Drop the SequenceMatcher-based fuzzy_match fallback that froze the server
when FTS returned empty results. FTS now returns empty set for zero results
(no fallback), and when the index is not yet ready, search returns empty
rather than scanning all items in Python.
hydrate_model_data replaces model_data with .metadata.json content which
may lack sha256 (corrupted file, concurrent write, etc.). Restore the
cached sha256 after hydration and persist the fix back to disk so
subsequent lookups don't hit the same error.
Also improve error log to include file_path for debugging.
LM Studio and some other OpenAI-compatible servers reject
response_format=json_object but accept json_schema. Switch to the
equivalent json_schema format and add a fallback that retries
without response_format when the provider rejects the format type.
- Use safe .get() in RecipeCache._resort_locked instead of itemgetter to prevent KeyError when recipe missing created_date; align sort key with _sort_cache_sync (prefer modified, fallback created_date, fallback 0)
- Add base_model to allowed_fields in persistence_service.update_recipe() so the field passes validation
- Route bulk base model updates through updateRecipeMetadata() on recipes page instead of generic saveModelMetadata(), matching existing isRecipesPage pattern used in setBulkFavorites and saveBulkTags
- Add enable_civarchive_api toggle (default on) to allow disabling
CivArchive to avoid its rate-limit windows entirely
- Add metadata_provider_order dropdown with two presets:
CivitAI → CivArchive → Archive DB (default) and
CivitAI → Archive DB → CivArchive
- Wire both settings through backend (metadata_service, settings_manager,
misc_handlers) and frontend (SettingsManager, state, settings modal)
- Reorder Metadata section in settings modal: toggles → status/management
→ fallback order, for natural top-down workflow
- Make update_metadata_providers() log the effective provider chain
using actually-registered providers rather than settings assumptions
- Add 5 test cases covering all provider-combination paths
- Complete i18n translations for 6 new keys across all 9 non-English locales
- Add PersistentModelCache.update_single_model() for lightweight targeted
SQL update (single row + incremental tag/hash deltas, no full table scan)
- Add ModelScanner.sync_cache_from_metadata() with compare-first logic:
skips entirely when cache is already in sync; when stale, updates the
entry in-place (O(1) instead of O(n) remove+append), incrementally
adjusts tag counts/hash index/version index, and resorts only when
sort-relevant fields changed
- Wire sync_cache_from_metadata() into BaseModelService.get_model_metadata()
via fire-and-forget asyncio.create_task — disk I/O is already paid for
- Include identity re-validation guard against concurrent cache replacement
- Add 16 tests covering _cache_entries_differ, sync_cache_from_metadata
(no-change, in-place, fallback, conditional resort), and
update_single_model (insert, tag delta, hash delta)
When Civitai returns 404 for /models/{id} (e.g. due to Civitai API bug
where un-deleted models still get 404), the fallback to CivArchive
provides metadata. However CivArchive may return mirrors with every
entry marked deletedAt, while the file's downloadUrl is still valid.
Before this fix, _build_download_urls_from_file_info used an if/else
that skipped the downloadUrl fallback whenever the mirrors array was
non-empty, even when all mirrors were filtered out. Now downloadUrl
is always tried when no usable mirror remains.
Also deduplicated the inline mirror-processing code at the second call
site by replacing it with a call to the shared helper.
Refactor _create_session() to make-before-break: snapshot old session,
assign new one first, then close old. Previously, concurrent download
retries called _create_session() without the session lock (violating its
docstring contract) and closed the old session while other coroutines
held active references — causing aiohttp to raise "NoneType has no
attribute connect" when dereferencing the torn-down connector.
Also wrap the two _create_session() calls in the integrity-retry and
network-retry paths with self._session_lock to match the locking
discipline used by the session property and refresh_session().
In add_to_queue, check download_history before INSERT OR IGNORE. Without
this check, a fire-and-forget /queue/complete failure on the extension side
would allow the same download_id to be re-inserted after complete_download()
deleted it from the queue — creating phantom queued entries for already-
finished downloads.
Add a pure frontend node that shows filename and editable notes for
a selected LoRA. Connect any output from a LoRA Loader/Stacker/Randomizer/
WanVideoSelect to the lora_source input — selecting a LoRA in the source
widget updates the info display automatically.
