",
+                        save_path,
+                    )
+                elif os.path.exists(save_path):
                     try:
                         os.remove(save_path)
                     except Exception as e:

py/services/download_queue_service.py

@@ -31,7 +31,7 @@ class DownloadQueueService:
     _instance: Optional[DownloadQueueService] = None
     _class_lock: asyncio.Lock = asyncio.Lock()
 
-    _SCHEMA = """
+    _SCHEMA_TABLES = """
         CREATE TABLE IF NOT EXISTS download_queue (
             download_id TEXT PRIMARY KEY,
             model_id INTEGER,
@@ -76,6 +76,11 @@ class DownloadQueueService:
         CREATE INDEX IF NOT EXISTS idx_dh_status ON download_history(status);
     """
 
+    _CREATE_UNIQUE_INDEX = """
+        CREATE UNIQUE INDEX IF NOT EXISTS idx_dh_download_id
+            ON download_history(download_id) WHERE download_id IS NOT NULL;
+    """
+
     @classmethod
     async def get_instance(cls) -> DownloadQueueService:
         """Return the singleton instance, creating it if necessary."""
@@ -113,10 +118,39 @@ class DownloadQueueService:
         if self._schema_initialized:
             return
         with self._connect() as conn:
-            conn.executescript(self._SCHEMA)
+            conn.executescript(self._SCHEMA_TABLES)
+
+            # Creating the unique index on download_history.download_id can
+            # fail if pre-existing rows have duplicate values (e.g. from a
+            # previous version that lacked the index).  Deduplicate first so
+            # that the migration does not crash on startup.
+            if not self._index_exists(conn, "idx_dh_download_id"):
+                self._remove_duplicate_download_ids(conn)
+                conn.executescript(self._CREATE_UNIQUE_INDEX)
+
             conn.commit()
         self._schema_initialized = True
 
+    @staticmethod
+    def _index_exists(conn: sqlite3.Connection, name: str) -> bool:
+        return conn.execute(
+            "SELECT 1 FROM sqlite_master WHERE type='index' AND name=?",
+            (name,),
+        ).fetchone() is not None
+
+    @staticmethod
+    def _remove_duplicate_download_ids(conn: sqlite3.Connection) -> None:
+        conn.execute("""
+            DELETE FROM download_history
+            WHERE id NOT IN (
+                SELECT MIN(id)
+                FROM download_history
+                WHERE download_id IS NOT NULL
+                GROUP BY download_id
+            )
+            AND download_id IS NOT NULL
+        """)
+
     def get_database_path(self) -> str:
         """Return the resolved database file path."""
         return self._db_path
@@ -154,13 +188,23 @@ class DownloadQueueService:
         """Insert a new download into the queue.
 
         Returns the inserted row as a dict (or an empty dict if the
-        download_id already exists).
+        download_id already exists in the queue or has a terminal
+        record in history).
         """
         now = time.time()
         file_params_json = json.dumps(file_params) if file_params is not None else None
 
         async with self._lock:
             conn = self._get_conn()
+
+            # Reject download_ids that already have a terminal record in history.
+            history_row = conn.execute(
+                "SELECT 1 FROM download_history WHERE download_id = ? LIMIT 1",
+                (download_id,),
+            ).fetchone()
+            if history_row is not None:
+                return {}
+
             conn.execute(
                 """
                 INSERT OR IGNORE INTO download_queue (
@@ -380,7 +424,7 @@ class DownloadQueueService:
             )
             conn.execute(
                 """
-                INSERT INTO download_history (
+                INSERT OR IGNORE INTO download_history (
                     download_id, model_id, model_version_id, model_name,
                     version_name, thumbnail_url, status, error, file_path,
                     bytes_downloaded, total_bytes, completed_at
@@ -537,17 +581,27 @@ class DownloadQueueService:
             "offset": offset,
         }
 
-    async def delete_history_item(self, id: int) -> bool:
-        """Delete a single history entry by its *id*.
+    async def delete_history_item(
+        self, id: Optional[int] = None, download_id: Optional[str] = None
+    ) -> bool:
+        """Delete a single history entry by *download_id* (preferred) or *id*.
 
         Returns ``True`` if a row was deleted.
         """
         async with self._lock:
             conn = self._get_conn()
-            cursor = conn.execute(
-                "DELETE FROM download_history WHERE id = ?",
-                (id,),
-            )
+            if download_id:
+                cursor = conn.execute(
+                    "DELETE FROM download_history WHERE download_id = ?",
+                    (download_id,),
+                )
+            elif id is not None:
+                cursor = conn.execute(
+                    "DELETE FROM download_history WHERE id = ?",
+                    (id,),
+                )
+            else:
+                return False
             conn.commit()
         return cursor.rowcount > 0
 
@@ -604,21 +658,34 @@ class DownloadQueueService:
     # Retry
     # ------------------------------------------------------------------
 
-    async def retry_from_history(self, item_id: int) -> Optional[dict[str, Any]]:
+    async def retry_from_history(
+        self,
+        item_id: Optional[int] = None,
+        download_id: Optional[str] = None,
+    ) -> Optional[dict[str, Any]]:
         """Re-queue a failed or canceled download from history.
 
-        Looks up the history record by its primary key.  If the status is
-        ``failed`` or ``canceled`` a new queue entry is created with the
-        same model metadata and a fresh download id, and the original
-        history entry is **deleted** to prevent exponential growth when
-        the retried item is later canceled or fails again and re-retried.
+        Looks up the history record by *download_id* (preferred) or
+        *item_id*.  If the status is ``failed`` or ``canceled`` a new
+        queue entry is created with the same model metadata and a fresh
+        download id, and the original history entry is **deleted** to
+        prevent exponential growth when the retried item is later
+        canceled or fails again and re-retried.
         """
         async with self._lock:
             conn = self._get_conn()
-            row = conn.execute(
-                "SELECT * FROM download_history WHERE id = ?",
-                (item_id,),
-            ).fetchone()
+            if download_id:
+                row = conn.execute(
+                    "SELECT * FROM download_history WHERE download_id = ?",
+                    (download_id,),
+                ).fetchone()
+            elif item_id is not None:
+                row = conn.execute(
+                    "SELECT * FROM download_history WHERE id = ?",
+                    (item_id,),
+                ).fetchone()
+            else:
+                return None
             if row is None:
                 return None
             status = str(row["status"])
@@ -650,7 +717,7 @@ class DownloadQueueService:
             )
             conn.execute(
                 "DELETE FROM download_history WHERE id = ?",
-                (item_id,),
+                (row["id"],),
             )
             conn.commit()
             queued = conn.execute(

py/services/downloader.py

@@ -46,6 +46,30 @@ def is_ssl_cert_verify_error(exc: BaseException) -> bool:
     return "CERTIFICATE_VERIFY_FAILED" in str(exc)
 
 
+def _parse_retry_after(value: str) -> int:
+    """Parse a Retry-After header value into seconds.
+
+    Supports both integer seconds and HTTP-date formats.
+    Returns a default of 60 seconds on invalid/missing input.
+    """
+    if not value or not value.strip():
+        return 60
+
+    value = value.strip()
+    try:
+        return max(1, int(value))
+    except ValueError:
+        pass
+
+    try:
+        parsed = parsedate_to_datetime(value)
+        now = datetime.now().astimezone()
+        delta = (parsed - now).total_seconds()
+        return max(1, int(delta))
+    except (ValueError, OverflowError, OSError):
+        return 60
+
+
 @dataclass(frozen=True)
 class DownloadProgress:
     """Snapshot of a download transfer at a moment in time."""
@@ -246,14 +270,14 @@ class Downloader:
 
         Note: This is private and caller MUST hold self._session_lock.
         """
-        # Close existing session if any
-        if self._session is not None:
-            try:
-                await self._session.close()
-            except Exception as e:  # pragma: no cover
-                logger.warning(f"Error closing previous session: {e}")
-            finally:
-                self._session = None
+        # Snapshot and clear old session reference before creating the new
+        # one.  This ensures self._session is always valid (or None, which
+        # triggers a fresh creation) and avoids a race where concurrent
+        # requests hold a reference to a session whose connector has been
+        # torn down by a premature close() call — the root cause of the
+        # intermittent "NoneType has no attribute connect" crash.
+        old_session = self._session
+        self._session = None
 
         # Check for app-level proxy settings
         proxy_url = None  # http(s) proxy, passed via the per-request `proxy=` kwarg
@@ -348,6 +372,13 @@ class Downloader:
         self._proxy_url = proxy_url
         self._session_created_at = datetime.now()
 
+        # Close the previous session now that the replacement is live.
+        if old_session is not None:
+            try:
+                await old_session.close()
+            except Exception as e:  # pragma: no cover
+                logger.warning(f"Error closing previous session: {e}")
+
         logger.debug(
             "Created new HTTP session with proxy settings. App-level proxy: %s, System-level proxy (trust_env): %s",
             bool(proxy_url),
@@ -729,7 +760,8 @@ class Downloader:
                             else:
                                 resume_offset = 0
                                 total_size = 0
-                            await self._create_session()
+                            async with self._session_lock:
+                                await self._create_session()
                             continue
 
                         return False, integrity_error
@@ -819,7 +851,8 @@ class Downloader:
                         logger.info(f"Will resume from byte {resume_offset}")
 
                     # Refresh session to get new connection
-                    await self._create_session()
+                    async with self._session_lock:
+                        await self._create_session()
                     continue
                 else:
                     logger.error(f"Max retries exceeded for download: {e}")
@@ -911,6 +944,19 @@ class Downloader:
                 elif response.status == 404:
                     error_msg = "File not found"
                     return False, error_msg, None
+                elif response.status == 429:
+                    raw_retry_after = response.headers.get("Retry-After")
+                    retry_after = _parse_retry_after(raw_retry_after or "")
+                    if raw_retry_after:
+                        logger.warning(
+                            "Rate limited (429) for %s, Retry-After: %ss", url, retry_after
+                        )
+                    else:
+                        logger.warning(
+                            "Rate limited (429) for %s, no Retry-After header; defaulting to %ss",
+                            url, retry_after,
+                        )
+                    return False, f"Rate limited (429), retry after {retry_after}s", None
                 else:
                     error_msg = f"Download failed with status {response.status}"
                     return False, error_msg, None

py/services/embedding_service.py

@@ -1,6 +1,6 @@
 import os
 import logging
-from typing import Dict
+from typing import Dict, Optional
 
 from .base_model_service import BaseModelService
 from .auto_tag_service import extract_auto_tags
@@ -21,20 +21,37 @@ class EmbeddingService(BaseModelService):
         """
         super().__init__("embedding", scanner, EmbeddingMetadata, update_service=update_service)
     
-    async def format_response(self, embedding_data: Dict) -> Dict:
-        """Format Embedding data for API response"""
+    async def format_response(self, embedding_data: Dict) -> Optional[Dict]:
+        """Format Embedding data for API response.
+
+        Returns None when the entry is missing critical fields (corrupted cache
+        row), so the handler layer can filter it out. See issue #730.
+        """
+        # Guard against corrupted cache entries missing critical fields
+        file_path = embedding_data.get("file_path")
+        if not file_path or not isinstance(file_path, str):
+            logger.warning(
+                "Skipping corrupted embedding entry (missing file_path): %s",
+                embedding_data.get("file_name", "<unknown>"),
+            )
+            return None
+
         # Get sub_type from cache entry (new canonical field)
         sub_type = embedding_data.get("sub_type", "embedding")
-        
+
+        file_name = embedding_data.get("file_name") or ""
+        model_name = embedding_data.get("model_name") or file_name
+        folder = embedding_data.get("folder") or ""
+
         return {
-            "model_name": embedding_data["model_name"],
-            "file_name": embedding_data["file_name"],
+            "model_name": model_name,
+            "file_name": file_name,
             "preview_url": config.get_preview_static_url(embedding_data.get("preview_url", "")),
             "preview_nsfw_level": embedding_data.get("preview_nsfw_level", 0),
             "base_model": embedding_data.get("base_model", ""),
-            "folder": embedding_data["folder"],
+            "folder": folder,
             "sha256": embedding_data.get("sha256", ""),
-            "file_path": embedding_data["file_path"].replace(os.sep, "/"),
+            "file_path": file_path.replace(os.sep, "/"),
             "file_size": embedding_data.get("size", 0),
             "modified": embedding_data.get("modified", ""),
             "tags": embedding_data.get("tags", []),
@@ -48,6 +65,8 @@ class EmbeddingService(BaseModelService):
             "skip_metadata_refresh": bool(embedding_data.get("skip_metadata_refresh", False)),
             "civitai": self.filter_civitai_data(embedding_data.get("civitai", {}), minimal=True),
             "auto_tags": embedding_data.get("auto_tags") or extract_auto_tags(embedding_data),
+            "version_count": embedding_data.get("version_count"),
+            "hf_url": embedding_data.get("hf_url", ""),
         }
     
     def find_duplicate_hashes(self) -> Dict:

py/services/errors.py

@@ -25,3 +25,21 @@ class ResourceNotFoundError(RuntimeError):
 
     pass
 
+
+class LLMNotConfiguredError(RuntimeError):
+    """Raised when an LLM-dependent operation is attempted but no provider is configured."""
+
+    pass
+
+
+class LLMRateLimitError(RateLimitError):
+    """Raised when the LLM provider rejects a request due to rate limiting."""
+
+    pass
+
+
+class LLMResponseError(RuntimeError):
+    """Raised when the LLM returns an unparseable or schema-invalid response."""
+
+    pass
+

