Instead of always using default paths, downloads from the model versions
tab now target the same directory as the current in-library version.
Falls back silently to default paths if the current version path cannot
be resolved.
Store textarea reference on container element to allow cloned widgets to access inputEl when promoted to subgraph nodes. This ensures both original and cloned widgets can properly get and set values through the shared DOM element.
Update the help text for 'Metadata Refresh Skip Paths' setting to explicitly
state that paths should be relative to the 'model root directory' instead of
just saying 'relative folder paths', which was ambiguous.
Updated translations:
- English (en)
- Chinese Simplified (zh-CN)
- Chinese Traditional (zh-TW)
- Japanese (ja)
- Korean (ko)
- Russian (ru)
- German (de)
- French (fr)
- Spanish (es)
- Hebrew (he)
Replace _SYNC_KEYS (37 keys) with _NO_SYNC_KEYS (5 keys) in SettingsHandler.
New settings automatically sync to frontend unless explicitly excluded.
Changes:
- SettingsHandler now syncs all settings except those in _NO_SYNC_KEYS
- Added keys() method to SettingsManager for iteration
- Updated tests to use new behavior
Benefits:
- No more missing keys when adding new settings
- Reduced maintenance burden
- Explicit exclusions for sensitive/internal settings only
Fixes: #86
Complete TODO translations from previous commit:
- Add translations for hideEarlyAccessUpdates setting
- Add translations for EA time formatting (endingSoon, hours, days)
- Add translations for EA badges and tooltips
- Translate to: de, es, fr, he, ja, ko, ru, zh-CN, zh-TW
Closes#815 translations
Add Early Access version support with filtering and improved UI:
Backend:
- Add is_early_access and early_access_ends_at fields to ModelVersionRecord
- Implement two-phase EA detection (bulk API + single API enrichment)
- Add hide_early_access_updates setting to filter EA updates
- Update has_update() and has_updates_bulk() to respect EA filter setting
- Add _enrich_early_access_details() for precise EA time fetching
- Fix setting propagation through base_model_service and model_update_service
Frontend:
- Add smart relative time display for EA (in Xh, in Xd, or date)
- Replace EA label with clock icon in metadata (fa-clock)
- Show Download button with bolt icon for EA versions (fa-bolt)
- Change EA badge color to #F59F00 (CivitAI Buzz theme)
- Fix toggle UI for hide_early_access_updates setting
- Add translation keys for EA time formatting
Tests:
- Update all tests to pass with new EA functionality
- Add test coverage for EA filtering logic
Closes#815
- Backend: Support limit=0 to return all tags in top-tags API
- Frontend: Remove tags limit setting and fetch all tags by default
- UI: Implement virtual scrolling in TagsModal for performance
- Initial display 200 tags, load more on scroll
- Show all results when searching
- Remove lora_pool_tags_limit setting to simplify UX
Fixes#819
Fix KeyError when 'hashes', 'name', or 'model' fields are missing from
Civitai API responses. Use .get() with defaults instead of direct dict
access in:
- LoraMetadata.from_civitai_info()
- CheckpointMetadata.from_civitai_info()
- EmbeddingMetadata.from_civitai_info()
- RecipeScanner._get_hash_from_civitai()
- DownloadManager._process_download()
Fixes#820
Previously, long LoRA filenames were truncated from the right with ellipsis,
which hid important checkpoint step numbers (e.g., -00800, -01000) that users
need to distinguish between different training checkpoints.
Changes:
- Replace single-line truncation with multi-line display (max 3 lines)
- Add line-height and word-break properties for better readability
- Use -webkit-line-clamp to gracefully handle extremely long names
This ensures the step number suffix is always visible in the tooltip.