- Python node (LoraInfoLM): display-only, no workflow execution
- Vue widget: filename label, auto-sizing notes textarea, save button
with ComfyUI toast feedback on save
- Frontend extension: wire-based selection propagation with stale-response
race guard; clears display on wire disconnect
- Backend: get-notes endpoint now returns file_path alongside notes;
matching supports full-path lora syntax; fix NoneType crash in
trigger words endpoint; document cache file_name invariant
- Wired into all four lora widget nodes (Loader, Stacker, Randomizer,
WanVideoSelect)
Changes:
- Backend: _validate_folder_paths() now checks checkpoints↔unet overlap
within the same library using os.path.realpath() for symlink resolution
- Backend: set() calls _validate_folder_paths() for both folder_paths and
extra_folder_paths before writing
- Backend: extracted _normalize_path_set() helper to eliminate duplicated
normalization logic
- Frontend: inline error display with red border + error message below the
conflicting input, no save triggered
- Frontend: path normalization (strip trailing slash, lowercase) in pre-check
to reduce false negatives vs backend realpath
- Frontend: asymmetric error UX — message only on the user-edited side,
red border on the pre-existing conflict side
- CSS: has-error styles with hardcoded rgba fallback for older browsers
- i18n: checkpointUnetOverlap + checkpointUnetOverlapInline keys added to
all 10 locale files
- Bulk refresh filter now excludes models with hf_url
- Individual refresh for HF models only checks CivitAI API
- CivArchive client validates model IDs before querying
- Add onerror handler on <img> previews to fallback to no-preview.png
- Fire async cache cleanup when preview file returns 404
- Add ModelCache.clear_preview_by_path() for safe stale-url removal
- Downgrade /api/lm/previews 404 log from warning to debug
When Civitai returns 429 (Too Many Requests) during example image
downloads, the previous behavior treated all failures identically and
permanently removed the corresponding images from model metadata —
making them impossible to retry.
This commit adds:
- 429 detection + Retry-After header parsing in download_to_memory
- Exponential backoff retry (up to 3 attempts) in
download_model_images_with_tracking
- Separate tracking of rate-limited vs permanently failed URLs
- rate_limited_models progress tracking persisted to disk
- Rate-limited models are NOT added to failed_models/processed_models
so they are automatically retried on subsequent download runs
- Force mode clears failed_models when rate-limited images exist
- Parse limit.output from model catalog alongside model IDs
for per-model max output token limits
- Use catalog lookup in chat_completion_json() to set max_tokens;
fall back to 4096 for unknown models (e.g. local Ollama)
- Remove the JSON retry (response_format → plain text fallback);
keep _try_salvage_json as last-resort for truncated responses
- Reduce Ollama num_ctx from 32768 to 8192 (sufficient for
metadata enrichment, saves VRAM)
- Fix stale test comment referencing removed retry
The bare call inside _build_prompt_context
would raise NameError because class methods don't close over class-level
scope. Use instead to trigger attribute lookup.
Update enrich_hf_metadata prompt.md clue locations for better LLM accuracy.
Update baseline report to v2 (mean 69.0, 46 models, +2.2pp vs baseline 71.1%).
Consolidate README snapshots into baselines/readmes/.
- Rename py/agent_cli/ -> py/metadata_ops/ (module was never agent-related)
- Rename tests/agent_cli/ -> tests/metadata_ops/
- Remove 9 low-value/debug INFO log points across agent_handlers.py,
agent_service.py, llm_service.py, and metadata_ops/__init__.py
- Keep LLM raw response at DEBUG level for diagnostics
- Consolidate per-model progress + LLM result into single concise
log line with basename instead of full path
- Update package/class/method docstrings to clarify this is a
pipeline infrastructure, not a true agent loop
Three-part fix for enrich_hf_metadata failing to extract correct preview_url
from HuggingFace collection repos where models share flat heading levels:
1. _strip_standalone_images() now converts <img> tags to markdown image
syntax  instead of stripping the URL entirely, so the LLM
can still extract preview URLs.
2. _extract_section() uses a line-count-based forward window (stopping at
<a id> anchors) for non-heading matches, instead of stopping at the
very next heading. This prevents same-level sub-headings (# Download,
# Trigger, # Sample prompt within a single model section) from
truncating the window before sample images are included.
3. Post-processor preview fallback now filters gallery images to the
model-specific README section before falling back to the repo-wide
first image.