py/services/llm_service.py

@@ -0,0 +1,729 @@
+"""Centralized LLM API client with BYOK (bring-your-own-key) provider support.
+
+Reads provider configuration from :class:`SettingsManager` and makes
+OpenAI-compatible ``/chat/completions`` calls.  Supports any provider that
+implements the OpenAI Chat Completions API surface area (OpenAI, Ollama,
+vLLM, LM Studio, etc.).
+"""
+
+from __future__ import annotations
+
+import asyncio
+import json
+import logging
+from typing import Any, Dict, List, Optional
+
+import aiohttp
+
+from .errors import LLMNotConfiguredError, LLMRateLimitError, LLMResponseError
+
+logger = logging.getLogger(__name__)
+
+# ---------------------------------------------------------------------------
+# Model catalog sourced from opencode's maintained model registry.
+# maps provider_id -> list of model IDs.
+# ---------------------------------------------------------------------------
+_MODEL_CATALOG_URL = "https://models.dev/api.json"
+
+# In-memory cache: maps provider slug -> list of model ID strings.
+_catalog_cache: Optional[Dict[str, List[str]]] = None
+
+# Per-model max output token limits parsed from the catalog.
+# ``{provider_id: {model_id: max_output_tokens}}``.
+_model_output_limits: Dict[str, Dict[str, int]] = {}
+
+_CATALOG_TIMEOUT = aiohttp.ClientTimeout(total=30)
+
+
+async def _load_model_catalog() -> Dict[str, List[str]]:
+    """Fetch and parse the model catalog.
+
+    Returns ``{provider_id: [model_id, ...]}`` and also populates
+    :data:`_model_output_limits` with per-model ``limit.output`` values
+    for use by :func:`_get_model_max_output`.
+
+    The JSON at ``_MODEL_CATALOG_URL`` is a dict keyed by provider slug; each
+    value has a ``models`` sub-dict keyed by model ID.  The result is cached
+    in memory after the first successful fetch.
+    Subsequent calls return the cached data immediately.
+    """
+    global _catalog_cache, _model_output_limits
+    if _catalog_cache is not None:
+        return _catalog_cache
+
+    try:
+        async with aiohttp.ClientSession(timeout=_CATALOG_TIMEOUT) as session:
+            async with session.get(_MODEL_CATALOG_URL) as resp:
+                if resp.status != 200:
+                    logger.warning("Model catalog returned HTTP %s", resp.status)
+                    return _catalog_cache or {}
+                data = await resp.json()
+    except (aiohttp.ClientError, asyncio.TimeoutError, json.JSONDecodeError) as exc:
+        logger.warning("Failed to fetch model catalog: %s", exc)
+        return _catalog_cache or {}
+
+    if not isinstance(data, dict):
+        logger.warning("Model catalog is not a dict, got %s", type(data).__name__)
+        return _catalog_cache or {}
+
+    result: Dict[str, List[str]] = {}
+    output_limits: Dict[str, Dict[str, int]] = {}
+    for provider_id, provider_info in data.items():
+        if not isinstance(provider_info, dict):
+            continue
+        models_dict = provider_info.get("models")
+        if not isinstance(models_dict, dict):
+            continue
+        model_ids: List[str] = []
+        provider_limits: Dict[str, int] = {}
+        for mid, model_info in models_dict.items():
+            if not isinstance(mid, str):
+                continue
+            model_ids.append(mid)
+            if isinstance(model_info, dict):
+                limit = model_info.get("limit")
+                if isinstance(limit, dict):
+                    output = limit.get("output")
+                    if isinstance(output, (int, float)) and output > 0:
+                        provider_limits[mid] = int(output)
+        if model_ids:
+            result[provider_id] = model_ids
+        if provider_limits:
+            output_limits[provider_id] = provider_limits
+
+    _catalog_cache = result
+    _model_output_limits = output_limits
+    logger.debug(
+        "Loaded model catalog: %d providers, %d total models "
+        "(%d providers have output limits)",
+        len(result),
+        sum(len(m) for m in result.values()),
+        len(output_limits),
+    )
+    return result
+
+
+def _get_model_max_output(provider: str, model: str) -> Optional[int]:
+    """Return the model's max output token limit from the catalog, or ``None``.
+
+    Returns ``None`` when the provider or model is not found in the catalog
+    (e.g. local Ollama models, custom models, or user-typed model names).
+    Callers should fall back to a safe default.
+    """
+    return _model_output_limits.get(provider, {}).get(model)
+
+
+# Short timeout for Ollama's local API
+_OLLAMA_API_TIMEOUT = aiohttp.ClientTimeout(total=8)
+
+
+async def fetch_ollama_models(api_base: str) -> List[str]:
+    """Fetch locally available models from a running Ollama instance.
+
+    Uses Ollama's OpenAI-compatible ``GET {api_base}/models`` endpoint.
+    Returns an empty list if Ollama is not reachable (not running).
+    """
+    url = f"{api_base.rstrip('/')}/models"
+    try:
+        async with aiohttp.ClientSession(timeout=_OLLAMA_API_TIMEOUT) as session:
+            async with session.get(url) as resp:
+                if resp.status != 200:
+                    logger.debug("Ollama API returned HTTP %s from %s", resp.status, api_base)
+                    return []
+                data = await resp.json()
+    except (aiohttp.ClientError, asyncio.TimeoutError, json.JSONDecodeError) as exc:
+        logger.debug("Ollama not reachable at %s: %s", api_base, exc)
+        return []
+
+    raw = data.get("data") if isinstance(data, dict) else None
+    if not isinstance(raw, list):
+        return []
+
+    return [
+        str(entry["id"]) for entry in raw
+        if isinstance(entry, dict) and isinstance(entry.get("id"), str)
+    ]
+
+
+async def get_provider_model_ids(provider_id: str) -> List[str]:
+    """Return the list of known model IDs for *provider_id* from the catalog.
+
+    The catalog is loaded on first call and cached thereafter.  If the
+    provider is not found an empty list is returned (never raises).
+    """
+    catalog = await _load_model_catalog()
+    return catalog.get(provider_id, [])
+
+
+async def get_all_provider_models(
+    provider_ids: List[str],
+) -> Dict[str, List[str]]:
+    """Return model lists for a subset of providers in one call.
+
+    Loads the catalog (cached) and returns only the requested providers.
+    Handy for embedding lightweight data into the template context.
+    """
+    catalog = await _load_model_catalog()
+    return {
+        pid: catalog.get(pid, [])
+        for pid in provider_ids
+    }
+
+
+# Provider preset definitions.
+# Each entry contains display metadata and defaults for the UI.
+# The key is the internal provider id stored in ``llm_provider``.
+# Models are NOT listed here — they come from the opencode model catalog at
+# runtime (see :func:`get_provider_model_ids`).
+PROVIDER_PRESETS: Dict[str, Dict[str, Any]] = {
+    "openai": {
+        "name": "OpenAI",
+        "api_base": "https://api.openai.com/v1",
+        "requires_key": True,
+    },
+    "ollama": {
+        "name": "Ollama (local)",
+        "api_base": "http://localhost:11434/v1",
+        "requires_key": False,
+    },
+    "deepseek": {
+        "name": "DeepSeek",
+        "api_base": "https://api.deepseek.com/v1",
+        "requires_key": True,
+    },
+    "groq": {
+        "name": "Groq",
+        "api_base": "https://api.groq.com/openai/v1",
+        "requires_key": True,
+    },
+    "openrouter": {
+        "name": "OpenRouter",
+        "api_base": "https://openrouter.ai/api/v1",
+        "requires_key": True,
+    },
+    "opencode-go": {
+        "name": "OpenCode Go",
+        "api_base": "https://opencode.ai/zen/go/v1",
+        "requires_key": True,
+    },
+    # "custom" is handled specially (no preset api_base, requires user input)
+}
+
+# Legacy lookup derived from PROVIDER_PRESETS for backward compat.
+_PROVIDER_DEFAULTS: Dict[str, str] = {
+    pid: info["api_base"]
+    for pid, info in PROVIDER_PRESETS.items()
+    if info.get("api_base")
+}
+
+# Request timeout for LLM calls (seconds)
+_LLM_TIMEOUT = aiohttp.ClientTimeout(total=120)
+
+
+class LLMService:
+    """Centralized LLM API client.
+
+    All LLM-based enrichment features call through this service so
+    that BYOK config, retry logic, and error handling live in one place.
+    """
+
+    _instance: Optional["LLMService"] = None
+    _lock: asyncio.Lock = asyncio.Lock()
+
+    def __init__(self, settings_service) -> None:
+        self._settings = settings_service
+
+    # ------------------------------------------------------------------
+    # Singleton access
+    # ------------------------------------------------------------------
+
+    @classmethod
+    async def get_instance(cls) -> "LLMService":
+        """Return the lazily-initialised global ``LLMService`` instance."""
+
+        if cls._instance is None:
+            async with cls._lock:
+                if cls._instance is None:
+                    from .settings_manager import get_settings_manager
+
+                    cls._instance = cls(get_settings_manager())
+                    # Start preloading the model catalog in the background so
+                    # the settings UI never blocks on it.  The catalog is
+                    # cached after the first fetch (see _load_model_catalog).
+                    asyncio.create_task(_load_model_catalog())
+        return cls._instance
+
+    @classmethod
+    def reset_instance(cls) -> None:
+        """Reset the cached singleton — primarily for tests."""
+
+        cls._instance = None
+
+    # ------------------------------------------------------------------
+    # Configuration helpers
+    # ------------------------------------------------------------------
+
+    def _get_config(self) -> Dict[str, Any]:
+        """Read the current LLM configuration from settings."""
+
+        return {
+            "provider": self._settings.get("llm_provider", "openai"),
+            "api_key": self._settings.get("llm_api_key", ""),
+            "api_base": self._settings.get("llm_api_base", ""),
+            "model": self._settings.get("llm_model", ""),
+        }
+
+    @staticmethod
+    def _provider_requires_key(provider: str) -> bool:
+        """Return ``False`` when the given provider id does not need an API key."""
+        preset = PROVIDER_PRESETS.get(provider, {})
+        return bool(preset.get("requires_key", True))
+
+    def is_configured(self) -> bool:
+        """Return ``True`` when the LLM provider is minimally configured.
+
+        A provider is considered configured when ``llm_model`` is set,
+        an API key is configured for providers that require one (e.g.
+        Ollama does not), and an API base URL is set for providers that
+        have no preset default (e.g. ``custom``).
+        """
+
+        cfg = self._get_config()
+        has_model = bool(cfg["model"])
+        has_key = bool(cfg["api_key"]) or not self._provider_requires_key(cfg["provider"])
+        has_base = bool(cfg["api_base"]) or bool(_PROVIDER_DEFAULTS.get(cfg["provider"]))
+        return has_model and has_key and has_base
+
+    def _resolve_api_base(self, provider: str, api_base: str) -> str:
+        """Resolve the API base URL for the given provider.
+
+        If ``api_base`` is explicitly set (non-empty), it takes priority.
+        Otherwise the default from :data:`PROVIDER_PRESETS` is used.
+        """
+
+        if api_base:
+            return api_base.rstrip("/")
+        return _PROVIDER_DEFAULTS.get(provider, "").rstrip("/")
+
+    def _build_headers(self, api_key: str) -> Dict[str, str]:
+        """Build HTTP headers for the LLM API request."""
+
+        headers = {"Content-Type": "application/json"}
+        if api_key:
+            headers["Authorization"] = f"Bearer {api_key}"
+        return headers
+
+    def _ensure_configured(self) -> Dict[str, Any]:
+        """Validate configuration and return it, or raise.
+
+        A provider is considered configured when ``llm_model`` is set,
+        an API key is configured for providers that require one, and
+        an API base URL is set for providers without a preset default.
+        """
+
+        cfg = self._get_config()
+        has_model = bool(cfg["model"])
+        needs_key = self._provider_requires_key(cfg["provider"])
+        has_key = bool(cfg["api_key"]) or not needs_key
+        has_base = bool(cfg["api_base"]) or bool(_PROVIDER_DEFAULTS.get(cfg["provider"]))
+        if not (has_model and has_key and has_base):
+            parts = []
+            if not has_model:
+                parts.append("No LLM model specified")
+            if not has_key and needs_key:
+                parts.append("No LLM API key configured")
+            if not has_base:
+                parts.append(
+                    f"No API base URL for provider '{cfg['provider']}'"
+                )
+            detail = "; ".join(parts) if parts else "LLM provider is not configured"
+            raise LLMNotConfiguredError(
+                f"{detail}. Configure it in Settings → AI Provider."
+            )
+        return cfg
+
+    # ------------------------------------------------------------------
+    # Core API call
+    # ------------------------------------------------------------------
+
+    async def chat_completion(
+        self,
+        *,
+        messages: List[Dict[str, str]],
+        model: Optional[str] = None,
+        temperature: float = 0.3,
+        response_format: Optional[Dict[str, Any]] = None,
+        max_tokens: Optional[int] = None,
+        retry_on_rate_limit: bool = True,
+    ) -> Dict[str, Any]:
+        """Call the configured LLM provider's ``/chat/completions`` endpoint.
+
+        Args:
+            messages: OpenAI-format message list
+            model: Override the configured model name
+            temperature: Sampling temperature
+            response_format: Optional ``{"type": "json_object"}`` for structured output
+            max_tokens: Optional max output tokens
+            retry_on_rate_limit: Retry once after a 429 with backoff
+
+        Returns:
+            Dict with ``content`` (str), ``usage`` (dict), ``model`` (str)
+
+        Raises:
+            LLMNotConfiguredError: Provider not enabled / missing config
+            LLMRateLimitError: Rate limited and retry exhausted
+            LLMResponseError: Non-200 response or parse failure
+        """
+
+        cfg = self._ensure_configured()
+        api_base = self._resolve_api_base(cfg["provider"], cfg["api_base"])
+        model_name = model or cfg["model"]
+
+        is_ollama = cfg["provider"] == "ollama"
+
+        if is_ollama:
+            # Use Ollama's native /api/chat endpoint which does NOT expose
+            # a separate reasoning/thinking field (the model's full output
+            # lands directly in message.content).  The OpenAI-compatible
+            # endpoint splits thinking into the "reasoning" field, making
+            # content empty when thinking consumes all available tokens.
+            base = api_base.rstrip("/")
+            if base.endswith("/v1"):
+                base = base[:-3]
+            url = f"{base}/api/chat"
+        else:
+            url = f"{api_base}/chat/completions"
+
+        payload: Dict[str, Any]
+        if is_ollama:
+            payload = {
+                "model": model_name,
+                "messages": messages,
+                "stream": False,
+                # Suppress separate thinking trace — thinking still happens
+                # internally (accuracy preserved) but output goes directly to
+                # message.content instead of being split across content +
+                # thinking.  Without this the model can exhaust num_predict
+                # on thinking alone and leave content empty.
+                "think": False,
+                "options": {
+                    "temperature": temperature,
+                    # 8K context is sufficient for metadata enrichment
+                    # (prompt ~2-5K, output ~0.2-1K tokens).  The old 32K
+                    # value was excessive for this use case and increased
+                    # Ollama VRAM usage unnecessarily.
+                    "num_ctx": 8192,
+                },
+            }
+            if response_format is not None:
+                payload["format"] = "json"
+            if max_tokens is not None:
+                payload["options"]["num_predict"] = max_tokens
+        else:
+            payload = {
+                "model": model_name,
+                "messages": messages,
+                "temperature": temperature,
+            }
+            if response_format is not None:
+                payload["response_format"] = response_format
+            if max_tokens is not None:
+                payload["max_tokens"] = max_tokens
+
+        if is_ollama:
+            logger.info(
+                "Ollama request: model=%s num_ctx=%s num_predict=%s format=%s think=%s",
+                payload.get("model"),
+                payload.get("options", {}).get("num_ctx"),
+                payload.get("options", {}).get("num_predict"),
+                payload.get("format", "none"),
+                payload.get("think"),
+            )
+
+        headers = self._build_headers(cfg["api_key"])
+
+        attempt = 0
+        max_attempts = 2 if retry_on_rate_limit else 1
+        while attempt < max_attempts:
+            attempt += 1
+            try:
+                async with aiohttp.ClientSession(timeout=_LLM_TIMEOUT) as session:
+                    async with session.post(
+                        url, json=payload, headers=headers
+                    ) as resp:
+                        if resp.status == 429:
+                            if attempt < max_attempts:
+                                retry_after = float(
+                                    resp.headers.get("Retry-After", "5")
+                                )
+                                logger.warning(
+                                    "LLM rate limited, retrying after %.1fs",
+                                    retry_after,
+                                )
+                                await asyncio.sleep(retry_after)
+                                continue
+                            raise LLMRateLimitError(
+                                f"LLM provider rate limited (HTTP 429)",
+                                provider=cfg["provider"],
+                            )
+
+                        if resp.status != 200:
+                            body = await resp.text()
+                            raise LLMResponseError(
+                                f"LLM API returned HTTP {resp.status}: "
+                                f"{body[:500]}"
+                            )
+
+                        data = await resp.json()
+
+            except aiohttp.ClientError as exc:
+                raise LLMResponseError(f"Network error calling LLM API: {exc}") from exc
+
+            # Parse response
+            try:
+                if is_ollama:
+                    content = (data.get("message") or {}).get("content") or ""
+                    usage = {"completion_tokens": data.get("eval_count", 0)}
+                    finish_reason = data.get("done_reason", "")
+                    if not content:
+                        logger.warning(
+                            "LLM returned empty content. Provider=ollama, "
+                            "done_reason=%s, eval_count=%s",
+                            finish_reason,
+                            data.get("eval_count", 0),
+                        )
+                else:
+                    content = data["choices"][0]["message"].get("content") or ""
+                    usage = data.get("usage", {})
+                    if not content:
+                        logger.warning(
+                            "LLM returned empty content. Full response truncated: %s",
+                            json.dumps(data, ensure_ascii=False)[:1000],
+                        )
+                return {
+                    "content": content,
+                    "usage": usage,
+                    "model": data.get("model", model_name),
+                }
+            except (KeyError, IndexError) as exc:
+                raise LLMResponseError(
+                    f"Unexpected LLM response structure: {json.dumps(data)[:500]}"
+                ) from exc
+
+        # Should not reach here, but satisfy type checker
+        raise LLMRateLimitError("Rate limit retry exhausted", provider=cfg["provider"])
+
+    # ------------------------------------------------------------------
+    # Structured output convenience
+    # ------------------------------------------------------------------
+
+    async def chat_completion_json(
+        self,
+        *,
+        system_prompt: str,
+        user_prompt: str,
+        model: Optional[str] = None,
+        temperature: float = 0.3,
+        max_tokens: Optional[int] = None,
+    ) -> Dict[str, Any]:
+        """Call the LLM with ``response_format=json_object`` and return parsed JSON.
+
+        ``max_tokens`` is resolved in this order:
+        1. Explicit caller-supplied ``max_tokens``
+        2. Per-model ``limit.output`` from the model catalog
+        3. A safe default of 4096 (sufficient for metadata enrichment)
+
+        If the response content is empty or not valid JSON, attempts
+        :func:`_try_salvage_json` before raising.
+
+        Args:
+            system_prompt: System-level instructions
+            user_prompt: User-level query
+            model: Override the configured model name
+            temperature: Sampling temperature
+            max_tokens: Optional max output tokens
+
+        Returns:
+            Parsed JSON dict from the LLM response
+
+        Raises:
+            LLMNotConfiguredError: Provider not configured
+            LLMRateLimitError: Rate limited
+            LLMResponseError: Empty response or JSON parse failure
+        """
+
+        messages = [
+            {"role": "system", "content": system_prompt},
+            {"role": "user", "content": user_prompt},
+        ]
+
+        # Resolve max_tokens: caller override → catalog lookup → safe default
+        if max_tokens is None:
+            cfg = self._get_config()
+            effective_max = _get_model_max_output(cfg["provider"], cfg["model"])
+        else:
+            effective_max = max_tokens
+        if effective_max is None:
+            effective_max = 4096
+
+        # Use json_schema (not json_object) for broader provider compatibility:
+        # LM Studio and some other OpenAI-compatible servers reject
+        # json_object but accept json_schema.  {"type": "object"} is
+        # functionally equivalent — it accepts any JSON object without
+        # constraining specific fields.
+        response_format = {
+            "type": "json_schema",
+            "json_schema": {
+                "name": "metadata",
+                "schema": {"type": "object"},
+            },
+        }
+
+        try:
+            result = await self.chat_completion(
+                messages=messages,
+                model=model,
+                temperature=temperature,
+                response_format=response_format,
+                max_tokens=effective_max,
+            )
+        except LLMResponseError as e:
+            # Only fall back when the provider rejects the response_format
+            # type value (e.g. "'response_format.type' must be...").  Avoid
+            # catching unrelated 400 errors whose body happens to mention
+            # "response_format" (e.g. "model does not support
+            # response_format restrictions on this endpoint").
+            if "'response_format.type'" not in str(e).lower():
+                raise
+            logger.info(
+                "Provider rejected response_format, retrying without it. "
+                "Falling back to prompt-only JSON mode. Error: %s",
+                e,
+            )
+            result = await self.chat_completion(
+                messages=messages,
+                model=model,
+                temperature=temperature,
+                response_format=None,
+                max_tokens=effective_max,
+            )
+
+        content = result.get("content", "") or ""
+        if not content:
+            raise LLMResponseError(
+                "LLM returned empty content. "
+                f"Raw response: {json.dumps(result)[:500]}"
+            )
+
+        try:
+            parsed = json.loads(content)
+            logger.debug(
+                "LLM raw content: %s",
+                json.dumps(parsed, ensure_ascii=False)[:2000],
+            )
+            return parsed
+        except (json.JSONDecodeError, TypeError) as exc:
+            logger.info(
+                "LLM raw response (first 800 chars): %s",
+                content[:800],
+            )
+
+        # Last resort: attempt to salvage partial/truncated JSON
+        salvaged = _try_salvage_json(content)
+        if salvaged is not None:
+            logger.warning(
+                "LLM JSON salvaged from partial content (%d chars raw)",
+                len(content),
+            )
+            return salvaged
+
+        raise LLMResponseError(
+            f"LLM response could not be parsed as JSON: {content[:200]}"
+        )
+
+
+def _try_salvage_json(raw: str) -> Dict[str, Any] | None:
+    """Attempt to repair and parse a truncated JSON string.
+
+    Handles common truncation patterns:
+
+    * Incomplete string value at the end (``"foo`` → ``"foo"``)
+    * Missing closing ``}`` or ``]`` (respecting nesting order)
+    * Trailing comma before closing bracket
+    * Extra text after the JSON object (e.g. markdown fences)
+
+    Returns the parsed dict on success, ``None`` if repair is impossible.
+    """
+    if not raw:
+        return None
+
+    text = raw.strip()
+
+    # Strip markdown fences if the LLM wrapped the JSON
+    if text.startswith("```"):
+        end = text.find("\n")
+        text = text[end + 1:] if end != -1 else text[3:]
+    if text.endswith("```"):
+        text = text[:-3].rstrip()
+
+    # Find the first '{' and strip everything before it
+    start = text.find("{")
+    if start == -1:
+        return None
+    text = text[start:]
+
+    # Try to close an incomplete string at the end (e.g. ``"https://huggingf``)
+    # Pattern: ends mid-string (last quote is open)
+    if text.count('"') % 2 == 1:
+        text += '"'
+
+    # Ensure trailing commas before closing braces work
+    text = _strip_trailing_commas(text)
+
+    # Walk through the text character by character to find unclosed
+    # brackets and close them in the correct (LIFO) order.
+    # We ignore brackets inside quoted strings.
+    stack: list[str] = []
+    in_string = False
+    escape = False
+    for ch in text:
+        if escape:
+            escape = False
+            continue
+        if ch == "\\":
+            escape = True
+            continue
+        if ch == '"':
+            in_string = not in_string
+            continue
+        if in_string:
+            continue
+        if ch in ("{", "["):
+            stack.append(ch)
+        elif ch == "}":
+            if stack and stack[-1] == "{":
+                stack.pop()
+            else:
+                return None  # Unmatched closer — unrecoverable
+        elif ch == "]":
+            if stack and stack[-1] == "[":
+                stack.pop()
+            else:
+                return None
+
+    # Close remaining open brackets in reverse order
+    for opener in reversed(stack):
+        text += "}" if opener == "{" else "]"
+
+    try:
+        return json.loads(text)
+    except (json.JSONDecodeError, ValueError):
+        return None
+
+
+def _strip_trailing_commas(text: str) -> str:
+    """Remove commas that appear before a closing brace/bracket."""
+    import re as _re
+    text = _re.sub(r",\s*}", "}", text)
+    text = _re.sub(r",\s*]", "]", text)
+    return text

py/services/lora_service.py

@@ -24,23 +24,41 @@ class LoraService(BaseModelService):
         """
         super().__init__("lora", scanner, LoraMetadata, update_service=update_service)
 
-    async def format_response(self, lora_data: Dict) -> Dict:
-        """Format LoRA data for API response"""
+    async def format_response(self, lora_data: Dict) -> Optional[Dict]:
+        """Format LoRA data for API response.
+
+        Returns None when the entry is missing critical fields (corrupted cache
+        row), so the handler layer can filter it out instead of crashing the
+        whole listing request. See issue #730.
+        """
+        # Guard against corrupted cache entries missing critical fields
+        file_path = lora_data.get("file_path")
+        if not file_path or not isinstance(file_path, str):
+            logger.warning(
+                "Skipping corrupted LoRA entry (missing file_path): %s",
+                lora_data.get("file_name", "<unknown>"),
+            )
+            return None
+
         # Resolve sub_type using priority: sub_type > model_type > civitai.model.type > default
         # Normalize to lowercase for consistent API responses
         sub_type = resolve_sub_type(lora_data).lower()
 
+        file_name = lora_data.get("file_name") or ""
+        model_name = lora_data.get("model_name") or file_name
+        folder = lora_data.get("folder") or ""
+
         return {
-            "model_name": lora_data["model_name"],
-            "file_name": lora_data["file_name"],
+            "model_name": model_name,
+            "file_name": file_name,
             "preview_url": config.get_preview_static_url(
                 lora_data.get("preview_url", "")
             ),
             "preview_nsfw_level": lora_data.get("preview_nsfw_level", 0),
             "base_model": lora_data.get("base_model", ""),
-            "folder": lora_data["folder"],
+            "folder": folder,
             "sha256": lora_data.get("sha256", ""),
-            "file_path": lora_data["file_path"].replace(os.sep, "/"),
+            "file_path": file_path.replace(os.sep, "/"),
             "file_size": lora_data.get("size", 0),
             "modified": lora_data.get("modified", ""),
             "tags": lora_data.get("tags", []),
@@ -59,6 +77,8 @@ class LoraService(BaseModelService):
                 lora_data.get("civitai", {}), minimal=True
             ),
             "auto_tags": lora_data.get("auto_tags") or extract_auto_tags(lora_data),
+            "version_count": lora_data.get("version_count"),
+            "hf_url": lora_data.get("hf_url", ""),
         }
 
     async def _apply_specific_filters(self, data: List[Dict], **kwargs) -> List[Dict]:
@@ -251,12 +271,16 @@ class LoraService(BaseModelService):
         return letters
 
     async def get_lora_trigger_words(self, lora_name: str) -> List[str]:
-        """Get trigger words for a specific LoRA file"""
+        """Get trigger words for a specific LoRA file.
+
+        Supports both simple names and full-path syntax.
+        """
         cache = await self.scanner.get_cached_data()
 
         for lora in cache.raw_data:
-            if lora["file_name"] == lora_name:
-                civitai_data = lora.get("civitai", {})
+            file_name = lora.get("file_name", "")
+            if file_name == lora_name or lora_name.endswith("/" + file_name) or lora_name.endswith("\\" + file_name):
+                civitai_data = lora.get("civitai") or {}
                 return civitai_data.get("trainedWords", [])
 
         return []

py/services/metadata_service.py

@@ -15,6 +15,17 @@ from .service_registry import ServiceRegistry
 
 logger = logging.getLogger(__name__)
 