- Remove max height setting, let ComfyUI handle widget sizing
- Widget now uses getMinHeight() to declare 150px minimum
- Container fills available space with overflow: auto for scrollbars
- Users can freely resize the node; content overflows show scrollbar
- Simplified renderTags by removing height calculation logic
Fixes#706
- Add new setting 'loramanager.trigger_word_max_height' (150-600px, default 300px)
- Add getTriggerWordMaxHeight() getter to retrieve setting value
- Update tags_widget to respect max height limit with scrollbar
- Add getMaxHeight callback for ComfyUI layout system
- Add tooltip note about requiring page reload
Fixes#706
- Extract progress file loading to async methods to run in executor
- Refactor start_download to reduce lock time by pre-loading data before entering lock
- Improve check_pending_models efficiency with single-pass model collection and async loading
- Add type hints to get_status method
- Add tests for download task callback execution and error handling
Translate all skip metadata refresh UI strings to all supported languages:
- zh-CN, zh-TW, ja, ko, de, fr, es, ru, he
Completes the translation TODOs from the previous commit.
- Reframe supporter access section to emphasize sustainability and gratitude
- Add CivArchive support announcement and image
- Document new dedicated download button and hide models feature in v0.4.8
- Improve readability and flow of the overview and supporter sections
- Add clear button inside autocomplete text widget that shows when text exists
- Support both Canvas mode and Vue DOM mode with appropriate styling
- Fix clear button visibility when value is changed externally (e.g., via 'send lora to workflow')
- Implement dual notification mechanism: CustomEvent + onSetValue callback
- Update widget interface to include onSetValue property
When no duplicate groups are detected, the duplicate manager now checks if it is currently in duplicate mode and calls `exitDuplicateMode()` to clear the display. This prevents the UI from showing stale duplicate information when no duplicates exist.
Add a segmented toggle in the Filter Panel to switch between 'Any' (OR)
and 'All' (AND) logic when filtering by multiple include tags.
Changes:
- Backend: Add tag_logic field to FilterCriteria and ModelFilterSet
- Backend: Parse tag_logic parameter in model handlers
- Frontend: Add segmented toggle UI in filter panel header
- Frontend: Add interaction logic and state management for tag logic
- Add translations for all supported languages
- Add comprehensive tests for the new feature
Closes#802
Add automatic cleanup of default values from settings.json to keep configuration files minimal and focused on user customizations. Introduces a threshold-based cleanup that only removes default values when the file contains a significant number of them (10+), preserving small template-based configurations while cleaning up legacy bloated files.
Key changes:
- Add DEFAULT_KEYS_CLEANUP_THRESHOLD constant to control cleanup aggressiveness
- Implement _cleanup_default_values_from_disk() method that removes default values from disk while keeping them available in memory
- Modify _ensure_default_settings() to only save when existing values are updated, not when defaults are inserted
- Update _serialize_settings_for_disk() to only persist settings that differ from defaults
- Add cleanup call during initialization for existing settings files
This reduces file size and noise in settings.json while maintaining full functionality at runtime.
- Added filter presets feature allowing users to save and quickly switch between filter combinations
- Fixed various bugs to improve overall stability
- Updated project version from 0.9.14 to 0.9.15 in pyproject.toml
- Use modelVersionId as fallback for all loras in fingerprint calculation (not just deleted)
- Add URL-based duplicate detection using source_path field
- Combine both fingerprint and URL-based duplicate detection in API response
- Fix _download_remote_media return type and unbound variable issue
- Hide delete button by default and show on hover for inactive presets
- Show delete button on active presets only when hovering over the preset
- Add ellipsis truncation for long preset names to prevent layout breakage
- Remove checkmark icon from active preset names for cleaner visual design
When checkpoints and unet folders point to the same physical location
(via symlinks), prioritize checkpoints for backward compatibility.
This prevents the 'Failed to load Checkpoint root' error that users
experience when they have incorrectly configured their ComfyUI paths.
Changes:
- Detect overlapping real paths between checkpoints and unet
- Log warning to inform users of the configuration issue
- Remove overlapping paths from unet_map, keeping checkpoints
Fixes #<issue-number>
Introduce a new private method `_normalize_trigger_words` to handle consistent splitting and cleaning of trigger word strings. This method splits input by both single and double commas, strips whitespace, and filters out empty strings, returning a set of normalized words. It is now used in `process_trigger_words` to compare trigger word overrides, ensuring accurate detection of changes by comparing normalized sets instead of raw strings.
The width of the repeat input field in the LoRA cycler settings view has been increased from 40px to 50px. This change improves usability by providing more space for user input, making the control easier to interact with and reducing visual crowding.