- agent_service._format_base_models: output bullet list instead of
JSON array for cleaner LLM parsing
- prompt.md mapping section: replace 14-row HF→CivitAI table with
compact rule set covering 14 mapping paths including new entries
for HiDream-ai, OnomaAIResearch/Illustrious, ideogram-ai/ideogram,
Tongyi-MAI/Z-Image-Turbo, and Wan-AI/Wan2.*
- base_model extraction instruction: add guidance to infer from
model filename, YAML tags, and README body text when YAML
frontmatter has no explicit base_model:
- Rename md_to_html.py → readme_processor.py (file no longer just HTML conversion)
- _extract_section: include YAML frontmatter, use heading-level-aware forward
walk (sub-headings under # are included), increase walk limit past 30 lines
- _is_heading: exclude </hN> closing tags from boundary detection
- _heading_level: new helper for heading-level-aware section matching
- css: yield 0 for heading like closing tags, was unexpectedly caught by _is_heading
- extract_gallery_images: fix YAML block scalar (text: >-) prompt extraction;
use endswith instead of == to detect the block marker
- _strip_widget_section: add to clean_readme_for_llm (widget text is handled
by post-processor, not needed in LLM prompt)
- _strip_standalone_images: keep markdown image URLs intact for LLM preview
extraction (was stripping to alt text only)
- list_base_models: switch from scanner-cache aggregation to
CivitaiBaseModelService.get_base_models() - always returns full list
- Ollama: add num_ctx=32768 to payload options so thinking models have room
to both reason and produce output
- Add tests/agent_cli/test_readme_processor.py: 59 tests covering extraction,
cleaning, section matching, heading detection
- Update existing tests for behavioral changes
- PostProcessor returns updates dict from enrich_hf_metadata
- AgentService includes updated_data per model in WebSocket progress events
- Convert preview_url to HTTP URL via config.get_preview_static_url()
- LoraContextMenu: showEnhancedProgress + updateSingleItem per model
- BulkContextMenu: same pattern, remove window.location.reload()
- Guard empty updated_data and clean up callbacks on HTTP error
- Add clean_readme_for_llm() to strip noise from README before LLM injection
- Keep widget section text (valuable tag signal) and unmarked code blocks (trigger words)
- Preserve standalone image alt text instead of removing entirely
- Switch Ollama to native /api/chat with think:false to fix empty content on thinking models
- Extract Sample Gallery table images and deduplicate with widget images
- Only strip code blocks with explicit language tags (bash)
- Add notes and usage_tips fields to SKILL.md output format and post-processor
- Clean up dead code, fix regex edge cases, remove double type annotation
- Replace hardcoded provider list with PROVIDER_PRESETS (OpenAI, Ollama,
DeepSeek, Groq, OpenRouter, OpenCode Go, Custom)
- Load model lists from models.dev/api.json catalog at startup
- Add Combobox vanilla JS component for model/base-URL selection
- Fetch local Ollama models via live API instead of catalog
- Hide API key values from frontend (boolean-only llm_api_key_set)
- Add i18n translations for all 9+ locales
- Update snapshot tests for new response fields
Widget entries with unquoted multi-line YAML scalars (e.g. "text: two samurais...\n continuation") were not parsed, leaving gallery image prompts empty. Add a third branch for plain scalar format alongside the existing quoted and >- folded block handlers.
- Add extract_gallery_images() to parse YAML widget entries from README
frontmatter, convert relative image URLs to absolute HF URLs, and
build civitai.images-compatible entries with prompt metadata
- LLM now extracts recommended_width/recommended_height from README
(e.g. "Best Dimensions"), used as gallery image dimensions
- extract_gallery_images() accepts default_width/height parameters,
falling back to 512x512 when LLM provides no recommendation
- Frontend ShowcaseView.js: defensive NaN guard for 0 width/height
- post_processor: consistently merge civitai updates across triggers,
description, and gallery blocks with distinct variable names
- SKILL.md: add recommended_width/recommended_height to output schema
- 62 tests pass, including gallery extraction and dimension tests
- Add identify_model_type() helper to determine lora/checkpoint/embedding
- Pass priority_tags from user settings to LLM prompt for tag relevance
- SKILL.md: instruct LLM to exclude technical/generic HF tags, cross-reference
against priority_tags; forbid ['None'] placeholder for trigger words
- post_processor: fix preview_url not updated after download (now writes local
.webp path to metadata); write trigger words to civitai.trainedWords instead
of top-level; sanitize ['None']/'null'/'n/a' placeholder values to []
- download_preview() now returns str | None (local path) instead of bool
- Update tests for new return type and nested civitai.trainedWords structure
Merge skill.yaml (metadata) and prompt.md (prompt template) into a
single SKILL.md file with YAML frontmatter, matching the agent-skill
convention used by opencode and Claude Code.
- Add frontmatter parser (_parse_skill_file) to SkillRegistry
- Remove skill.yaml, prompt.md, empty skills/__init__.py
- Remove obsolete load_handler method
- Update tests for new format and cleaned-up fields