+_PROVIDER_DISPLAY_NAMES = {
+    "civitai_api": "CivitAI",
+    "civarchive_api": "CivArchive",
+    "sqlite": "Archive DB",
+}
+
+_PRESET_PROVIDER_ORDERS = {
+    "civitai_archive_sqlite": ["civitai_api", "civarchive_api", "sqlite"],
+    "civitai_sqlite_archive": ["civitai_api", "sqlite", "civarchive_api"],
+}
+
 async def initialize_metadata_providers():
     """Initialize and configure all metadata providers based on settings"""
     provider_manager = await ModelMetadataProviderManager.get_instance()
@@ -26,7 +37,9 @@ async def initialize_metadata_providers():
     # Get settings
     settings_manager = get_settings_manager()
     enable_archive_db = settings_manager.get('enable_metadata_archive_db', False)
-    
+    enable_civarchive_api = settings_manager.get('enable_civarchive_api', True)
+    provider_order = settings_manager.get('metadata_provider_order', 'civitai_archive_sqlite')
+
     providers = []
     
     # Initialize archive database provider if enabled
@@ -59,27 +72,48 @@ async def initialize_metadata_providers():
     except Exception as e:
         logger.error(f"Failed to initialize Civitai API metadata provider: {e}")
 
-    # Register CivArchive provider, and all add to fallback providers
-    try:
-        civarchive_client = await ServiceRegistry.get_civarchive_client()
-        civarchive_provider = CivArchiveModelMetadataProvider(civarchive_client)
-        provider_manager.register_provider('civarchive_api', civarchive_provider)
-        providers.append(('civarchive_api', civarchive_provider))
-        logger.debug("CivArchive metadata provider registered (also included in fallback)")
-    except Exception as e:
-        logger.error(f"Failed to initialize CivArchive metadata provider: {e}")
+    # Register CivArchive provider when enabled. Civitai API is always
+    # preferred (better metadata); CivArchive mainly recovers metadata for
+    # models deleted from Civitai, so it can be turned off to avoid its long
+    # rate-limit windows entirely.
+    if enable_civarchive_api:
+        try:
+            civarchive_client = await ServiceRegistry.get_civarchive_client()
+            civarchive_provider = CivArchiveModelMetadataProvider(civarchive_client)
+            provider_manager.register_provider('civarchive_api', civarchive_provider)
+            providers.append(('civarchive_api', civarchive_provider))
+            logger.debug("CivArchive metadata provider registered (also included in fallback)")
+        except Exception as e:
+            logger.error(f"Failed to initialize CivArchive metadata provider: {e}")
+    else:
+        logger.debug("CivArchive metadata provider disabled by setting 'enable_civarchive_api'")
+
+    # Preset fallback orderings (see module-level _PRESET_PROVIDER_ORDERS).
+    # civitai_api is always first (better metadata); the remaining providers
+    # are arranged by the configured preset.  Providers that are not
+    # registered (disabled/unavailable) are simply skipped, so each preset
+    # degrades gracefully.
+    desired_order = _PRESET_PROVIDER_ORDERS.get(
+        provider_order, _PRESET_PROVIDER_ORDERS["civitai_archive_sqlite"]
+    )
 
     # Set up fallback provider based on available providers
     if len(providers) > 1:
-        # Always use Civitai API (it has better metadata), then CivArchive API, then Archive DB
         ordered_providers: list[tuple[str, ModelMetadataProvider]] = []
-        ordered_providers.extend([p for p in providers if p[0] == 'civitai_api'])
-        ordered_providers.extend([p for p in providers if p[0] == 'civarchive_api'])
-        ordered_providers.extend([p for p in providers if p[0] == 'sqlite'])
-        
+        for name in desired_order:
+            ordered_providers.extend([p for p in providers if p[0] == name])
+        # Include any provider not covered by the preset (defensive) at the end
+        for p in providers:
+            if p not in ordered_providers:
+                ordered_providers.append(p)
+
         if ordered_providers:
             fallback_provider = FallbackMetadataProvider(ordered_providers)
             provider_manager.register_provider('fallback', fallback_provider, is_default=True)
+            logger.debug(
+                "Metadata fallback provider order: %s",
+                ", ".join(name for name, _ in ordered_providers),
+            )
     elif len(providers) == 1:
         # Only one provider available, set it as default
         provider_name, provider = providers[0]
@@ -96,11 +130,30 @@ async def update_metadata_providers():
         # Get current settings
         settings_manager = get_settings_manager()
         enable_archive_db = settings_manager.get('enable_metadata_archive_db', False)
+        enable_civarchive_api = settings_manager.get('enable_civarchive_api', True)
+        provider_order = settings_manager.get('metadata_provider_order', 'civitai_archive_sqlite')
         
         # Reinitialize all providers with new settings
         provider_manager = await initialize_metadata_providers()
         
-        logger.info(f"Updated metadata providers, archive_db enabled: {enable_archive_db}")
+        # Build effective provider chain for logging (use actually-registered
+        # providers, not just settings, so a failed init is reflected correctly)
+        registered = set(provider_manager.providers.keys())
+        desired = _PRESET_PROVIDER_ORDERS.get(
+            provider_order, _PRESET_PROVIDER_ORDERS["civitai_archive_sqlite"]
+        )
+        chain = " → ".join(
+            _PROVIDER_DISPLAY_NAMES[p]
+            for p in desired
+            if p in registered and p in _PROVIDER_DISPLAY_NAMES
+        )
+        
+        logger.info(
+            "Updated metadata providers: archive_db=%s, civarchive_api=%s, chain=%s",
+            enable_archive_db,
+            enable_civarchive_api,
+            chain,
+        )
         return provider_manager
     except Exception as e:
         logger.error(f"Failed to update metadata providers: {e}")

py/services/metadata_sync_service.py

@@ -209,20 +209,40 @@ class MetadataSyncService:
                         error_msg = "CivitAI model is deleted and no archive provider is available"
                     return False, error_msg
             else:
-                provider_attempts.append((None, await self._get_default_provider()))
+                is_hf_source = bool(model_data.get("hf_url"))
+                if is_hf_source:
+                    # HF-sourced model: only check CivitAI API directly.
+                    # CivArchive is almost guaranteed to have no record, and
+                    # hitting it wastes rate-limit budget.
+                    # Use a distinct provider name ("civitai_api" not None) so
+                    # downstream code does NOT interpret a "Model not found"
+                    # response as civitai_api_not_found — which would mark the
+                    # model civitai_deleted=True when it was never on CivitAI.
+                    try:
+                        provider_attempts.append(("civitai_api", await self._get_provider("civitai_api")))
+                    except Exception as exc:  # pragma: no cover - provider resolution fault
+                        logger.debug("Unable to resolve civitai_api provider: %s", exc)
+                if not provider_attempts:
+                    provider_attempts.append((None, await self._get_default_provider()))
 
             civitai_metadata: Optional[Dict[str, Any]] = None
             metadata_provider: Optional[MetadataProviderProtocol] = None
             provider_used: Optional[str] = None
             last_error: Optional[str] = None
             civitai_api_not_found = False
+            any_rate_limited = False
 
             for provider_name, provider in provider_attempts:
                 try:
                     civitai_metadata_candidate, error = await provider.get_model_by_hash(sha256)
                 except RateLimitError as exc:
-                    exc.provider = exc.provider or (provider_name or provider.__class__.__name__)
-                    raise
+                    logger.warning(
+                        "Provider %s is rate-limited (retry_after=%.0fs); skipping to next provider",
+                        provider_name or provider.__class__.__name__,
+                        exc.retry_after or 0,
+                    )
+                    any_rate_limited = True
+                    continue
                 except Exception as exc:  # pragma: no cover - defensive logging
                     logger.error("Provider %s failed for hash %s: %s", provider_name, sha256, exc)
                     civitai_metadata_candidate, error = None, str(exc)
@@ -258,6 +278,14 @@ class MetadataSyncService:
                     model_data["last_checked_at"] = datetime.now().timestamp()
                     needs_save = True
 
+                # When the model was already classified as "not on CivitAI" via
+                # .metadata.json (civitai_deleted=True) but the SQLite cache is
+                # stale (because the pre-fix code never persisted these flags),
+                # ensure the flags are written to the scanner cache + SQLite.
+                if not needs_save and model_data.get("civitai_deleted") is True:
+                    model_data["last_checked_at"] = datetime.now().timestamp()
+                    needs_save = True
+
                 # Save metadata if any state was updated
                 if needs_save:
                     data_to_save = model_data.copy()
@@ -266,6 +294,7 @@ class MetadataSyncService:
                     if "last_checked_at" not in data_to_save:
                         data_to_save["last_checked_at"] = datetime.now().timestamp()
                     await self._metadata_manager.save_metadata(file_path, data_to_save)
+                    await update_cache_func(file_path, file_path, data_to_save)
 
                 default_error = (
                     "CivitAI model is deleted and metadata archive DB is not enabled"
@@ -276,17 +305,18 @@ class MetadataSyncService:
                 )
 
                 resolved_error = last_error or default_error
+                if any_rate_limited and "Rate limited" not in resolved_error:
+                    resolved_error = "Rate limited"
                 if is_expected_offline_error(resolved_error):
                     resolved_error = OFFLINE_FRIENDLY_MESSAGE
 
                 error_msg = (
                     f"Error fetching metadata: {resolved_error} "
-                    f"(model_name={model_data.get('model_name', '')})"
+                    f"(file={os.path.basename(file_path)}, sha256={sha256})"
                 )
-                if is_expected_offline_error(resolved_error):
-                    logger.info(error_msg)
-                else:
-                    logger.error(error_msg)
+                # Use case layer (BulkMetadataRefreshUseCase) logs failed models at WARNING level,
+                # so this level is demoted to DEBUG to avoid duplicate user-visible logging.
+                logger.debug(error_msg)
                 return False, error_msg
 
             model_data["from_civitai"] = True
@@ -411,7 +441,18 @@ class MetadataSyncService:
         metadata = await metadata_loader(metadata_path)
 
         for key, value in updates.items():
-            if isinstance(value, dict) and isinstance(metadata.get(key), dict):
+            if key == "tags" and isinstance(value, list):
+                # Normalize tags: trim, lowercase, deduplicate
+                normalized = []
+                seen = set()
+                for tag in value:
+                    if isinstance(tag, str):
+                        t = tag.strip().lower()
+                        if t and t not in seen:
+                            normalized.append(t)
+                            seen.add(t)
+                metadata[key] = normalized
+            elif isinstance(value, dict) and isinstance(metadata.get(key), dict):
                 metadata[key].update(value)
             else:
                 metadata[key] = value

py/services/model_cache.py

@@ -18,6 +18,8 @@ SUPPORTED_SORT_MODES = [
     ('size', 'desc'),
     ('usage', 'asc'),
     ('usage', 'desc'),
+    ('versions_count', 'asc'),
+    ('versions_count', 'desc'),
 ]
 # Is this in use?
 
@@ -263,6 +265,17 @@ class ModelCache:
                 ),
                 reverse=reverse
             )
+        elif sort_key == 'versions_count':
+            # Pre-dedup sort: fall back to name sort.
+            # Actual re-sort by version_count happens in get_paginated_data after dedup.
+            result = natsorted(
+                data,
+                key=lambda x: (
+                    self._get_display_name(x).lower(),
+                    x.get('file_path', '').lower()
+                ),
+                reverse=reverse
+            )
         else:
             # Fallback: no sort
             result = list(data)
@@ -324,4 +337,25 @@ class ModelCache:
             else:
                 return False  # Model not found
                     
-            return True
+            return True
+
+    async def clear_preview_by_path(self, preview_file_path: str) -> int:
+        """Clear ``preview_url`` for every cached entry referencing a file path.
+
+        When a preview file has been deleted from disk, this removes its
+        reference from all matching cache entries so the next list-API
+        response returns an empty ``preview_url`` instead of a stale URL
+        that produces 404s.
+
+        Returns the number of entries that were updated.
+        """
+        normalized = preview_file_path.replace("\\", "/")
+        cleared = 0
+        async with self._lock:
+            for item in self.raw_data:
+                cached_url = item.get("preview_url", "")
+                if cached_url.replace("\\", "/") == normalized:
+                    item["preview_url"] = ""
+                    item["preview_nsfw_level"] = 0
+                    cleared += 1
+        return cleared

py/services/model_file_service.py

@@ -8,6 +8,7 @@ from abc import ABC, abstractmethod
 from ..utils.utils import calculate_relative_path_for_model, remove_empty_dirs
 from ..utils.constants import AUTO_ORGANIZE_BATCH_SIZE
 from ..services.settings_manager import get_settings_manager
+from ..services.model_lifecycle_service import _require_path_in_library_roots
 
 logger = logging.getLogger(__name__)
 
@@ -493,6 +494,9 @@ class ModelMoveService:
             Dictionary with move result
         """
         try:
+            _require_path_in_library_roots(file_path, self.scanner, label="Source path")
+            _require_path_in_library_roots(target_path, self.scanner, label="Target path")
+
             if use_default_paths:
                 # Find the model in cache to get metadata
                 cache = await self.scanner.get_cached_data()

py/services/model_lifecycle_service.py

@@ -48,6 +48,36 @@ async def delete_model_artifacts(
     return deleted
 
 
+def _require_path_in_library_roots(file_path: str, scanner, *, label: str = "path") -> None:
+    """Raise ``ValueError`` if *file_path* is not inside a configured model root.
+
+    Uses ``os.path.abspath()`` (NOT ``realpath``) to resolve ``..`` and ``.``
+    while preserving symlinks — this keeps the check in business-path space.
+    Skips when the scanner does not expose ``get_model_roots`` or the list
+    is empty.
+    """
+
+    roots = None
+    if hasattr(scanner, "get_model_roots"):
+        try:
+            roots = scanner.get_model_roots()
+        except NotImplementedError:
+            roots = None
+    if not roots:
+        return
+
+    resolved = os.path.abspath(os.path.normpath(file_path))
+
+    for root in roots:
+        root_resolved = os.path.abspath(os.path.normpath(root))
+        if resolved == root_resolved or resolved.startswith(root_resolved + os.sep):
+            return
+
+    raise ValueError(
+        f"{label} '{file_path}' is outside configured library directories"
+    )
+
+
 class ModelLifecycleService:
     """Co-ordinate destructive and mutating model operations."""
 
@@ -74,6 +104,8 @@ class ModelLifecycleService:
         if not file_path:
             raise ValueError("Model path is required")
 
+        _require_path_in_library_roots(file_path, self._scanner, label="File path")
+
         cache = await self._scanner.get_cached_data()
 
         cached_entry = None
@@ -182,6 +214,8 @@ class ModelLifecycleService:
         if not file_path:
             raise ValueError("Model path is required")
 
+        _require_path_in_library_roots(file_path, self._scanner, label="File path")
+
         metadata_path = os.path.splitext(file_path)[0] + ".metadata.json"
         metadata = await self._metadata_loader(metadata_path)
         metadata["exclude"] = True
@@ -229,6 +263,8 @@ class ModelLifecycleService:
         if not file_path:
             raise ValueError("Model path is required")
 
+        _require_path_in_library_roots(file_path, self._scanner, label="File path")
+
         if not os.path.exists(file_path):
             raise ValueError("Model file does not exist")
 
@@ -270,6 +306,9 @@ class ModelLifecycleService:
         if not file_paths:
             raise ValueError("No file paths provided for deletion")
 
+        for path in file_paths:
+            _require_path_in_library_roots(path, self._scanner, label="File path")
+
         return await self._scanner.bulk_delete_models(file_paths)
 
     async def rename_model(
@@ -280,6 +319,8 @@ class ModelLifecycleService:
         if not file_path or not new_file_name:
             raise ValueError("File path and new file name are required")
 
+        _require_path_in_library_roots(file_path, self._scanner, label="File path")
+
         invalid_chars = {"/", "\\", ":", "*", "?", '"', "<", ">", "|"}
         if any(char in new_file_name for char in invalid_chars):
             raise ValueError("Invalid characters in file name")

py/services/model_metadata_provider.py

@@ -65,7 +65,14 @@ class _RateLimitRetryHelper:
                 return await func(*args, **kwargs)
             except RateLimitError as exc:
                 attempt += 1
-                if attempt >= self._retry_limit:
+
+                # Determine effective retry limit based on rate-limit magnitude
+                effective_retry_limit = self._retry_limit  # default: 3
+                if exc.retry_after is not None and exc.retry_after >= 120.0:
+                    # Long rate-limit window (>=2 min) — retries are futile
+                    effective_retry_limit = 1  # total 1 attempt = 0 retries
+
+                if attempt >= effective_retry_limit:
                     exc.provider = exc.provider or label
                     raise
 
@@ -478,8 +485,12 @@ class FallbackMetadataProvider(ModelMetadataProvider):
                 if result:
                     return result, error
             except RateLimitError as exc:
-                exc.provider = exc.provider or label
-                raise exc
+                logger.warning(
+                    "Provider %s is rate-limited (retry_after=%.0fs); skipping to next provider",
+                    label,
+                    exc.retry_after or 0,
+                )
+                continue
             except Exception as e:
                 logger.debug("Provider %s failed for get_model_by_hash: %s", label, e)
                 continue
@@ -497,16 +508,12 @@ class FallbackMetadataProvider(ModelMetadataProvider):
                 if result:
                     return result
             except RateLimitError as exc:
-                if not_found_confirmed:
-                    logger.debug(
-                        "Suppressing rate limit from %s for model %s: "
-                        "already confirmed as not found by another provider",
-                        label,
-                        model_id,
-                    )
-                    return None
-                exc.provider = exc.provider or label
-                raise exc
+                logger.warning(
+                    "Provider %s is rate-limited (retry_after=%.0fs); skipping to next provider",
+                    label,
+                    exc.retry_after or 0,
+                )
+                continue
             except ResourceNotFoundError:
                 not_found_confirmed = True
                 logger.debug(
@@ -532,8 +539,12 @@ class FallbackMetadataProvider(ModelMetadataProvider):
                 if result:
                     return result
             except RateLimitError as exc:
-                exc.provider = exc.provider or label
-                raise exc
+                logger.warning(
+                    "Provider %s is rate-limited (retry_after=%.0fs); skipping to next provider",
+                    label,
+                    exc.retry_after or 0,
+                )
+                continue
             except Exception as e:
                 logger.debug("Provider %s failed for get_model_version: %s", label, e)
                 continue
@@ -550,8 +561,12 @@ class FallbackMetadataProvider(ModelMetadataProvider):
                 if result:
                     return result, error
             except RateLimitError as exc:
-                exc.provider = exc.provider or label
-                raise exc
+                logger.warning(
+                    "Provider %s is rate-limited (retry_after=%.0fs); skipping to next provider",
+                    label,
+                    exc.retry_after or 0,
+                )
+                continue
             except Exception as e:
                 logger.debug("Provider %s failed for get_model_version_info: %s", label, e)
                 continue
@@ -572,8 +587,12 @@ class FallbackMetadataProvider(ModelMetadataProvider):
             except NotImplementedError:
                 continue
             except RateLimitError as exc:
-                exc.provider = exc.provider or label
-                raise exc
+                logger.warning(
+                    "Provider %s is rate-limited (retry_after=%.0fs); skipping to next provider",
+                    label,
+                    exc.retry_after or 0,
+                )
+                continue
             except Exception as e:
                 logger.debug(
                     "Provider %s failed for get_model_versions_by_hashes: %s",
@@ -594,8 +613,12 @@ class FallbackMetadataProvider(ModelMetadataProvider):
                 if result is not None:
                     return result
             except RateLimitError as exc:
-                exc.provider = exc.provider or label
-                raise exc
+                logger.warning(
+                    "Provider %s is rate-limited (retry_after=%.0fs); skipping to next provider",
+                    label,
+                    exc.retry_after or 0,
+                )
+                continue
             except Exception as e:
                 logger.debug("Provider %s failed for get_user_models: %s", label, e)
                 continue

py/services/model_query.py

@@ -294,12 +294,14 @@ class ModelFilterSet:
                 for tag, state in tag_filters.items():
                     if not tag:
                         continue
+                    # Normalize to lowercase for case-insensitive matching
+                    normalized = tag.strip().lower()
                     if state == "exclude":
-                        exclude_tags.add(tag)
+                        exclude_tags.add(normalized)
                     else:
-                        include_tags.add(tag)
+                        include_tags.add(normalized)
             else:
-                include_tags = {tag for tag in tag_filters if tag}
+                include_tags = {tag.strip().lower() for tag in tag_filters if tag}
 
             if include_tags:
                 tag_logic = criteria.tag_logic.lower() if criteria.tag_logic else "any"
@@ -318,13 +320,17 @@ class ModelFilterSet:
                                 return True
                             # Otherwise, check if all non-special tags match
                             if non_special_tags:
-                                return all(tag in (item_tags or []) for tag in non_special_tags)
+                                # Case-insensitive: normalize item tags too
+                                normalized_item_tags = {t.strip().lower() for t in (item_tags or []) if isinstance(t, str)}
+                                return all(tag in normalized_item_tags for tag in non_special_tags)
                             return True
-                        # Normal case: all tags must match
-                        return all(tag in (item_tags or []) for tag in non_special_tags)
+                        # Normal case: all tags must match (case-insensitive)
+                        normalized_item_tags = {t.strip().lower() for t in (item_tags or []) if isinstance(t, str)}
+                        return all(tag in normalized_item_tags for tag in non_special_tags)
                     else:
-                        # OR logic (default): item must have ANY include tag
-                        return any(tag in include_tags for tag in (item_tags or []))
+                        # OR logic (default): item must have ANY include tag (case-insensitive)
+                        normalized_item_tags = {t.strip().lower() for t in (item_tags or []) if isinstance(t, str)}
+                        return bool(normalized_item_tags & include_tags)
 
                 items = [item for item in items if matches_include(item.get("tags"))]
 
@@ -333,7 +339,9 @@ class ModelFilterSet:
                 def matches_exclude(item_tags):
                     if not item_tags and "__no_tags__" in exclude_tags:
                         return True
-                    return any(tag in exclude_tags for tag in (item_tags or []))
+                    # Case-insensitive: normalize item tags
+                    normalized_item_tags = {t.strip().lower() for t in (item_tags or []) if isinstance(t, str)}
+                    return bool(normalized_item_tags & exclude_tags)
 
                 items = [
                     item for item in items if not matches_exclude(item.get("tags"))

py/services/model_scanner.py

@@ -14,7 +14,7 @@ from ..utils.metadata_manager import MetadataManager
 from ..utils.civitai_utils import resolve_license_info
 from .model_cache import ModelCache
 from .model_hash_index import ModelHashIndex
-from .model_lifecycle_service import delete_model_artifacts
+from .model_lifecycle_service import delete_model_artifacts, _require_path_in_library_roots
 from .service_registry import ServiceRegistry
 from .websocket_manager import ws_manager
 from .persistent_model_cache import get_persistent_cache
@@ -227,6 +227,11 @@ class ModelScanner:
 
         entry: Dict[str, Any] = {
             'file_path': normalized_path,
+            # file_name is always stored WITHOUT extension (e.g. "OWSMianne_ANIMA_V1",
+            # not "OWSMianne_ANIMA_V1.safetensors"). All upstream population points
+            # (MetadataManager, from_civitai_info, download manager, etc.) strip the
+            # extension via os.path.splitext before writing. Code consuming this field
+            # should match against names that are likewise extension-free.
             'file_name': get_value('file_name', '') or '',
             'model_name': get_value('model_name', '') or '',
             'folder': normalized_folder,
@@ -248,6 +253,7 @@ class ModelScanner:
             'civitai': civitai_slim,
             'civitai_deleted': bool(get_value('civitai_deleted', False)),
             'skip_metadata_refresh': bool(get_value('skip_metadata_refresh', False)),
+            'hf_url': get_value('hf_url', '') or '',
         }
 
         license_source: Dict[str, Any] = {}
@@ -476,11 +482,20 @@ class ModelScanner:
             for tag in adjusted_item.get('tags') or []:
                 tags_count[tag] = tags_count.get(tag, 0) + 1
 
-        # Validate cache entries and check health
+        # Validate cache entries and check health.
+        # Always use the validated/repaired entries — even when there are no
+        # invalid entries, auto_repair may have filled in missing optional
+        # fields (model_name, file_name, folder) with safe defaults on a copied
+        # working_entry.  Without this unconditional replacement the repaired
+        # copies are discarded and None values propagate to format_response.
+        # See issue #730.
         valid_entries, invalid_entries = CacheEntryValidator.validate_batch(
             adjusted_raw_data, auto_repair=True
         )
 
+        # Always use the validated entries (repaired copies)
+        adjusted_raw_data = valid_entries
+
         if invalid_entries:
             monitor = CacheHealthMonitor()
             report = monitor.check_health(adjusted_raw_data, auto_repair=True)
@@ -532,6 +547,13 @@ class ModelScanner:
         if not scan_result or not getattr(self, '_persistent_cache', None):
             return
 
+        if self.is_cancelled():
+            logger.info(
+                f"{self.model_type.capitalize()} Scanner: Skipping _save_persistent_cache "
+                "after cancellation"
+            )
+            return
+
         hash_snapshot = self._build_hash_index_snapshot(scan_result.hash_index)
         loop = asyncio.get_event_loop()
         try:
@@ -705,14 +727,20 @@ class ModelScanner:
             # Determine the page type based on model type
             # Scan for new data
             scan_result = await self._gather_model_data()
-            await self._apply_scan_result(scan_result)
-            await self._save_persistent_cache(scan_result)
-            await self._sync_download_history(scan_result.raw_data, source='scan')
+            if not self.is_cancelled():
+                await self._apply_scan_result(scan_result)
+                await self._save_persistent_cache(scan_result)
+                await self._sync_download_history(scan_result.raw_data, source='scan')
 
-            logger.info(
-                f"{self.model_type.capitalize()} Scanner: Cache initialization completed in {time.time() - start_time:.2f} seconds, "
-                f"found {len(scan_result.raw_data)} models"
-            )
+                logger.info(
+                    f"{self.model_type.capitalize()} Scanner: Cache initialization completed in {time.time() - start_time:.2f} seconds, "
+                    f"found {len(scan_result.raw_data)} models"
+                )
+            else:
+                logger.info(
+                    f"{self.model_type.capitalize()} Scanner: Cache initialization cancelled "
+                    f"after {time.time() - start_time:.2f} seconds"
+                )
         except Exception as e:
             logger.error(f"{self.model_type.capitalize()} Scanner: Error initializing cache: {e}")
             # Ensure cache is at least an empty structure on error
@@ -899,6 +927,25 @@ class ModelScanner:
                 # Update cache data
                 self._cache.raw_data = [item for item in self._cache.raw_data if item['file_path'] not in missing_files]
             
+            dedup_removed = 0
+            seen_paths: set = set()
+            deduped: list = []
+            for item in reversed(self._cache.raw_data):
+                path = item.get('file_path', '')
+                if path not in seen_paths:
+                    seen_paths.add(path)
+                    deduped.append(item)
+                else:
+                    for tag in item.get('tags', []):
+                        if tag in self._tags_count:
+                            self._tags_count[tag] = max(0, self._tags_count[tag] - 1)
+                            if self._tags_count[tag] == 0:
+                                del self._tags_count[tag]
+                    dedup_removed += 1
+            if dedup_removed > 0:
+                self._cache.raw_data = list(reversed(deduped))
+                total_removed += dedup_removed
+            
             # Resort cache if changes were made
             if total_added > 0 or total_removed > 0:
                 # Update folders list
@@ -1067,8 +1114,11 @@ class ModelScanner:
 
         model_data = self._build_cache_entry(metadata, folder=normalized_folder)
 
-        # Compute SHA256 hash when metadata provided none (e.g., CivitAI API response has empty hashes)
-        if not model_data.get('sha256') and file_path:
+        # Compute SHA256 hash when metadata provided none (e.g., CivitAI API response has empty hashes).
+        # Respect hash_status='pending' (set by CheckpointScanner for large models) to defer
+        # hash calculation until on-demand — avoids reading entire checkpoint files at startup.
+        hash_status = model_data.get('hash_status', '')
+        if not model_data.get('sha256') and hash_status != 'pending' and file_path:
             try:
                 logger.info(f"Computing SHA256 hash for {file_path} (was empty from metadata)")
                 sha256 = await calculate_sha256(file_path)
@@ -1093,6 +1143,13 @@ class ModelScanner:
         if scan_result is None:
             return
 
+        if self.is_cancelled():
+            logger.info(
+                f"{self.model_type.capitalize()} Scanner: Skipping _apply_scan_result "
+                "after cancellation"
+            )
+            return
+
         self._hash_index = scan_result.hash_index
         self._tags_count = dict(scan_result.tags_count)
         self._excluded_models = list(scan_result.excluded_models)
@@ -1314,18 +1371,25 @@ class ModelScanner:
             # Update folder in metadata
             metadata_dict['folder'] = folder
             
-            # Add to cache
-            self._cache.raw_data.append(metadata_dict)
-            self._cache.add_to_version_index(metadata_dict)
+            file_path = metadata_dict.get('file_path', '')
+            if file_path:
+                old_entries = [item for item in self._cache.raw_data if item.get('file_path') == file_path]
+                for old_entry in old_entries:
+                    for tag in old_entry.get('tags', []):
+                        if tag in self._tags_count:
+                            self._tags_count[tag] = max(0, self._tags_count[tag] - 1)
+                            if self._tags_count[tag] == 0:
+                                del self._tags_count[tag]
+                self._hash_index.remove_by_path(file_path)
+                self._cache.raw_data = [item for item in self._cache.raw_data if item.get('file_path') != file_path]
+
+            for tag in metadata_dict.get('tags', []):
+                self._tags_count[tag] = self._tags_count.get(tag, 0) + 1
+
+            self._cache.raw_data.append(metadata_dict)
 
-            # Resort cache data
             await self._cache.resort()
             
-            # Update folders list
-            all_folders = set(self._cache.folders)
-            all_folders.add(folder)
-            self._cache.folders = sorted(list(all_folders), key=lambda x: x.lower())
-            
             # Update the hash index
             self._hash_index.add_entry(metadata_dict['sha256'], metadata_dict['file_path'])
             await self._persist_current_cache()
@@ -1356,6 +1420,9 @@ class ModelScanner:
                 
             base_name = os.path.splitext(os.path.basename(source_path))[0]
             source_dir = os.path.dirname(source_path)
+
+            _require_path_in_library_roots(source_path, self, label="Source path")
+            _require_path_in_library_roots(target_path, self, label="Target path")
             
             os.makedirs(target_path, exist_ok=True)
             
@@ -1528,6 +1595,218 @@ class ModelScanner:
 
         return cache_entry if metadata else True
         
+    async def sync_cache_from_metadata(
+        self, file_path: str, metadata_dict: Dict[str, Any]
+    ) -> bool:
+        """Opportunistically sync in-memory and persistent caches from metadata.
+
+        Builds a prospective cache entry from *metadata_dict* (deserialized
+        ``.metadata.json`` content) and compares it against the current cache
+        entry.  When the two are already identical this method returns
+        ``False`` without touching anything — avoiding the overhead of
+        ``update_single_model_cache``, which always removes and re-inserts
+        the entry, triggers a full resort, and persists via the heavyweight
+        ``save_cache()``.
+
+        When differences are detected the update is applied **in-place** with
+        targeted operations:
+
+        * The existing ``raw_data`` entry is modified rather than removed and
+          re-appended (O(1) instead of O(n)).
+        * Tag counts and the hash index are updated incrementally.
+        * The version index is rebuilt only for the affected entry.
+        * ``resort()`` is called **only** when a sort-relevant field changed
+          (``model_name`` / ``file_name`` for name-sort, ``modified`` for
+          date-sort, ``size`` for size-sort).
+        * The persistent (SQLite) cache receives a targeted single-row update
+          via :meth:`PersistentModelCache.update_single_model` rather than a
+          full-table ``save_cache()``.
+
+        Returns:
+            ``True`` if any cache update was performed, ``False`` if the
+            caches were already in sync.
+
+        .. note::
+
+            This is a **best-effort** operation.  Failures are logged but
+            never propagated — callers should fire-and-forget via
+            :func:`asyncio.create_task`.
+        """
+        try:
+            return await self._sync_cache_from_metadata_impl(
+                file_path, metadata_dict
+            )
+        except Exception:
+            logger.warning(
+                "sync_cache_from_metadata failed for %s",
+                file_path,
+                exc_info=True,
+            )
+            return False
+
+    async def _sync_cache_from_metadata_impl(
+        self, file_path: str, metadata_dict: Dict[str, Any]
+    ) -> bool:
+        cache = await self.get_cached_data()
+
+        # Locate the existing cache entry -----------------------------------
+        existing_idx: Optional[int] = None
+        existing_entry: Optional[Dict[str, Any]] = None
+        for i, item in enumerate(cache.raw_data):
+            if item.get("file_path") == file_path:
+                existing_entry = item
+                existing_idx = i
+                break
+
+        # Build the desired entry from metadata ------------------------------
+        folder_value = (
+            existing_entry.get("folder", "")
+            if existing_entry
+            else self._calculate_folder(file_path)
+        )
+        desired_entry = self._build_cache_entry(
+            metadata_dict,
+            folder=folder_value,
+            file_path_override=file_path,
+        )
+
+        # Ensure sha256 is populated (defensive — metadata should have it)
+        if (
+            not desired_entry.get("sha256")
+            and file_path
+            and os.path.exists(file_path)
+        ):
+            try:
+                sha256 = await calculate_sha256(file_path)
+                if sha256:
+                    desired_entry["sha256"] = sha256.lower()
+            except Exception:
+                pass
+
+        # Not in cache at all — delegate to the full update path ------------
+        if existing_entry is None:
+            result = await self.update_single_model_cache(
+                file_path, file_path, metadata_dict
+            )
+            return bool(result)
+
+        # Compare — skip everything if already in sync -----------------------
+        if not self._cache_entries_differ(existing_entry, desired_entry):
+            return False
+
+        # Re-validate: the cache may have been replaced concurrently
+        # (e.g. by _apply_scan_result).  Use identity check, not equality,
+        # so we detect when the raw_data list was swapped out from under us.
+        if self._cache is None or not any(
+            item is existing_entry for item in self._cache.raw_data
+        ):
+            return False
+
+        # ---- Differences detected: apply targeted, in-place updates --------
+
+        # Snapshot old values for delta computations
+        old_tags = list(existing_entry.get("tags") or [])
+        old_sha256: str = existing_entry.get("sha256", "") or ""
+        old_model_name: str = existing_entry.get("model_name", "") or ""
+        old_file_name: str = existing_entry.get("file_name", "") or ""
+        old_modified: float = float(existing_entry.get("modified", 0.0) or 0.0)
+        old_size: int = int(existing_entry.get("size", 0) or 0)
+        old_civitai = existing_entry.get("civitai")
+
+        # ---- In-place update of the cache entry ----
+        existing_entry.clear()
+        existing_entry.update(desired_entry)
+
+        # ---- Incremental tag count update ----
+        new_tags: set = set(desired_entry.get("tags") or [])
+        old_tag_set: set = set(old_tags)
+        for tag in old_tag_set - new_tags:
+            current = self._tags_count.get(tag, 0)
+            if current <= 1:
+                self._tags_count.pop(tag, None)
+            else:
+                self._tags_count[tag] = current - 1
+        for tag in new_tags - old_tag_set:
+            self._tags_count[tag] = self._tags_count.get(tag, 0) + 1
+
+        # ---- Incremental hash index update ----
+        new_sha = (desired_entry.get("sha256", "") or "").lower()
+        old_sha = (old_sha256 or "").lower()
+        if new_sha != old_sha:
+            if old_sha:
+                self._hash_index.remove_by_path(file_path)
+            if new_sha:
+                self._hash_index.add_entry(new_sha, file_path)
+
+        # ---- Incremental version index update ----
+        new_civitai = desired_entry.get("civitai")
+        if old_civitai != new_civitai:
+            temp_old = {
+                "file_path": file_path,
+                "file_name": old_file_name,
+                "civitai": old_civitai,
+            }
+            cache.remove_from_version_index(temp_old)
+            cache.add_to_version_index(existing_entry)
+
+        # ---- Conditional resort (only when sort-key fields changed) ----
+        need_resort = False
+        _last = cache._last_sort
+        sort_key: Optional[str] = _last[0] if _last != (None, None) else None
+        if sort_key == "name":
+            if (
+                old_model_name != desired_entry.get("model_name", "")
+                or old_file_name != desired_entry.get("file_name", "")
+            ):
+                need_resort = True
+        elif sort_key == "date":
+            if old_modified != float(desired_entry.get("modified", 0.0) or 0.0):
+                need_resort = True
+        elif sort_key == "size":
+            if old_size != int(desired_entry.get("size", 0) or 0):
+                need_resort = True
+
+        if need_resort:
+            await cache.resort()
+
+        # ---- Targeted SQL update (single row, not full save_cache) ----
+        persistent = getattr(self, "_persistent_cache", None)
+        if persistent is not None:
+            old_item_for_sql: Dict[str, Any] = {
+                "file_path": file_path,
+                "tags": old_tags,
+                "sha256": old_sha256,
+            }
+            await asyncio.get_event_loop().run_in_executor(
+                None,
+                persistent.update_single_model,
+                self.model_type,
+                desired_entry,
+                old_item_for_sql,
+            )
+
+        return True
+
+    @staticmethod
+    def _cache_entries_differ(a: Dict[str, Any], b: Dict[str, Any]) -> bool:
+        """Return ``True`` when two cache-entry dicts differ in any field.
+
+        Tag lists are compared order-insensitively; all other keys use
+        standard equality.
+        """
+        a_tags = sorted(a.get("tags") or [])
+        b_tags = sorted(b.get("tags") or [])
+        if a_tags != b_tags:
+            return True
+
+        all_keys = set(a.keys()) | set(b.keys())
+        for key in all_keys:
+            if key == "tags":
+                continue
+            if a.get(key) != b.get(key):
+                return True
+        return False
+
     def has_hash(self, sha256: str) -> bool:
         """Check if a model with given hash exists"""
         return self._hash_index.has_hash(sha256.lower())
@@ -1580,7 +1859,32 @@ class ModelScanner:
         if limit == 0:
             return sorted_tags
         return sorted_tags[:limit]
-        
+
+    async def search_tags(
+        self, query: str, limit: int = 50
+    ) -> List[Dict[str, any]]:
+        """Search tags by case-insensitive substring match, sorted by count.
+
+        If query is empty, behaves like get_top_tags (returns top ``limit``
+        tags). If limit is 0, all matching tags are returned.
+        """
+        await self.get_cached_data()
+
+        normalized_query = (query or "").strip().lower()
+        if not normalized_query:
+            return await self.get_top_tags(limit if limit > 0 else 20)
+
+        matched = [
+            {"tag": tag, "count": count}
+            for tag, count in self._tags_count.items()
+            if normalized_query in tag.lower()
+        ]
+        matched.sort(key=lambda x: x["count"], reverse=True)
+
+        if limit == 0:
+            return matched
+        return matched[:limit]
+
     async def get_base_models(self, limit: int = 20) -> List[Dict[str, any]]:
         """Get base models sorted by count. If limit is 0, return all."""
         cache = await self.get_cached_data()
@@ -1696,6 +2000,8 @@ class ModelScanner:
                     break
 
                 try:
+                    _require_path_in_library_roots(file_path, self, label="File path")
+
                     target_dir = os.path.dirname(file_path)
                     base_name = os.path.basename(file_path)
                     file_name, main_extension = os.path.splitext(base_name)
@@ -1761,6 +2067,13 @@ class ModelScanner:
         """
         if not file_paths or self._cache is None:
             return False
+
+        if self.is_cancelled():
+            logger.info(
+                f"{self.model_type.capitalize()} Scanner: Skipping cache update "
+                "after cancelled bulk delete"
+            )
+            return False
             
         try:
             # Get all models that need to be removed from cache

py/services/model_update_service.py

@@ -724,6 +724,16 @@ class ModelUpdateService:
             "Refreshing update metadata for %d %s models", total_models, model_type
         )
 
+        # When filtering by folder, also collect the cross-folder version set
+        # so that versions already present in other folders are not reported
+        # as available updates. See issue #997.
+        all_local_versions: Optional[Dict[int, List[int]]] = None
+        if folder_path is not None:
+            all_local_versions = await self._collect_local_versions(
+                scanner,
+                target_model_ids=target_filter,
+            )
+
         results: Dict[int, ModelUpdateRecord] = {}
         prefetched: Dict[int, Mapping] = {}
 
@@ -762,6 +772,12 @@ class ModelUpdateService:
         for index, (model_id, version_ids) in enumerate(
             local_versions.items(), start=1
         ):
+            # Use cross-folder version IDs for is_in_library if available
+            all_vids: Sequence[int] = (
+                all_local_versions.get(model_id, [])
+                if all_local_versions is not None
+                else version_ids
+            )
             record = await self._refresh_single_model(
                 model_type,
                 model_id,
@@ -769,6 +785,7 @@ class ModelUpdateService:
                 metadata_provider,
                 force_refresh=force_refresh,
                 prefetched_response=prefetched.get(model_id),
+                all_local_version_ids=all_vids,
             )
             if scanner.is_cancelled():
                 logger.info(f"{model_type.capitalize()} Update Service: Refresh cancelled by user")
@@ -964,8 +981,16 @@ class ModelUpdateService:
         *,
         force_refresh: bool = False,
         prefetched_response: Optional[Mapping] = None,
+        all_local_version_ids: Optional[Sequence[int]] = None,
     ) -> Optional[ModelUpdateRecord]:
         normalized_local = self._normalize_sequence(local_versions)
+        # When folder-filtering, this carries the cross-folder version set
+        # for is_in_library; otherwise it falls back to normalized_local.
+        normalized_all = (
+            self._normalize_sequence(all_local_version_ids)
+            if all_local_version_ids is not None
+            else normalized_local
+        )
         now = time.time()
         async with self._lock:
             existing = self._get_record(model_type, model_id)
@@ -973,6 +998,7 @@ class ModelUpdateService:
                 record = self._merge_with_local_versions(
                     existing,
                     normalized_local,
+                    all_local_version_ids=normalized_all,
                 )
                 self._upsert_record(record)
                 return record
@@ -1048,6 +1074,7 @@ class ModelUpdateService:
                 record = self._merge_with_local_versions(
                     existing,
                     normalized_local,
+                    all_local_version_ids=normalized_all,
                 )
                 self._upsert_record(record)
                 return record
@@ -1059,6 +1086,7 @@ class ModelUpdateService:
                     model_type=model_type,
                     model_id=model_id,
                     last_checked_at=now,
+                    all_local_version_ids=normalized_all,
                 )
                 record = replace(record, should_ignore_model=True)
                 self._upsert_record(record)
@@ -1077,6 +1105,7 @@ class ModelUpdateService:
                     fetched_versions,
                     existing,
                     now,
+                    all_local_version_ids=normalized_all,
                 )
             else:
                 record = self._merge_with_local_versions(
@@ -1085,6 +1114,7 @@ class ModelUpdateService:
                     model_type=model_type,
                     model_id=model_id,
                     last_checked_at=existing.last_checked_at if existing else None,
+                    all_local_version_ids=normalized_all,
                 )
             self._upsert_record(record)
             return record
@@ -1322,12 +1352,20 @@ class ModelUpdateService:
         existing: Optional[ModelUpdateRecord],
         normalized_local: Sequence[int],
         *,
+        all_local_version_ids: Optional[Sequence[int]] = None,
         model_type: Optional[str] = None,
         model_id: Optional[int] = None,
         last_checked_at: Optional[float] = None,
         version_info: Optional[Mapping] = None,
     ) -> ModelUpdateRecord:
         local_set = set(normalized_local)
+        # When folder-filtering, also consider versions in other folders
+        # as in-library so they are not reported as available updates.
+        effective_local_set: set[int] = (
+            local_set | set(all_local_version_ids)
+            if all_local_version_ids is not None
+            else local_set
+        )
         versions: List[ModelVersionRecord] = []
         ignore_map: Dict[int, bool] = {}
         if existing:
@@ -1339,7 +1377,7 @@ class ModelUpdateService:
                 versions.append(
                     replace(
                         version,
-                        is_in_library=version.version_id in local_set,
+                        is_in_library=version.version_id in effective_local_set,
                     )
                 )
         elif model_type is None or model_id is None:
@@ -1386,8 +1424,17 @@ class ModelUpdateService:
         remote_versions: Sequence[ModelVersionRecord],
         existing: Optional[ModelUpdateRecord],
         timestamp: float,
+        *,
+        all_local_version_ids: Optional[Sequence[int]] = None,
     ) -> ModelUpdateRecord:
         local_set = set(local_versions)
+        # When folder-filtering, also consider versions in other folders
+        # as in-library so they are not reported as available updates.
+        effective_local_set: set[int] = (
+            local_set | set(all_local_version_ids)
+            if all_local_version_ids is not None
+            else local_set
+        )
         ignore_map = {version.version_id: version.should_ignore for version in existing.versions} if existing else {}
         preview_map = {version.version_id: version.preview_url for version in existing.versions} if existing else {}
         sort_map = {version.version_id: version.sort_index for version in existing.versions} if existing else {}
@@ -1406,7 +1453,7 @@ class ModelUpdateService:
                     released_at=remote_version.released_at,
                     size_bytes=remote_version.size_bytes,
                     preview_url=remote_version.preview_url or preview_map.get(version_id),
-                    is_in_library=version_id in local_set,
+                    is_in_library=version_id in effective_local_set,
                     should_ignore=ignore_map.get(version_id, remote_version.should_ignore),
                     sort_index=sort_map.get(version_id, index),
                     early_access_ends_at=remote_version.early_access_ends_at,

py/services/persistent_model_cache.py

@@ -57,6 +57,7 @@ class PersistentModelCache:
         "db_checked",
         "last_checked_at",
         "hash_status",
+        "hf_url",
     )
     _MODEL_UPDATE_COLUMNS: Tuple[str, ...] = _MODEL_COLUMNS[2:]
     _instances: Dict[str, "PersistentModelCache"] = {}
@@ -165,8 +166,8 @@ class PersistentModelCache:
 
             item = {
                 "file_path": file_path,
-                "file_name": row["file_name"],
-                "model_name": row["model_name"],
+                "file_name": row["file_name"] or "",
+                "model_name": row["model_name"] or "",
                 "folder": row["folder"] or "",
                 "size": row["size"] or 0,
                 "modified": row["modified"] or 0.0,
@@ -188,6 +189,7 @@ class PersistentModelCache:
                 "skip_metadata_refresh": bool(row["skip_metadata_refresh"]),
                 "license_flags": int(license_value),
                 "hash_status": row["hash_status"] or "completed",
+                "hf_url": row["hf_url"] or "",
             }
             raw_data.append(item)
 
@@ -452,6 +454,7 @@ class PersistentModelCache:
                             db_checked INTEGER,
                             last_checked_at REAL,
                             hash_status TEXT,
+                            hf_url TEXT DEFAULT '',
                             PRIMARY KEY (model_type, file_path)
                         );
 
@@ -500,6 +503,7 @@ class PersistentModelCache:
             # Persisting without explicit flags should assume CivitAI's documented defaults (0b111001 == 57).
             "license_flags": f"INTEGER DEFAULT {DEFAULT_LICENSE_FLAGS}",
             "hash_status": "TEXT DEFAULT 'completed'",
+            "hf_url": "TEXT DEFAULT ''",
         }
 
         for column, definition in required_columns.items():
@@ -548,19 +552,19 @@ class PersistentModelCache:
         return (
             model_type,
             item.get("file_path"),
-            item.get("file_name"),
-            item.get("model_name"),
-            item.get("folder"),
+            item.get("file_name") or "",
+            item.get("model_name") or "",
+            item.get("folder") or "",
             int(item.get("size") or 0),
             float(item.get("modified") or 0.0),
             (item.get("sha256") or "").lower() or None,
-            item.get("base_model"),
-            item.get("preview_url"),
+            item.get("base_model") or "",
+            item.get("preview_url") or "",
             int(item.get("preview_nsfw_level") or 0),
             1 if item.get("from_civitai", True) else 0,
             1 if item.get("favorite") else 0,
-            item.get("notes"),
-            item.get("usage_tips"),
+            item.get("notes") or "",
+            item.get("usage_tips") or "",
             metadata_source,
             civitai.get("id"),
             civitai.get("modelId"),
@@ -575,6 +579,7 @@ class PersistentModelCache:
             1 if item.get("db_checked") else 0,
             float(item.get("last_checked_at") or 0.0),
             item.get("hash_status", "completed"),
+            item.get("hf_url") or "",
         )
 
     def _insert_model_sql(self) -> str:
@@ -582,6 +587,95 @@ class PersistentModelCache:
         placeholders = ", ".join(["?"] * len(self._MODEL_COLUMNS))
         return f"INSERT INTO models ({columns}) VALUES ({placeholders})"
 
+    def update_single_model(
+        self,
+        model_type: str,
+        new_item: Dict,
+        old_item: Optional[Dict] = None,
+    ) -> None:
+        """Update a single model row in the persistent cache.
+
+        A lightweight alternative to :meth:`save_cache` that performs a targeted
+        DELETE + INSERT for the model row and computes incremental tag / hash-index
+        deltas from *old_item*.  When *old_item* is omitted the previous tags and
+        hash are not cleaned up (callers should only omit it for brand-new entries).
+
+        All operations run inside a single transaction so readers see a consistent
+        view.
+        """
+        if not self.is_enabled():
+            return
+        if not self._schema_initialized:
+            self._initialize_schema()
+        if not self._schema_initialized:
+            return
+
+        file_path: Optional[str] = new_item.get("file_path")
+        if not file_path:
+            return
+
+        try:
+            with self._db_lock:
+                conn = self._connect()
+                try:
+                    conn.execute("PRAGMA foreign_keys = ON")
+                    conn.execute("BEGIN")
+
+                    # --- model row (DELETE + INSERT = upsert) ---
+                    conn.execute(
+                        "DELETE FROM models WHERE model_type = ? AND file_path = ?",
+                        (model_type, file_path),
+                    )
+                    row = self._prepare_model_row(model_type, new_item)
+                    conn.execute(self._insert_model_sql(), row)
+
+                    # --- tags ---
+                    new_tags: set = set(new_item.get("tags") or [])
+                    old_tags: set = set(old_item.get("tags") or []) if old_item else set()
+                    tags_to_delete = old_tags - new_tags
+                    tags_to_insert = new_tags - old_tags
+
+                    if tags_to_delete:
+                        conn.executemany(
+                            "DELETE FROM model_tags WHERE model_type = ? AND file_path = ? AND tag = ?",
+                            [(model_type, file_path, t) for t in tags_to_delete],
+                        )
+                    if tags_to_insert:
+                        conn.executemany(
+                            "INSERT INTO model_tags (model_type, file_path, tag) VALUES (?, ?, ?)",
+                            [(model_type, file_path, t) for t in tags_to_insert],
+                        )
+
+                    # --- hash_index ---
+                    new_sha: Optional[str] = (new_item.get("sha256") or "").lower() or None
+                    old_sha: Optional[str] = (
+                        (old_item.get("sha256") or "").lower() or None
+                    ) if old_item else None
+                    if new_sha != old_sha:
+                        if old_sha:
+                            conn.execute(
+                                "DELETE FROM hash_index WHERE model_type = ? AND sha256 = ? AND file_path = ?",
+                                (model_type, old_sha, file_path),
+                            )
+                        if new_sha:
+                            conn.execute(
+                                "INSERT OR IGNORE INTO hash_index (model_type, sha256, file_path) VALUES (?, ?, ?)",
+                                (model_type, new_sha, file_path),
+                            )
+
+                    conn.execute("COMMIT")
+                except Exception:
+                    conn.execute("ROLLBACK")
+                    raise
+                finally:
+                    conn.close()
+        except Exception as exc:
+            logger.warning(
+                "Failed to update single model in persistent cache (%s): %s",
+                file_path,
+                exc,
+            )
+
     def _load_tags(self, conn: sqlite3.Connection, model_type: str) -> Dict[str, List[str]]:
         tag_rows = conn.execute(
             "SELECT file_path, tag FROM model_tags WHERE model_type = ?",

py/services/recipe_cache.py

@@ -1,7 +1,6 @@
 import asyncio
 from typing import Iterable, List, Dict, Optional
 from dataclasses import dataclass, field
-from operator import itemgetter
 from natsort import natsorted
 
 
@@ -149,5 +148,10 @@ class RecipeCache:
         )
         if not name_only:
             self.sorted_by_date = sorted(
-                self.raw_data, key=itemgetter("created_date", "file_path"), reverse=True
+                self.raw_data,
+                key=lambda x: (
+                    x.get("modified", x.get("created_date", 0)),
+                    x.get("file_path", ""),
+                ),
+                reverse=True,
             )

py/services/recipe_scanner.py

@@ -21,7 +21,7 @@ from .checkpoint_scanner import CheckpointScanner
 from .settings_manager import get_settings_manager
 from .recipes.errors import RecipeNotFoundError
 from ..utils.civitai_utils import extract_civitai_image_id
-from ..utils.utils import calculate_recipe_fingerprint, fuzzy_match
+from ..utils.utils import calculate_recipe_fingerprint
 from natsort import natsorted
 import sys
 import re
@@ -1020,13 +1020,16 @@ class RecipeScanner:
 
         try:
             result = self._fts_index.search(search, fields)
-            # Return None if empty to trigger fuzzy fallback
-            # Empty FTS results may indicate query syntax issues or need for fuzzy matching
+            # Return empty set for empty FTS results — do NOT fall back to
+            # Python fuzzy matching, which freezes the server with 10k+ recipes.
+            # FTS5 prefix matching with unicode61 tokenizer correctly handles
+            # compound tokens (e.g. "illustrious" matches "path/illustrious/model").
+            # If FTS returns nothing, there are genuinely no matching recipes.
             if not result:
-                return None
+                return set()
             return result
         except Exception as exc:
-            logger.debug("FTS search failed, falling back to fuzzy search: %s", exc)
+            logger.debug("FTS search failed, falling back to title-only search: %s", exc)
             return None
 
     def _update_fts_index_for_recipe(
@@ -2079,49 +2082,14 @@ class RecipeScanner:
                         if str(item.get("id", "")) in fts_matching_ids
                     ]
                 else:
-                    # Fallback to fuzzy_match (slower but always available)
-                    # Build the search predicate based on search options
-                    def matches_search(item):
-                        # Search in title if enabled
-                        if search_options.get("title", True):
-                            if fuzzy_match(str(item.get("title", "")), search):
-                                return True
-
-                        # Search in tags if enabled
-                        if search_options.get("tags", True) and "tags" in item:
-                            for tag in item["tags"]:
-                                if fuzzy_match(tag, search):
-                                    return True
-
-                        # Search in lora file names if enabled
-                        if search_options.get("lora_name", True) and "loras" in item:
-                            for lora in item["loras"]:
-                                if fuzzy_match(str(lora.get("file_name", "")), search):
-                                    return True
-
-                        # Search in lora model names if enabled
-                        if search_options.get("lora_model", True) and "loras" in item:
-                            for lora in item["loras"]:
-                                if fuzzy_match(str(lora.get("modelName", "")), search):
-                                    return True
-
-                        # Search in prompt and negative_prompt if enabled
-                        if search_options.get("prompt", True) and "gen_params" in item:
-                            gen_params = item["gen_params"]
-                            if fuzzy_match(str(gen_params.get("prompt", "")), search):
-                                return True
-                            if fuzzy_match(
-                                str(gen_params.get("negative_prompt", "")), search
-                            ):
-                                return True
-
-                        # No match found
-                        return False
-
-                    # Filter the data using the search predicate
-                    filtered_data = [
-                        item for item in filtered_data if matches_search(item)
-                    ]
+                    # FTS index not yet built — return empty rather than
+                    # scanning 42k+ items in Python. The FTS background build
+                    # finishes in seconds; by the time a user navigates here
+                    # and types a search, it is already available.
+                    logger.debug(
+                        "FTS index not ready — search '%s' returning empty", search
+                    )
+                    filtered_data = []
 
             # Apply additional filters
             if filters:

py/services/recipes/analysis_service.py

@@ -146,11 +146,38 @@ class RecipeAnalysisService:
                 ):
                     metadata = metadata["meta"]
 
-                # Include modelVersionIds from root level if available
-                # Civitai API returns modelVersionIds at root level, not in meta
+                # Include modelVersionIds from root level if available.
+                # CivitAI API returns modelVersionIds at root level, not in meta.
+                # When meta is null (None), create a minimal dict so downstream
+                # parsers can still discover LoRAs and checkpoints.
                 model_version_ids = image_info.get("modelVersionIds")
-                if model_version_ids and isinstance(metadata, dict):
-                    metadata["modelVersionIds"] = model_version_ids
+                if model_version_ids:
+                    if isinstance(metadata, dict):
+                        metadata["modelVersionIds"] = model_version_ids
+                    else:
+                        metadata = {"modelVersionIds": model_version_ids}
+
+                # Inject browsingLevel (canonical integer) so the recipe's
+                # preview_nsfw_level can be set, enabling proper NSFW blur
+                # of the preview image.  Fall back to nsfwLevel (string)
+                # when browsingLevel is absent.
+                if isinstance(metadata, dict):
+                    browsing_level = image_info.get("browsingLevel")
+                    nsfw_level_str = image_info.get("nsfwLevel")
+                    if isinstance(browsing_level, int) and browsing_level > 0:
+                        metadata["browsingLevel"] = browsing_level
+                    elif (
+                        isinstance(nsfw_level_str, str)
+                        and nsfw_level_str
+                        in (
+                            "PG", "PG13", "R", "X", "XXX", "Blocked",
+                        )
+                    ):
+                        from ...utils.constants import NSFW_LEVELS
+
+                        metadata["browsingLevel"] = NSFW_LEVELS.get(
+                            nsfw_level_str, 0
+                        )
 
                 # Validate that metadata contains meaningful recipe fields
                 # If not, treat as None to trigger EXIF extraction from downloaded image
@@ -171,12 +198,19 @@ class RecipeAnalysisService:
                 temp_path = self._create_temp_path(suffix=extension)
                 await self._download_image(url, temp_path)
 
-            if metadata is None and not is_video:
-                metadata = await asyncio.to_thread(
+            # Always extract EXIF from the downloaded image for generation
+            # params (prompt, negative prompt, sampler, steps, etc.).
+            # Previously this was gated on ``metadata is None``, but that
+            # skipped EXIF entirely when API metadata (modelVersionIds,
+            # browsingLevel) is present, losing all generation parameters.
+            exif_metadata = None
+            if not is_video:
+                exif_metadata = await asyncio.to_thread(
                     self._exif_utils.extract_image_metadata, temp_path
                 )
 
-            if not metadata and civitai_image_id and image_info:
+            # Fallback: try the original (non-optimized) image for EXIF data
+            if not exif_metadata and civitai_image_id and image_info:
                 original_url = image_info.get("url")
                 if original_url:
                     self._logger.debug(
@@ -187,15 +221,38 @@ class RecipeAnalysisService:
                     orig_temp_path = self._create_temp_path(suffix=".png")
                     try:
                         await self._download_image(original_url, orig_temp_path)
-                        metadata = await asyncio.to_thread(
+                        exif_metadata = await asyncio.to_thread(
                             self._exif_utils.extract_image_metadata,
                             orig_temp_path,
                         )
                     finally:
                         self._safe_cleanup(orig_temp_path)
 
+            # Parse EXIF data (typically a string like parameters/prompt/workflow)
+            # and API metadata (dict with modelVersionIds, browsingLevel) separately,
+            # then merge: API loras/checkpoint override, EXIF gen_params fill in gaps.
+            # This mirrors the two-pass approach in _do_import_from_url.
+            exif_parsed_result = None
+            if isinstance(exif_metadata, str):
+                exif_parser = self._recipe_parser_factory.create_parser(exif_metadata)
+                if exif_parser:
+                    exif_data = await exif_parser.parse_metadata(
+                        exif_metadata, recipe_scanner=recipe_scanner,
+                    )
+                    if exif_data and not exif_data.get("error"):
+                        exif_parsed_result = exif_data
+
+            # Merge API metadata (dict) with EXIF data (if dict) for the
+            # CivitaiApiMetadataParser.  If EXIF data is a string it was
+            # parsed above — don't try to merge a string into a dict.
+            merged = {}
+            if isinstance(exif_metadata, dict):
+                merged.update(exif_metadata)
+            if isinstance(metadata, dict):
+                merged.update(metadata)
+
             result = await self._parse_metadata(
-                metadata or {},
+                merged,
                 recipe_scanner=recipe_scanner,
                 image_path=temp_path,
                 include_image_base64=True,
@@ -203,13 +260,23 @@ class RecipeAnalysisService:
                 extension=extension,
             )
 
-            if civitai_image_id and image_info and not result.payload.get("error"):
-                mvid = image_info.get("modelVersionId")
-                if not mvid:
-                    mvids = image_info.get("modelVersionIds")
-                    if isinstance(mvids, list) and mvids:
-                        mvid = mvids[0]
+            # Merge EXIF string-parsed gen_params into the API result.
+            # API gen_params take priority (they come later via update).
+            if exif_parsed_result and not result.payload.get("error"):
+                exif_gp = exif_parsed_result.get("gen_params") or {}
+                result_gp = result.payload.get("gen_params") or {}
+                merged_gp = {**exif_gp, **result_gp}
+                if merged_gp:
+                    result.payload["gen_params"] = merged_gp
 
+            if civitai_image_id and image_info and not result.payload.get("error"):
+                # Use the metadata dict we built (may contain modelVersionIds
+                # and browsingLevel from the API root level).  Do NOT pass
+                # image_info.get("meta") — it is null for images whose meta
+                # lives at the root level only.  Also do NOT derive
+                # model_version_id from modelVersionIds[0] — that array mixes
+                # checkpoints, LoRAs, and other types without ordering
+                # guarantees; the parser already resolved them correctly.
                 recipe_for_enrich = {
                     "gen_params": result.payload.get("gen_params", {}),
                     "loras": result.payload.get("loras", []),
@@ -222,8 +289,10 @@ class RecipeAnalysisService:
                     recipe=recipe_for_enrich,
                     civitai_client=civitai_client,
                     request_params=None,
-                    prefetched_civitai_meta_raw=image_info.get("meta"),
-                    prefetched_model_version_id=mvid,
+                    prefetched_civitai_meta_raw=(
+                        metadata if isinstance(metadata, dict) else None
+                    ),
+                    prefetched_model_version_id=None,
                 )
 
                 result.payload["gen_params"] = recipe_for_enrich["gen_params"]
@@ -232,6 +301,12 @@ class RecipeAnalysisService:
                 if recipe_for_enrich.get("base_model"):
                     result.payload["base_model"] = recipe_for_enrich["base_model"]
 
+                # Extract browsingLevel from our constructed metadata for NSFW blur
+                if isinstance(metadata, dict):
+                    bl = metadata.get("browsingLevel")
+                    if isinstance(bl, int) and bl > 0:
+                        result.payload["preview_nsfw_level"] = bl
+
             return result
         finally:
             if temp_path:
@@ -314,6 +389,10 @@ class RecipeAnalysisService:
             "prompt_type",
             "positive",
             "negative",
+            # modelVersionIds is injected at the root level by CivitAI's image
+            # API when meta is null. It carries the version IDs of ALL models
+            # (checkpoint + LoRAs) used to generate the image.
+            "modelVersionIds",
         }
         return any(field in metadata for field in recipe_fields)
 

py/services/recipes/persistence_service.py

@@ -216,11 +216,12 @@ class RecipePersistenceService:
             "preview_nsfw_level",
             "favorite",
             "gen_params",
+            "base_model",
         )
 
         if not any(key in updates for key in allowed_fields):
             raise RecipeValidationError(
-                "At least one field to update must be provided (title or tags or source_path or preview_nsfw_level or favorite or gen_params)"
+                "At least one field to update must be provided (title or tags or source_path or preview_nsfw_level or favorite or gen_params or base_model)"
             )
 
         if "gen_params" in updates and not isinstance(updates["gen_params"], dict):

py/services/settings_manager.py

@@ -65,6 +65,8 @@ DEFAULT_SETTINGS: Dict[str, Any] = {
     "onboarding_completed": False,
     "dismissed_banners": [],
     "enable_metadata_archive_db": False,
+    "enable_civarchive_api": True,
+    "metadata_provider_order": "civitai_archive_sqlite",
     "proxy_enabled": False,
     "proxy_host": "",
     "proxy_port": "",
@@ -91,7 +93,6 @@ DEFAULT_SETTINGS: Dict[str, Any] = {
     "autoplay_on_hover": False,
     "display_density": "default",
     "card_info_display": "always",
-    "show_folder_sidebar": True,
     "include_trigger_words": False,
     "compact_mode": False,
     "priority_tags": DEFAULT_PRIORITY_TAG_CONFIG.copy(),
@@ -99,13 +100,20 @@ DEFAULT_SETTINGS: Dict[str, Any] = {
     "lora_syntax_format": "legacy",
     "model_card_footer_action": "replace_preview",
     "show_version_on_card": True,
-    "update_flag_strategy": "same_base",
+    "version_grouping": "same_base",
     "auto_organize_exclusions": [],
     "metadata_refresh_skip_paths": [],
     "skip_previously_downloaded_model_versions": False,
     "download_skip_base_models": [],
     "backup_auto_enabled": True,
     "backup_retention_count": 5,
+    "use_new_license_icons": True,
+    "group_by_model": False,
+    # AI / LLM provider configuration (BYOK)
+    "llm_provider": "openai",  # "openai" | "ollama" | "custom"
+    "llm_api_key": "",
+    "llm_api_base": "",  # empty = provider default
+    "llm_model": "",  # e.g. "gpt-4o-mini"
 }
 
 
@@ -134,6 +142,9 @@ class SettingsManager:
         self._template_path = (
             Path(__file__).resolve().parents[2] / "settings.json.example"
         )
+        # Known placeholder value in settings.json.example; any file containing
+        # this value should be treated as "not configured".
+        self._TEMPLATE_PLACEHOLDER_API_KEY = "your_civitai_api_key_here"
         self.settings = self._load_settings()
         self._migrate_setting_keys()
         self._ensure_default_settings()
@@ -143,6 +154,11 @@ class SettingsManager:
         self._check_environment_variables()
         self._collect_configuration_warnings()
 
+        if os.environ.get("LORA_MANAGER_PORTABLE", "0") == "1":
+            if not self.settings.get("use_portable_settings"):
+                self.settings["use_portable_settings"] = True
+                self._save_settings()
+
         if self._needs_initial_save:
             self._save_settings()
             self._needs_initial_save = False
@@ -165,6 +181,12 @@ class SettingsManager:
                     self._original_disk_payload = copy.deepcopy(data)
                     if self._matches_template_payload(data):
                         self._preserve_disk_template = True
+                    # Clean up the template placeholder so it is not treated
+                    # as a real key (affects both the frontend boolean and
+                    # the downloader's Authorization header).
+                    placeholder = self._TEMPLATE_PLACEHOLDER_API_KEY
+                    if data.get("civitai_api_key") == placeholder:
+                        data["civitai_api_key"] = ""
                 return data
             except json.JSONDecodeError as exc:
                 logger.error("Failed to parse settings.json: %s", exc)
@@ -610,12 +632,37 @@ class SettingsManager:
 
         return False
 
+    @staticmethod
+    def _normalize_path_set(paths: Iterable[str]) -> set[str]:
+        """Normalize an iterable of paths for set-based overlap comparison.
+
+        Resolves symlinks via ``os.path.realpath`` when the path exists on disk,
+        then applies ``os.path.normcase`` + ``os.path.normpath`` for consistent
+        cross-platform comparison.  Non-string / empty entries are skipped.
+        """
+        result: set[str] = set()
+        for p in paths:
+            if not isinstance(p, str):
+                continue
+            stripped = p.strip()
+            if not stripped:
+                continue
+            if os.path.exists(stripped):
+                stripped = os.path.normpath(os.path.realpath(stripped))
+            result.add(os.path.normcase(stripped))
+        return result
+
     def _validate_folder_paths(
         self,
         library_name: str,
         folder_paths: Mapping[str, Iterable[str]],
     ) -> None:
-        """Ensure folder paths do not overlap with other libraries."""
+        """Ensure folder paths do not overlap with other libraries.
+
+        Also detects checkpoints ↔ unet path overlap within the same library
+        (including via symlink resolution), which is a configuration error since
+        these model types must use separate physical folders.
+        """
         libraries = self.settings.get("libraries", {})
         normalized_new: Dict[str, Dict[str, str]] = {}
         for key, values in folder_paths.items():
@@ -653,6 +700,22 @@ class SettingsManager:
                         f"Folder path(s) {collisions} already assigned to library '{other_name}'"
                     )
 
+        # Checkpoints ↔ unet overlap within the same library
+        ckpt_paths = folder_paths.get("checkpoints", []) or []
+        unet_paths = folder_paths.get("unet", []) or []
+        if ckpt_paths and unet_paths:
+            ckpt_real = self._normalize_path_set(ckpt_paths)
+            unet_real = self._normalize_path_set(unet_paths)
+            overlap = ckpt_real & unet_real
+            if overlap:
+                collisions = ", ".join(sorted(overlap))
+                raise ValueError(
+                    f"Path(s) {collisions} are configured for both "
+                    f"'checkpoints' and 'unet' (diffusion models). "
+                    f"These model types must use separate physical folders. "
+                    f"Please remove one of the conflicting entries."
+                )
+
     def _update_active_library_entry(
         self,
         *,
@@ -735,6 +798,7 @@ class SettingsManager:
             "includeTriggerWords": "include_trigger_words",
             "compactMode": "compact_mode",
             "modelCardFooterAction": "model_card_footer_action",
+            "update_flag_strategy": "version_grouping",
         }
 
         updated = False
@@ -862,6 +926,23 @@ class SettingsManager:
             self.settings["civitai_api_key"] = env_api_key
             self._save_settings()
 
+        # LLM provider overrides
+        llm_env_map = {
+            "LLM_API_KEY": "llm_api_key",
+            "LLM_MODEL": "llm_model",
+            "LLM_API_BASE": "llm_api_base",
+            "LLM_PROVIDER": "llm_provider",
+        }
+        llm_changed = False
+        for env_var, settings_key in llm_env_map.items():
+            env_val = os.environ.get(env_var)
+            if env_val:
+                logger.info("Found %s environment variable", env_var)
+                self.settings[settings_key] = env_val
+                llm_changed = True
+        if llm_changed:
+            self._save_settings()
+
     def _default_settings_actions(self) -> List[Dict[str, Any]]:
         return [
             {
@@ -1509,8 +1590,12 @@ class SettingsManager:
             portable_switch_pending = True
             self._prepare_portable_switch(value)
         if key == "folder_paths" and isinstance(value, Mapping):
+            active_name = self.get_active_library_name()
+            self._validate_folder_paths(active_name, value)
             self._update_active_library_entry(folder_paths=value)  # type: ignore[arg-type]
         elif key == "extra_folder_paths" and isinstance(value, Mapping):
+            active_name = self.get_active_library_name()
+            self._validate_folder_paths(active_name, value)
             self._update_active_library_entry(extra_folder_paths=value)  # type: ignore[arg-type]
         elif key == "default_lora_root":
             self._update_active_library_entry(default_lora_root=str(value))
@@ -1557,7 +1642,7 @@ class SettingsManager:
         previous_dir = os.path.dirname(previous_path) or target_dir
 
         if os.path.abspath(previous_path) != os.path.abspath(target_path):
-            self._copy_model_cache_directory(previous_dir, target_dir)
+            self._migrate_settings_directory_content(previous_dir, target_dir)
             logger.info("Switching settings file to: %s", target_path)
 
         self._pending_portable_switch = {"other_path": other_path}
@@ -1592,46 +1677,52 @@ class SettingsManager:
         finally:
             self._pending_portable_switch = None
 
-    def _copy_model_cache_directory(self, source_dir: str, target_dir: str) -> None:
-        """Copy model_cache artifacts when switching storage locations."""
+    def _migrate_settings_directory_content(
+        self, source_dir: str, target_dir: str
+    ) -> None:
+        """Migrate settings directory subdirectories when switching storage locations.
+
+        Copies the canonical subdirectories (cache, backups, logs, stats, wildcards)
+        from the old settings directory to the new one. Legacy cache artifacts
+        (model_cache, recipe_cache, etc.) are migrated lazily by
+        ``resolve_cache_path_with_migration`` on first access.
+
+        Args:
+            source_dir: The previous settings directory path.
+            target_dir: The new settings directory path.
+        """
 
         if not source_dir or not target_dir:
             return
 
-        source_cache_dir = os.path.join(source_dir, "model_cache")
-        target_cache_dir = os.path.join(target_dir, "model_cache")
-        if os.path.isdir(source_cache_dir) and os.path.abspath(
-            source_cache_dir
-        ) != os.path.abspath(target_cache_dir):
-            try:
-                shutil.copytree(
-                    source_cache_dir,
-                    target_cache_dir,
-                    dirs_exist_ok=True,
-                    ignore=shutil.ignore_patterns("*.sqlite-shm", "*.sqlite-wal"),
-                )
-            except Exception as exc:
-                logger.warning(
-                    "Failed to copy model_cache directory from %s to %s: %s",
-                    source_cache_dir,
-                    target_cache_dir,
-                    exc,
-                )
+        def _copy_dir(name: str) -> None:
+            source = os.path.join(source_dir, name)
+            target = os.path.join(target_dir, name)
+            if os.path.isdir(source) and os.path.abspath(source) != os.path.abspath(
+                target
+            ):
+                try:
+                    shutil.copytree(
+                        source,
+                        target,
+                        dirs_exist_ok=True,
+                        ignore=shutil.ignore_patterns("*.sqlite-shm", "*.sqlite-wal"),
+                    )
+                except Exception as exc:
+                    logger.warning(
+                        "Failed to copy directory %s from %s to %s: %s",
+                        name,
+                        source,
+                        target,
+                        exc,
+                    )
 
-        source_cache_file = os.path.join(source_dir, "model_cache.sqlite")
-        target_cache_file = os.path.join(target_dir, "model_cache.sqlite")
-        if os.path.isfile(source_cache_file) and os.path.abspath(
-            source_cache_file
-        ) != os.path.abspath(target_cache_file):
-            try:
-                shutil.copy2(source_cache_file, target_cache_file)
-            except Exception as exc:
-                logger.warning(
-                    "Failed to copy model_cache.sqlite from %s to %s: %s",
-                    source_cache_file,
-                    target_cache_file,
-                    exc,
-                )
+        # Managed subdirectories under settings_dir
+        _copy_dir("cache")
+        _copy_dir("backups")
+        _copy_dir("logs")
+        _copy_dir("stats")
+        _copy_dir("wildcards")
 
     def _get_user_config_directory(self) -> str:
         """Return the user configuration directory, falling back to ~/.config."""
@@ -1758,6 +1849,9 @@ class SettingsManager:
                 if key in self.settings:
                     minimal[key] = copy.deepcopy(self.settings[key])
 
+            if self.settings.get("use_portable_settings"):
+                minimal["use_portable_settings"] = True
+
             if self._seed_template:
                 for key, value in self._seed_template.items():
                     minimal.setdefault(key, copy.deepcopy(value))

py/services/tag_update_service.py

@@ -36,9 +36,9 @@ class TagUpdateService:
             if isinstance(tag, str) and tag.strip():
                 # Convert all tags to lowercase to avoid case sensitivity issues on Windows
                 normalized = tag.strip().lower()
-                if normalized.lower() not in existing_lower:
+                if normalized not in existing_lower:
                     existing_tags.append(normalized)
-                    existing_lower.append(normalized.lower())
+                    existing_lower.append(normalized)
                     tags_added.append(normalized)
 
         metadata["tags"] = existing_tags

py/services/use_cases/bulk_metadata_refresh_use_case.py

@@ -3,6 +3,7 @@
 from __future__ import annotations
 
 import logging
+import time
 from typing import Any, Dict, List, Optional, Protocol, Sequence
 
 from ..metadata_sync_service import MetadataSyncService
@@ -50,6 +51,10 @@ class BulkMetadataRefreshUseCase:
             if not model.get("skip_metadata_refresh", False)
             and not self._is_in_skip_path(model.get("folder", ""), skip_paths)
             and (not model.get("civitai") or not model["civitai"].get("id"))
+            # Skip models downloaded from Hugging Face — they are not on
+            # CivitAI / CivArchive.  Users can still refresh them individually
+            # via the right-click context menu.
+            and not model.get("hf_url", "")
             and not (
                 # Skip models confirmed not on CivitAI when no need to retry
                 model.get("from_civitai") is False
@@ -62,26 +67,48 @@ class BulkMetadataRefreshUseCase:
         ]
 
         total_to_process = len(to_process)
+        initial_skipped = total_models - total_to_process  # models excluded from fetch queue
         processed = 0
         success = 0
+        skipped_count = initial_skipped
+        handled_count = initial_skipped
         needs_resort = False
+        start_time = time.monotonic()
+        failures: List[Dict[str, str]] = []
 
         self._service.scanner.reset_cancellation()
 
         async def emit(status: str, **extra: Any) -> None:
             if progress_callback is None:
                 return
-            payload = {"status": status, "total": total_to_process, "processed": processed, "success": success}
+            payload = {
+                "status": status,
+                "total": total_models,
+                "processed": processed,
+                "success": success,
+                "failure_count": len(failures),
+                "skipped_count": skipped_count,
+                "handled": handled_count,
+                "elapsed_seconds": int(time.monotonic() - start_time),
+            }
+            # Only include full failure details in terminal emits (completed,
+            # cancelled, rate_limited) to avoid serializing the list on every
+            # per-model progress update.
+            if failures and status in ("completed", "cancelled", "rate_limited"):
+                payload["failures"] = failures
             payload.update(extra)
             await progress_callback.on_progress(payload)
 
         await emit("started")
 
+        RATE_LIMIT_ABORT_THRESHOLD = 3
+        consecutive_rate_limits = 0
+
         for model in to_process:
             if self._service.scanner.is_cancelled():
                 self._logger.info("Bulk metadata refresh cancelled by user")
                 await emit("cancelled", processed=processed, success=success)
-                return {"success": False, "message": "Operation cancelled", "processed": processed, "updated": success, "total": total_models}
+                return {"success": False, "message": "Operation cancelled", "processed": processed, "updated": success, "total": total_models, "failures": failures, "failure_count": len(failures), "skipped_count": skipped_count, "elapsed_seconds": int(time.monotonic() - start_time)}
             try:
                 original_name = model.get("model_name")
 
@@ -99,33 +126,89 @@ class BulkMetadataRefreshUseCase:
                         if sha256:
                             model["sha256"] = sha256
                             model["hash_status"] = "completed"
+                            hash_status = "completed"
                         else:
                             self._logger.error(f"Failed to calculate hash for {file_path}")
+                            failures.append({"name": model.get("model_name", file_path or "Unknown"), "error": "Failed to calculate hash"})
                             processed += 1
+                            handled_count += 1
                             continue
                     else:
                         self._logger.warning(f"Scanner does not support lazy hash calculation for {file_path}")
+                        skipped_count += 1
                         processed += 1
+                        handled_count += 1
                         continue
 
                 # Skip models without valid hash
                 if not model.get("sha256"):
                     self._logger.warning(f"Skipping model without hash: {file_path}")
+                    skipped_count += 1
                     processed += 1
+                    handled_count += 1
                     continue
 
                 await MetadataManager.hydrate_model_data(model)
-                result, _ = await self._metadata_sync.fetch_and_update_model(
+
+                # hydrate_model_data replaces model with .metadata.json content,
+                # which may lack sha256. Restore from cache and persist the fix.
+                if not model.get("sha256"):
+                    model["sha256"] = sha256
+                    model["hash_status"] = model.get("hash_status", hash_status)
+                    data_to_save = model.copy()
+                    data_to_save.pop("folder", None)
+                    await MetadataManager.save_metadata(file_path, data_to_save)
+
+                result, error_msg = await self._metadata_sync.fetch_and_update_model(
                     sha256=model["sha256"],
                     file_path=model["file_path"],
                     model_data=model,
                     update_cache_func=self._service.scanner.update_single_model_cache,
                 )
+
+                if not result and error_msg and "Rate limited" in error_msg:
+                    consecutive_rate_limits += 1
+                else:
+                    consecutive_rate_limits = 0
+
+                if not result:
+                    current_name = model.get("model_name", file_path or "Unknown")
+                    failures.append({"name": current_name, "error": error_msg or "Unknown error"})
+                    self._logger.warning("Failed to fetch metadata for %s: %s", current_name, error_msg)
+
+                if consecutive_rate_limits >= RATE_LIMIT_ABORT_THRESHOLD:
+                    # The current model was attempted and failed due to rate limiting;
+                    # count it before aborting so the summary is consistent.
+                    processed += 1
+                    handled_count += 1
+                    self._logger.warning(
+                        "Bulk metadata refresh aborted: %d consecutive rate limits detected. "
+                        "Processed %d/%d models.",
+                        consecutive_rate_limits,
+                        processed,
+                        total_to_process,
+                    )
+                    await emit(
+                        "rate_limited",
+                    )
+                    return {
+                        "success": False,
+                        "message": f"Rate limit detected; {total_to_process - processed} models skipped",
+                        "processed": processed,
+                        "updated": success,
+                        "total": total_models,
+                        "failures": failures,
+                        "failure_count": len(failures),
+                        "skipped_count": skipped_count,
+                        "elapsed_seconds": int(time.monotonic() - start_time),
+                    }
+
                 if result:
                     success += 1
                     if original_name != model.get("model_name"):
                         needs_resort = True
                 processed += 1
+                handled_count += 1
                 await emit(
                     "processing",
                     processed=processed,
@@ -134,6 +217,9 @@ class BulkMetadataRefreshUseCase:
                 )
             except Exception as exc:  # pragma: no cover - logging path
                 processed += 1
+                handled_count += 1
+                current_name = model.get("model_name", model.get("file_path", "Unknown"))
+                failures.append({"name": current_name, "error": str(exc)})
                 self._logger.error(
                     "Error fetching CivitAI data for %s: %s",
                     model.get("file_path"),
@@ -150,7 +236,7 @@ class BulkMetadataRefreshUseCase:
             f"{success} of {processed} processed {self._service.model_type}s (total: {total_models})"
         )
 
-        return {"success": True, "message": message, "processed": processed, "updated": success, "total": total_models}
+        return {"success": True, "message": message, "processed": processed, "updated": success, "total": total_models, "failures": failures, "failure_count": len(failures), "skipped_count": skipped_count, "elapsed_seconds": int(time.monotonic() - start_time)}
 
     @staticmethod
     def _is_in_skip_path(folder: str, skip_paths: List[str]) -> bool:

py/services/wildcard_service.py

@@ -19,7 +19,7 @@ logger = logging.getLogger(__name__)
 _WILDCARD_PATTERN = re.compile(r"__([\w\s.\-+/*\\]+?)__")
 _OPTION_PATTERN = re.compile(r"{([^{}]*?)}")
 _TRIGGER_WORD_PATTERN = re.compile(r"^trigger_words\d+$")
-_WEIGHTED_OPTION_PATTERN = re.compile(r"^\s*([0-9.]+)::")
+_WEIGHTED_OPTION_PATTERN = re.compile(r"^\s*-?\d+(\.\d+)?::")
 _NUMERIC_PATTERN = re.compile(r"^-?\d+(\.\d+)?$")
 
 
@@ -390,7 +390,7 @@ class WildcardService:
     ) -> str | None:
         keyword = _normalize_wildcard_key(raw_key)
         if keyword in wildcard_dict:
-            return rng.choice(wildcard_dict[keyword])
+            return self._pick_weighted_or_plain(wildcard_dict[keyword], rng)
 
         if "*" in keyword:
             regex_pattern = keyword.replace("*", ".*").replace("+", r"\+")
@@ -400,7 +400,7 @@ class WildcardService:
                 if compiled.match(key):
                     aggregated.extend(values)
             if aggregated:
-                return rng.choice(aggregated)
+                return self._pick_weighted_or_plain(aggregated, rng)
 
         if "/" not in keyword:
             fallback_keyword = _normalize_wildcard_key(f"*/{keyword}")
@@ -409,6 +409,39 @@ class WildcardService:
 
         return None
 
+    def _pick_weighted_or_plain(
+        self, values: list[str], rng: random.Random
+    ) -> str:
+        """Pick a value from the list, respecting N::weight prefix if present.
+
+        When any value in the list uses the ``N::value`` weighted syntax with a
+        weight different from 1, the pick uses weighted random selection.  When
+        no such weighting is present, a plain ``rng.choice`` is used (preserving
+        backward compatibility for unweighted wildcard files).
+
+        In either case the ``N::`` prefix is always stripped from the returned
+        value, matching the behaviour of ``{...}`` option groups.
+        """
+        # Fast path: skip weighting logic entirely when no :: syntax exists
+        if not any("::" in v for v in values):
+            return rng.choice(values)
+
+        weighted_options: list[tuple[float, str]] = []
+        for value in values:
+            weight = 1.0
+            parts = value.split("::", 1)
+            if len(parts) == 2 and _is_numeric_string(parts[0].strip()):
+                weight = float(parts[0].strip())
+            weighted_options.append((weight, value))
+
+        any_weighted = any(w != 1.0 for w, _ in weighted_options)
+        if any_weighted:
+            picked = self._weighted_choice(weighted_options, rng)
+        else:
+            picked = rng.choice(values)
+
+        return self._strip_weight_prefix(picked)
+
 
 def is_trigger_words_input(name: str) -> bool:
     return bool(_TRIGGER_WORD_PATTERN.match(name))

py/utils/constants.py

@@ -12,6 +12,7 @@ NODE_TYPES = {
     "Lora Loader (LoraManager)": 1,
     "Lora Stacker (LoraManager)": 2,
     "WanVideo Lora Select (LoraManager)": 3,
+    "Create Hook LoRA (LoraManager)": 4,
 }
 
 # Default ComfyUI node color when bgcolor is null
@@ -31,6 +32,8 @@ PREVIEW_EXTENSIONS = [
     ".mp4",
     ".gif",
     ".webm",
+    ".avif",
+    ".jxl",
 ]
 
 # Card preview image width
@@ -41,10 +44,24 @@ EXAMPLE_IMAGE_WIDTH = 832
 
 # Supported media extensions for example downloads
 SUPPORTED_MEDIA_EXTENSIONS = {
-    "images": [".jpg", ".jpeg", ".png", ".webp", ".gif"],
+    "images": [".jpg", ".jpeg", ".png", ".webp", ".gif", ".avif", ".jxl"],
     "videos": [".mp4", ".webm"],
 }
 
+# Model weight file extensions recognised by scanners.
+# This is the union of all scanner extensions (lora, checkpoint, embedding).
+MODEL_FILE_EXTENSIONS = {
+    ".safetensors",
+    ".ckpt",
+    ".pt",
+    ".pt2",
+    ".bin",
+    ".pth",
+    ".pkl",
+    ".sft",
+    ".gguf",
+}
+
 # Valid sub-types for each scanner type
 VALID_LORA_SUB_TYPES = ["lora", "locon", "dora"]
 VALID_CHECKPOINT_SUB_TYPES = ["checkpoint", "diffusion_model"]
@@ -145,6 +162,8 @@ DIFFUSION_MODEL_BASE_MODELS = frozenset(
         "Qwen",
         "ZImageBase",
         "ZImageTurbo",
+        # Krea 2 — loaded via UNETLoader in ComfyUI
+        "Krea 2",
     ]
 )
 
@@ -208,8 +227,21 @@ SUPPORTED_DOWNLOAD_SKIP_BASE_MODELS = frozenset(
         "Wan Video 2.5 I2V",
         "Hunyuan Video",
         "Anima",
+        "ACE Audio",
+        "Boogu",
         "Ernie",
         "Ernie Turbo",
+        "Grok",
+        "HappyHorse",
+        "HiDream-O1",
+        "Ideogram 4.0",
+        "Krea 2",
+        "Lens",
+        "MAI",
         "Nucleus",
+        "Qwen 2",
+        "Upscaler",
+        "Wan Image 2.7",
+        "Wan Video 2.7",
     ]
 )

py/utils/example_images_download_manager.py

@@ -72,6 +72,7 @@ class _DownloadProgress(dict):
             refreshed_models=set(),
             failed_models=set(),
             reprocessed_models=set(),
+            rate_limited_models=set(),
         )
 
     def snapshot(self) -> dict:
@@ -82,6 +83,7 @@ class _DownloadProgress(dict):
         snapshot["refreshed_models"] = list(self["refreshed_models"])
         snapshot["failed_models"] = list(self["failed_models"])
         snapshot["reprocessed_models"] = list(self.get("reprocessed_models", set()))
+        snapshot["rate_limited_models"] = list(self.get("rate_limited_models", set()))
         return snapshot
 
 
@@ -153,13 +155,15 @@ class DownloadManager:
         # Step 3: Load progress file (I/O operation, done outside lock)
         processed_models = set()
         failed_models = set()
+        rate_limited_models = set()
         
         try:
-            progress_file, processed_models, failed_models = await self._load_progress_file(output_dir)
+            progress_file, processed_models, failed_models, rate_limited_models = await self._load_progress_file(output_dir)
             logger.debug(
-                "Loaded previous progress, %s models already processed, %s models marked as failed",
+                "Loaded previous progress, %s models already processed, %s models marked as failed, %s models rate-limited",
                 len(processed_models),
                 len(failed_models),
+                len(rate_limited_models),
             )
         except Exception as e:
             logger.error(f"Failed to load progress file: {e}")
@@ -175,6 +179,7 @@ class DownloadManager:
                 self._progress.reset()
                 self._progress["processed_models"] = processed_models
                 self._progress["failed_models"] = failed_models
+                self._progress["rate_limited_models"] = rate_limited_models
                 self._stop_requested = False
                 self._progress["status"] = "running"
                 self._progress["start_time"] = time.time()
@@ -242,8 +247,8 @@ class DownloadManager:
             "status": self._progress.snapshot(),
         }
 
-    async def _load_progress_file(self, output_dir: str) -> tuple[str, set, set]:
-        """Load progress file from disk. Returns (progress_file_path, processed_models, failed_models).
+    async def _load_progress_file(self, output_dir: str) -> tuple[str, set, set, set]:
+        """Load progress file from disk. Returns (progress_file_path, processed_models, failed_models, rate_limited_models).
 
         This is a separate async method to allow running in executor to avoid blocking event loop.
         """
@@ -252,8 +257,12 @@ class DownloadManager:
             None, self._load_progress_file_sync, output_dir
         )
 
-    def _load_progress_file_sync(self, output_dir: str) -> tuple[str, set, set]:
-        """Synchronous implementation of progress file loading."""
+    def _load_progress_file_sync(self, output_dir: str) -> tuple[str, set, set, set]:
+        """Synchronous implementation of progress file loading.
+
+        Returns:
+            tuple: (progress_file_path, processed_models, failed_models, rate_limited_models)
+        """
         progress_file = os.path.join(output_dir, ".download_progress.json")
         progress_source = progress_file
 
@@ -289,6 +298,7 @@ class DownloadManager:
 
         processed_models = set()
         failed_models = set()
+        rate_limited_models = set()
 
         if os.path.exists(progress_source):
             try:
@@ -296,11 +306,11 @@ class DownloadManager:
                     saved_progress = json.load(f)
                     processed_models = set(saved_progress.get("processed_models", []))
                     failed_models = set(saved_progress.get("failed_models", []))
+                    rate_limited_models = set(saved_progress.get("rate_limited_models", []))
             except Exception:
-                # Return empty sets on error
                 pass
 
-        return progress_file, processed_models, failed_models
+        return progress_file, processed_models, failed_models, rate_limited_models
 
     def _load_progress_sets_sync(self, progress_file: str) -> tuple[set, set]:
         """Load only the processed and failed model sets from progress file.
@@ -732,11 +742,13 @@ class DownloadManager:
                     success,
                     is_stale,
                     failed_images,
+                    rate_limited_images,
                 ) = await ExampleImagesProcessor.download_model_images_with_tracking(
                     model_hash, model_name, images, model_dir, optimize, downloader
                 )
 
                 failed_urls: Set[str] = set(failed_images)
+                rate_limited_urls: Set[str] = set(rate_limited_images)
 
                 # If metadata is stale, try to refresh it
                 if is_stale and model_hash not in self._progress["refreshed_models"]:
@@ -760,6 +772,7 @@ class DownloadManager:
                             success,
                             _,
                             additional_failed,
+                            additional_rate_limited,
                         ) = await ExampleImagesProcessor.download_model_images_with_tracking(
                             model_hash,
                             model_name,
@@ -770,29 +783,50 @@ class DownloadManager:
                         )
 
                         failed_urls.update(additional_failed)
+                        rate_limited_urls.update(additional_rate_limited)
 
                     self._progress["refreshed_models"].add(model_hash)
 
-                if failed_urls:
+                # Separate permanent failures from rate-limited ones
+                permanent_failures = failed_urls - rate_limited_urls
+
+                if permanent_failures:
                     await self._remove_failed_images_from_metadata(
                         model_hash,
                         model_name,
                         model_dir,
-                        failed_urls,
+                        permanent_failures,
                         scanner,
                     )
 
-                if failed_urls:
+                if rate_limited_urls:
+                    self._progress["rate_limited_models"].add(model_hash)
+                    logger.warning(
+                        "%d example images for %s are rate-limited (429), will retry next time",
+                        len(rate_limited_urls),
+                        model_name,
+                    )
+                    # Clear failed_models so non-force runs can retry
+                    if force and model_hash in self._progress["failed_models"]:
+                        self._progress["failed_models"].discard(model_hash)
+                        logger.info(
+                            f"Removed {model_name} from failed_models after force retry with rate-limited images"
+                        )
+
+                if rate_limited_urls:
+                    # Don't mark as failed or fully processed — rate-limited
+                    # images will be retried next time.
+                    pass
+                elif permanent_failures:
                     self._progress["failed_models"].add(model_hash)
                     self._progress["processed_models"].add(model_hash)
                     logger.info(
                         "Removed %s failed example images for %s",
-                        len(failed_urls),
+                        len(permanent_failures),
                         model_name,
                     )
                 elif success:
                     self._progress["processed_models"].add(model_hash)
-                    # Remove from failed_models if force mode enabled and model was previously failed
                     if force and model_hash in self._progress["failed_models"]:
                         self._progress["failed_models"].discard(model_hash)
                         logger.info(
@@ -850,6 +884,7 @@ class DownloadManager:
                 "processed_models": list(self._progress["processed_models"]),
                 "refreshed_models": list(self._progress["refreshed_models"]),
                 "failed_models": list(self._progress["failed_models"]),
+                "rate_limited_models": list(self._progress.get("rate_limited_models", set())),
                 "completed": self._progress["completed"],
                 "total": self._progress["total"],
                 "last_update": time.time(),
@@ -1155,11 +1190,13 @@ class DownloadManager:
                     success,
                     is_stale,
                     failed_images,
+                    rate_limited_images,
                 ) = await ExampleImagesProcessor.download_model_images_with_tracking(
                     model_hash, model_name, images, model_dir, optimize, downloader
                 )
 
                 failed_urls: Set[str] = set(failed_images)
+                rate_limited_urls: Set[str] = set(rate_limited_images)
 
                 # If metadata is stale, try to refresh it
                 if is_stale and model_hash not in self._progress["refreshed_models"]:
@@ -1183,6 +1220,7 @@ class DownloadManager:
                             success,
                             _,
                             additional_failed_images,
+                            additional_rate_limited,
                         ) = await ExampleImagesProcessor.download_model_images_with_tracking(
                             model_hash,
                             model_name,
@@ -1192,21 +1230,35 @@ class DownloadManager:
                             downloader,
                         )
 
-                        # Combine failed images from both attempts
                         failed_urls.update(additional_failed_images)
+                        rate_limited_urls.update(additional_rate_limited)
 
                     self._progress["refreshed_models"].add(model_hash)
 
-                # For forced downloads, remove failed images from metadata
-                if failed_urls:
+                # Separate permanent failures from rate-limited ones
+                permanent_failures = failed_urls - rate_limited_urls
+
+                # Only remove permanently failed images from metadata
+                if permanent_failures:
                     await self._remove_failed_images_from_metadata(
-                        model_hash, model_name, model_dir, failed_urls, scanner
+                        model_hash, model_name, model_dir, permanent_failures, scanner
                     )
 
-                # Mark as processed
-                if (
-                    success or failed_urls
-                ):  # Mark as processed if we successfully downloaded some images or removed failed ones
+                if rate_limited_urls:
+                    self._progress["rate_limited_models"].add(model_hash)
+                    logger.warning(
+                        "%d example images for %s are rate-limited (429), will retry next time",
+                        len(rate_limited_urls),
+                        model_name,
+                    )
+
+                # Mark as processed only when no rate-limited images remain
+                if rate_limited_urls:
+                    pass
+                elif permanent_failures:
+                    self._progress["processed_models"].add(model_hash)
+                    self._progress["failed_models"].add(model_hash)
+                elif success:
                     self._progress["processed_models"].add(model_hash)
 
                 return True  # Return True to indicate a remote download happened
@@ -1229,15 +1281,20 @@ class DownloadManager:
         model_dir: str,
         failed_images: Iterable[str],
         scanner,
+        error_type: str = "not_found",
     ) -> None:
-        """Mark failed images in model metadata so they won't be retried."""
+        """Mark failed images in model metadata so they won't be retried.
+
+        Args:
+            error_type: Reason string stored in the image's ``downloadError`` field
+                        (default ``"not_found"``).
+        """
 
         failed_set: Set[str] = {url for url in failed_images if url}
         if not failed_set:
             return
 
         try:
-            # Get current model data
             model_data = await MetadataUpdater.get_updated_model(model_hash, scanner)
             if not model_data:
                 logger.warning(
@@ -1268,7 +1325,7 @@ class DownloadManager:
                     continue
 
                 image["downloadFailed"] = True
-                image.setdefault("downloadError", "not_found")
+                image.setdefault("downloadError", error_type)
                 logger.debug(
                     "Marked example image %s for %s as failed due to missing remote asset",
                     image_url,

py/utils/example_images_metadata.py

@@ -475,13 +475,19 @@ class MetadataUpdater:
             return False
 
         model_folder = get_model_folder(model_hash)
-        if not model_folder:
+        if not model_folder or not os.path.isdir(model_folder):
             return False
 
         civitai = getattr(metadata, "civitai", None)
         if not isinstance(civitai, dict):
             return False
 
+        # Read the directory listing once so every image entry reuses it.
+        try:
+            dir_entries = os.listdir(model_folder)
+        except OSError:
+            dir_entries = []
+
         has_changes = False
 
         custom_images = civitai.get("customImages")
@@ -493,24 +499,15 @@ class MetadataUpdater:
                 if not img_id:
                     continue
 
-                if not os.path.isdir(model_folder):
+                prefix = f"custom_{img_id}"
+                found = any(
+                    f.startswith(prefix) and os.path.isfile(
+                        os.path.join(model_folder, f)
+                    )
+                    for f in dir_entries
+                )
+                if not found:
                     stale.append(idx)
-                else:
-                    found = False
-                    try:
-                        prefix = f"custom_{img_id}"
-                        for fname in os.listdir(model_folder):
-                            if fname.startswith(prefix) and os.path.isfile(
-                                os.path.join(model_folder, fname)
-                            ):
-                                found = True
-                                break
-                    except OSError:
-                        stale.append(idx)
-                        continue
-
-                    if not found:
-                        stale.append(idx)
 
             if stale:
                 for idx in reversed(stale):
@@ -532,22 +529,9 @@ class MetadataUpdater:
                     # is gone.
                     continue
 
-                if not os.path.isdir(model_folder):
+                prefix = f"image_{idx}."
+                if not any(f.startswith(prefix) for f in dir_entries):
                     stale.append(idx)
-                else:
-                    found = False
-                    try:
-                        prefix = f"image_{idx}."
-                        for fname in os.listdir(model_folder):
-                            if fname.startswith(prefix):
-                                found = True
-                                break
-                    except OSError:
-                        stale.append(idx)
-                        continue
-
-                    if not found:
-                        stale.append(idx)
 
             if stale:
                 for idx in reversed(stale):

py/utils/example_images_migration.py

@@ -3,9 +3,16 @@ import logging
 import os
 import re
 import json
+import shutil
 from ..services.settings_manager import get_settings_manager
 from ..services.service_registry import ServiceRegistry
-from ..utils.example_images_paths import iter_library_roots
+from ..utils.example_images_paths import (
+    get_example_images_root,
+    is_hash_folder,
+    iter_library_roots,
+    uses_library_scoped_folders,
+    _library_folder_has_only_hash_dirs,
+)
 from ..utils.metadata_manager import MetadataManager
 from ..utils.example_images_processor import ExampleImagesProcessor
 from ..utils.constants import SUPPORTED_MEDIA_EXTENSIONS
@@ -36,6 +43,90 @@ settings = _SettingsProxy()
 class ExampleImagesMigration:
     """Handles migrations for example images naming conventions"""
     
+    @staticmethod
+    def _consolidate_library_folders():
+        """Move hash folders from library-named subdirectories back to root.
+
+        When a user switches from multi-library mode back to single-library
+        mode, example images previously stored under e.g.
+        ``<root>/default/<hash>/`` need to be moved back to
+        ``<root>/<hash>/``.  Running this once at startup removes the need
+        for ``get_model_folder()`` to perform directory scans on every
+        request.
+        """
+        if uses_library_scoped_folders():
+            return
+
+        root = get_example_images_root()
+        if not root or not os.path.isdir(root):
+            return
+
+        moved: list[str] = []
+        cleaned: list[str] = []
+
+        try:
+            for entry in os.listdir(root):
+                # Fast regex checks first — no filesystem I/O.
+                if is_hash_folder(entry) or entry == "_deleted":
+                    continue
+
+                entry_path = os.path.join(root, entry)
+                if not os.path.isdir(entry_path):
+                    continue
+                if not _library_folder_has_only_hash_dirs(entry_path):
+                    continue
+
+                try:
+                    for hash_entry in os.listdir(entry_path):
+                        hash_path = os.path.join(entry_path, hash_entry)
+                        if not os.path.isdir(hash_path) or not is_hash_folder(hash_entry):
+                            continue
+                        target = os.path.join(root, hash_entry)
+                        if not os.path.exists(target):
+                            try:
+                                shutil.move(hash_path, target)
+                                moved.append(hash_entry)
+                            except (OSError, shutil.Error) as exc:
+                                logger.error(
+                                    "Failed to move '%s' → '%s': %s",
+                                    hash_path, target, exc,
+                                )
+                except OSError as exc:
+                    logger.error(
+                        "Failed to list library subdirectory '%s': %s",
+                        entry_path, exc,
+                    )
+
+                try:
+                    remaining = os.listdir(entry_path)
+                except OSError:
+                    remaining = []
+                if not remaining:
+                    try:
+                        os.rmdir(entry_path)
+                        cleaned.append(entry)
+                    except OSError as exc:
+                        logger.debug(
+                            "Could not remove empty library dir '%s': %s",
+                            entry_path, exc,
+                        )
+        except OSError as exc:
+            logger.error(
+                "Failed to list example images root during consolidation: %s",
+                exc,
+            )
+
+        if moved:
+            logger.info(
+                "Consolidated %d example image folder(s) to root",
+                len(moved),
+            )
+        if cleaned:
+            logger.info(
+                "Removed %d empty library directories",
+                len(cleaned),
+            )
+
     @staticmethod
     async def check_and_run_migrations():
         """Check if migrations are needed and run them in background"""
@@ -44,6 +135,10 @@ class ExampleImagesMigration:
             logger.debug("No example images path configured or path doesn't exist, skipping migrations")
             return
 
+        # Run library-to-root consolidation once at startup so the hot
+        # path (get_model_folder) stays a pure-path computation.
+        ExampleImagesMigration._consolidate_library_folders()
+
         for library_name, library_path in iter_library_roots():
             if not library_path or not os.path.exists(library_path):
                 continue

py/utils/example_images_paths.py

@@ -12,6 +12,18 @@ from ..services.settings_manager import get_settings_manager
 
 _HEX_PATTERN = re.compile(r"[a-fA-F0-9]{64}")
 
+# Filesystem/metadata files that are never created by the example images system
+# and are safe to ignore during validation. The cleanup service only operates on
+# directories, so these files pose no data-loss risk.
+_SAFE_FILENAMES: frozenset[str] = frozenset({
+    ".DS_Store",       # macOS folder metadata
+    "Thumbs.db",       # Windows thumbnail cache
+    "desktop.ini",     # Windows folder customization
+    ".localized",      # macOS folder name localization
+    ".gitkeep",        # Placeholder to keep empty dirs in git
+    ".gitignore",      # Git ignore rules
+})
+
 logger = logging.getLogger(__name__)
 
 
@@ -71,7 +83,12 @@ def ensure_library_root_exists(library_name: Optional[str] = None) -> str:
 
 
 def get_model_folder(model_hash: str, library_name: Optional[str] = None) -> str:
-    """Return the folder path for a model's example images."""
+    """Return the folder path for a model's example images.
+
+    Multi-library ↔ single-library consolidation is handled once at startup by
+    ``ExampleImagesMigration._consolidate_library_folders`` — this function is a
+    pure path computation on the hot path (no directory scans).
+    """
 
     if not model_hash:
         return ""
@@ -180,6 +197,22 @@ def is_hash_folder(name: str) -> bool:
     return bool(_HEX_PATTERN.fullmatch(name or ""))
 
 
+def _is_safe_ignorable_entry(item: str, item_path: str) -> bool:
+    """Return True if *item* is a harmless system/hidden file we can skip.
+
+    These files are never created by the example images system and are safe to
+    ignore because the cleanup/delete operations only act on **directories**,
+    never on individual files (other than ``.download_progress.json``).
+    """
+    if item in _SAFE_FILENAMES:
+        return True
+    # Hide Unix hidden files (dotfiles) that are regular files,
+    # since the cleanup system never deletes or moves files.
+    if item.startswith(".") and os.path.isfile(item_path):
+        return True
+    return False
+
+
 def is_valid_example_images_root(folder_path: str) -> bool:
     """Check whether a folder looks like a dedicated example images root."""
 
@@ -190,9 +223,16 @@ def is_valid_example_images_root(folder_path: str) -> bool:
 
     for item in items:
         item_path = os.path.join(folder_path, item)
+
+        # .download_progress.json is an expected metadata file — check before
+        # the generic dotfile rule so it stays explicitly documented.
         if item == ".download_progress.json" and os.path.isfile(item_path):
             continue
 
+        # Skip harmless system/hidden files — cleanup only touches directories
+        if _is_safe_ignorable_entry(item, item_path):
+            continue
+
         if os.path.isdir(item_path):
             if is_hash_folder(item):
                 continue
@@ -211,6 +251,41 @@ def is_valid_example_images_root(folder_path: str) -> bool:
     return True
 
 
+def find_non_compliant_items_in_example_images_root(folder_path: str) -> list[str]:
+    """Return the names of items that prevent *folder_path* from being a valid
+    example images root, or an empty list if the folder is valid.
+
+    This mirrors ``is_valid_example_images_root`` but **returns** the offending
+    names instead of a boolean, so callers can produce actionable error messages.
+    """
+    try:
+        items = os.listdir(folder_path)
+    except OSError as exc:
+        return [f"<cannot list directory: {exc}>"]
+
+    offending: list[str] = []
+
+    for item in items:
+        item_path = os.path.join(folder_path, item)
+
+        # Same skip rules as is_valid_example_images_root
+        if item == ".download_progress.json" and os.path.isfile(item_path):
+            continue
+        if _is_safe_ignorable_entry(item, item_path):
+            continue
+        if os.path.isdir(item_path):
+            if is_hash_folder(item):
+                continue
+            if item == "_deleted":
+                continue
+            if _library_folder_has_only_hash_dirs(item_path):
+                continue
+
+        offending.append(item)
+
+    return offending
+
+
 def _library_folder_has_only_hash_dirs(path: str) -> bool:
     """Return True when a library subfolder only contains hash folders or metadata files."""
 

py/utils/example_images_processor.py

@@ -1,3 +1,4 @@
+import asyncio
 import logging
 import os
 import re
@@ -62,6 +63,10 @@ class ExampleImagesProcessor:
                 return '.gif'
             elif content.startswith(b'RIFF') and b'WEBP' in content[:12]:
                 return '.webp'
+            elif len(content) >= 12 and content[4:8] == b'ftyp' and b'avif' in content[8:24]:
+                return '.avif'
+            elif content.startswith(b'\x00\x00\x00\x0cJXL \x0d\x0a\x87\x0a'):
+                return '.jxl'
             elif content.startswith(b'\x00\x00\x00\x18ftypmp4') or content.startswith(b'\x00\x00\x00\x20ftypmp4'):
                 return '.mp4'
             elif content.startswith(b'\x1A\x45\xDF\xA3'):
@@ -75,6 +80,8 @@ class ExampleImagesProcessor:
                 'image/png': '.png',
                 'image/gif': '.gif',
                 'image/webp': '.webp',
+                'image/avif': '.avif',
+                'image/jxl': '.jxl',
                 'video/mp4': '.mp4',
                 'video/webm': '.webm',
                 'video/quicktime': '.mov'
@@ -188,16 +195,22 @@ class ExampleImagesProcessor:
         
         return model_success, False  # (success, is_metadata_stale)
     
+    @staticmethod
+    def _extract_retry_after(error_message: str) -> int:
+        if not error_message:
+            return 60
+        match = re.search(r"retry after (\d+)s", str(error_message))
+        if match:
+            return max(1, int(match.group(1)))
+        return 60
+
     @staticmethod
     async def download_model_images_with_tracking(model_hash, model_name, model_images, model_dir, optimize, downloader):
-        """Download images for a single model with tracking of failed image URLs
-        
-        Returns:
-            tuple: (success, is_stale_metadata, failed_images) - whether download was successful, whether metadata is stale, list of failed image URLs
-        """
         model_success = True
         failed_images = []
-        
+        rate_limited_images = []
+        any_successful_download = False
+
         for i, image in enumerate(model_images):
             image_url = image.get('url')
             if not image_url:
@@ -215,64 +228,110 @@ class ExampleImagesProcessor:
             original_url = image_url
             if optimize and 'civitai.com' in image_url:
                 image_url = ExampleImagesProcessor.get_civitai_optimized_url(image_url)
-            
-            # Download the file first to determine the actual file type
-            try:
-                logger.debug(f"Downloading media file {i} for {model_name}")
-                
-                # Download using the unified downloader with headers
-                success, content, headers = await downloader.download_to_memory(
+
+            async def _attempt_download() -> tuple:
+                logger.debug("Downloading media file %s for %s", i, model_name)
+                return await downloader.download_to_memory(
                     image_url,
-                    use_auth=False,  # Example images don't need auth
-                    return_headers=True
+                    use_auth=False,
+                    return_headers=True,
                 )
-                
+
+            try:
+                success, content, headers = await _attempt_download()
+
                 if success:
-                    # Determine file extension from content or headers
                     media_ext = ExampleImagesProcessor._get_file_extension_from_content_or_headers(
                         content, headers, original_url, image.get("type")
                     )
-                    
-                    # Check if the detected file type is supported
                     is_image = media_ext in SUPPORTED_MEDIA_EXTENSIONS['images']
                     is_video = media_ext in SUPPORTED_MEDIA_EXTENSIONS['videos']
-                    
+
                     if not (is_image or is_video):
-                        logger.debug(f"Skipping unsupported file type: {media_ext}")
+                        logger.debug("Skipping unsupported file type: %s", media_ext)
                         continue
-                    
-                    # Use 0-based indexing with the detected extension
+
                     save_filename = f"image_{i}{media_ext}"
                     save_path = os.path.join(model_dir, save_filename)
-                    
-                    # Check if already downloaded
+
                     if os.path.exists(save_path):
-                        logger.debug(f"File already exists: {save_path}")
+                        logger.debug("File already exists: %s", save_path)
                         continue
-                    
-                    # Save the file
+
                     with open(save_path, 'wb') as f:
                         f.write(content)
-                    
+                    any_successful_download = True
+
                 elif ExampleImagesProcessor._is_not_found_error(content):
                     error_msg = f"Failed to download file: {image_url}, status code: 404 - Model metadata might be stale"
                     logger.warning(error_msg)
-                    model_success = False  # Mark the model as failed due to 404 error
-                    failed_images.append(image_url)  # Track failed URL
-                    # Return early to trigger metadata refresh attempt
-                    return False, True, failed_images  # (success, is_metadata_stale, failed_images)
+                    model_success = False
+                    failed_images.append(image_url)
+                    return False, True, failed_images, rate_limited_images
+
+                elif "Rate limited (429)" in str(content):
+                    max_attempts = 3
+                    for attempt in range(1, max_attempts + 1):
+                        wait = ExampleImagesProcessor._extract_retry_after(str(content)) * (2 ** (attempt - 1))
+                        logger.warning(
+                            "Rate limited (429) for %s, retry %d/%d after %ds",
+                            image_url, attempt, max_attempts, wait,
+                        )
+                        await asyncio.sleep(wait)
+
+                        success, content, headers = await _attempt_download()
+                        if success:
+                            media_ext = ExampleImagesProcessor._get_file_extension_from_content_or_headers(
+                                content, headers, original_url, image.get("type")
+                            )
+                            is_image = media_ext in SUPPORTED_MEDIA_EXTENSIONS['images']
+                            is_video = media_ext in SUPPORTED_MEDIA_EXTENSIONS['videos']
+
+                            if not (is_image or is_video):
+                                logger.debug("Skipping unsupported file type: %s", media_ext)
+                                break
+
+                            save_filename = f"image_{i}{media_ext}"
+                            save_path = os.path.join(model_dir, save_filename)
+                            if os.path.exists(save_path):
+                                logger.debug("File already exists: %s", save_path)
+                                break
+
+                            with open(save_path, 'wb') as f:
+                                f.write(content)
+                            any_successful_download = True
+                            break
+                        elif "Rate limited (429)" in str(content):
+                            continue
+                        elif ExampleImagesProcessor._is_not_found_error(content):
+                            logger.warning("Failed to download file: %s, status code: 404", image_url)
+                            model_success = False
+                            failed_images.append(image_url)
+                            break
+                        else:
+                            logger.warning("Failed to download file: %s, error: %s", image_url, content)
+                            model_success = False
+                            failed_images.append(image_url)
+                            break
+                    else:
+                        logger.warning(
+                            "Giving up on %s after %d retries due to rate limiting",
+                            image_url, max_attempts,
+                        )
+                        rate_limited_images.append(image_url)
+                        model_success = False
                 else:
                     error_msg = f"Failed to download file: {image_url}, error: {content}"
                     logger.warning(error_msg)
-                    model_success = False  # Mark the model as failed
-                    failed_images.append(image_url)  # Track failed URL
+                    model_success = False
+                    failed_images.append(image_url)
             except Exception as e:
                 error_msg = f"Error downloading file {image_url}: {str(e)}"
                 logger.error(error_msg)
-                model_success = False  # Mark the model as failed
-                failed_images.append(image_url)  # Track failed URL
-        
-        return model_success, False, failed_images  # (success, is_metadata_stale, failed_images)
+                model_success = False
+                failed_images.append(image_url)
+
+        return any_successful_download or model_success, False, failed_images, rate_limited_images
     
     @staticmethod
     async def process_local_examples(model_file_path, model_file_name, model_name, model_dir, optimize):

py/utils/exif_utils.py

@@ -1,17 +1,125 @@
 import json
 import logging
 import os
+import struct
 from io import BytesIO
 from typing import Any, Optional
 
 import piexif
 from PIL import Image, PngImagePlugin
 
+try:
+    import brotli
+    _BROTLI_AVAILABLE = True
+except ImportError:
+    brotli = None
+    _BROTLI_AVAILABLE = False
+
 logger = logging.getLogger(__name__)
 
 class ExifUtils:
     """Utility functions for working with EXIF data in images"""
 
+    @staticmethod
+    def _parse_isobmff_boxes(data: bytes, offset: int = 0) -> list[dict]:
+        boxes = []
+        while offset + 8 <= len(data):
+            size = struct.unpack('>I', data[offset:offset + 4])[0]
+            box_type = data[offset + 4:offset + 8]
+            if size == 0:
+                break
+            if size < 8 or offset + size > len(data):
+                break
+            box_data = data[offset + 8:offset + size]
+            boxes.append({'type': box_type, 'data': box_data, 'size': size})
+            offset += size
+        return boxes
+
+    @staticmethod
+    def _is_jxl_container(data: bytes) -> bool:
+        if len(data) < 32:
+            return False
+        return (
+            struct.unpack('>I', data[:4])[0] == 12
+            and data[4:8] == b'JXL '
+            and data[8:12] == bytes([0x0d, 0x0a, 0x87, 0x0a])
+            and struct.unpack('>I', data[12:16])[0] >= 16
+            and data[16:20] == b'ftyp'
+            and data[20:24] == b'jxl '
+        )
+
+    @staticmethod
+    def _is_avif_container(data: bytes) -> bool:
+        if len(data) < 16:
+            return False
+        for box in ExifUtils._parse_isobmff_boxes(data):
+            if box['type'] == b'ftyp' and b'avif' in box['data']:
+                return True
+        return False
+
+    # Max decompressed size for brotli metadata (2 MB)
+    _BROTLI_MAX_DECOMPRESSED = 2 * 1024 * 1024
+
+    @staticmethod
+    def _extract_isobmff_brotli(image_path: str) -> Optional[dict]:
+        try:
+            with open(image_path, 'rb') as f:
+                data = f.read()
+        except Exception:
+            return None
+
+        if ExifUtils._is_jxl_container(data):
+            boxes = ExifUtils._parse_isobmff_boxes(data, offset=12)
+        elif ExifUtils._is_avif_container(data):
+            boxes = ExifUtils._parse_isobmff_boxes(data)
+        else:
+            return None
+
+        brob = None
+        for box in boxes:
+            if box['type'] == b'brob':
+                brob = box
+                break
+        if brob is None:
+            return None
+
+        payload = brob['data']
+        if payload[:4] != b'comf':
+            return None
+        compressed = payload[4:]
+
+        if _BROTLI_AVAILABLE:
+            try:
+                decompressed = brotli.decompress(compressed)
+                if len(decompressed) > ExifUtils._BROTLI_MAX_DECOMPRESSED:
+                    logger.warning(
+                        "Brotli metadata too large (%d bytes, max %d), ignoring",
+                        len(decompressed),
+                        ExifUtils._BROTLI_MAX_DECOMPRESSED,
+                    )
+                    decompressed = None
+            except Exception:
+                decompressed = None
+        else:
+            decompressed = None
+
+        raw = decompressed if decompressed is not None else compressed
+        try:
+            meta = json.loads(raw.decode('utf-8'))
+        except Exception:
+            return None
+
+        result = {"parameters": None, "prompt": None, "workflow": None, "comment": None}
+        if isinstance(meta.get("prompt"), (dict, list)):
+            result["prompt"] = json.dumps(meta["prompt"])
+        elif isinstance(meta.get("prompt"), str):
+            result["prompt"] = meta["prompt"]
+        if isinstance(meta.get("workflow"), (dict, list)):
+            result["workflow"] = json.dumps(meta["workflow"])
+        elif isinstance(meta.get("workflow"), str):
+            result["workflow"] = meta["workflow"]
+        return result
+
     @staticmethod
     def _decode_user_comment(user_comment: Any) -> Optional[str]:
         if user_comment is None:
@@ -43,6 +151,12 @@ class ExifUtils:
             "comment": None,
         }
 
+        ext = os.path.splitext(image_path)[1].lower()
+        if ext in ('.avif', '.jxl'):
+            brotli_meta = ExifUtils._extract_isobmff_brotli(image_path)
+            if brotli_meta:
+                return brotli_meta
+
         with Image.open(image_path) as img:
             info = getattr(img, "info", {}) or {}
 
@@ -149,7 +263,6 @@ class ExifUtils:
             Optional[str]: Extracted metadata or None if not found
         """
         try:
-            # Skip for video files
             if image_path:
                 ext = os.path.splitext(image_path)[1].lower()
                 if ext in ['.mp4', '.webm']:
@@ -177,10 +290,9 @@ class ExifUtils:
             str: Path to the updated image
         """
         try:
-            # Skip for video files
             if image_path:
                 ext = os.path.splitext(image_path)[1].lower()
-                if ext in ['.mp4', '.webm']:
+                if ext in ['.mp4', '.webm', '.avif', '.jxl']:
                     return image_path
 
             metadata_fields = ExifUtils._load_structured_metadata(image_path)
@@ -212,10 +324,9 @@ class ExifUtils:
     def append_recipe_metadata(image_path, recipe_data) -> str:
         """Append recipe metadata to an image's EXIF data"""
         try:
-            # Skip for video files
             if image_path:
                 ext = os.path.splitext(image_path)[1].lower()
-                if ext in ['.mp4', '.webm']:
+                if ext in ['.mp4', '.webm', '.avif', '.jxl']:
                     return image_path
 
             # First, extract existing metadata
@@ -327,10 +438,9 @@ class ExifUtils:
             Tuple of (optimized_image_data, extension)
         """
         try:
-            # Skip for video files early if it's a file path
             if isinstance(image_data, str) and os.path.exists(image_data):
                 ext = os.path.splitext(image_data)[1].lower()
-                if ext in ['.mp4', '.webm']:
+                if ext in ['.mp4', '.webm', '.avif', '.jxl']:
                     try:
                         with open(image_data, 'rb') as f:
                             return f.read(), ext

py/utils/file_utils.py

@@ -34,12 +34,26 @@ def _get_hash_chunk_size_bytes() -> int:
 
 
 async def calculate_sha256(file_path: str) -> str:
-    """Calculate SHA256 hash of a file (full file content)."""
+    """Calculate SHA256 hash of a file (full file content).
+
+    Uses ``posix_fadvise`` with ``POSIX_FADV_DONTNEED`` to avoid polluting the OS page
+    cache — critical on WSL where cached file pages live inside the VM and are not
+    accounted for in guest ``used`` memory, causing VmmemWSL to balloon.
+
+    On Windows/macOS where ``posix_fadvise`` is not available the hint is silently
+    skipped.
+    """
     sha256_hash = hashlib.sha256()
     chunk_size = _get_hash_chunk_size_bytes()
     with open(file_path, "rb") as f:
+        fd = f.fileno()
         for byte_block in iter(lambda: f.read(chunk_size), b""):
             sha256_hash.update(byte_block)
+        # Evict pages after reading so the data doesn't linger in the kernel page
+        # cache — on WSL this otherwise appears as unreclaimable VmmemWSL growth.
+        # Guard against platforms (Windows, macOS) that lack posix_fadvise.
+        if hasattr(os, "posix_fadvise") and hasattr(os, "POSIX_FADV_DONTNEED"):
+            os.posix_fadvise(fd, 0, 0, os.POSIX_FADV_DONTNEED)
     return sha256_hash.hexdigest()