When a deleted model is checked against the SQLite archive and not found, the `db_checked` flag was set in memory but never saved to disk. This occurred because the save operation was only triggered when `civitai_api_not_found` was True, which is not the case for deleted models (since the CivitAI API is not attempted). As a result, deleted models would be rechecked on every refresh instead of being skipped.
Changes:
- Introduce a `needs_save` flag to track when metadata state is updated
- Save metadata whenever `db_checked` is set to True, regardless of API status
- Ensure `last_checked_at` is set for SQLite-only attempts
- Add regression test to verify the fix
- Add LoRA Cycler node with iteration support
- Enhance Prompt node with tag autocomplete (Danbooru + e621)
- Add command system (/char, /artist, /ac, /noac) for tag operations
- Reference Lora Cycler and Lora Manager Basic template workflows
- Bug fixes and stability improvements
Refactor recipe lookup logic to improve efficiency from O(n²) to O(n + m):
- Build recipe_by_id dictionary for O(1) recipe ID lookups
- Simplify persisted_by_path construction using recipe_id extraction
- Add fallback lookup by recipe_id when path lookup fails
- Maintain same functionality while reducing computational complexity
When enable_metadata_archive_db=True, the previous filter logic would
repeatedly try to fetch metadata for models that were already confirmed
to not exist on CivitAI (from_civitai=False, civitai_deleted=True).
The fix adds a skip condition to exclude models that:
1. Are confirmed not from CivitAI (from_civitai=False)
2. Are marked as deleted/not found on CivitAI (civitai_deleted=True)
3. Either have no archive DB enabled, or have already been checked (db_checked=True)
This prevents unnecessary API calls to CivArchive for user-trained models
or models from non-CivitAI sources.
Fixes repeated "Error fetching version of CivArchive model by hash" logs
for models that will never be found on CivitAI/CivArchive.
- Add /api/example-images/check-pending endpoint to quickly check models needing downloads
- Improve DownloadManager.start_download() to return immediately without blocking
- Add _handle_download_task_done callback for proper error handling and progress saving
- Add check_pending_models() method for lightweight pre-download validation
- Update frontend ExampleImagesManager to use new check-pending endpoint
- Add comprehensive tests for new functionality
Introduce comprehensive documentation for the new `lora-manager-e2e` skill, which provides end-to-end testing workflows for LoRa Manager. The skill enables automated validation of standalone mode, including server management, UI interaction via Chrome DevTools MCP, and frontend-to-backend integration testing.
Key additions:
- Detailed skill description and prerequisites
- Quick start workflow for server setup and browser debugging
- Common E2E test patterns for page load verification, server restart, and API testing
- Example test flows demonstrating step-by-step validation procedures
- Scripts and MCP command examples for practical implementation
This documentation supports automated testing of LoRa Manager's web interface and backend functionality, ensuring reliable end-to-end validation of features.
- Add cache entry validator service for data integrity checks
- Add cache health monitor service for periodic health checks
- Enhance model cache and scanner with validation support
- Update websocket manager for health status broadcasting
- Add initialization banner service for cache health alerts
- Add comprehensive test coverage for new services
- Update translations across all locales
- Refactor sync translation keys script
- Modify custom words search to extract last space-separated token from search term
- Add `_getLastSpaceToken` helper method for token extraction
- Update selection replacement logic to only replace last token in multi-word prompts
- Enables searching "hello 1gi" to find "1girl" and replace only "1gi" with "1girl"
- Maintains full command replacement for command mode (e.g., "/char miku")
JavaScript floating point arithmetic causes values like 1.1 to become
1.1000000000000014. Add precision limiting to 2 decimal places in
snapToStep function for both sliders.
- Add LoraListModal component with search and preview tooltip
- Make 'Next LoRA' name clickable to open selector modal
- Integrate PreviewTooltip with custom resolver for Vue widgets
- Disable selector when prompts are queued (consistent with pause button)
- Fix tooltip z-index to display above modal backdrop
Fixes issue: users couldn't easily identify which index corresponds
to specific LoRA in large lists
- Replace REP badge with segmented progress bar for repeat indicator
- Reorganize Starting Index & Repeat controls into aligned groups
- Change repeat format from '× [count] times' to '[count] ×' for better alignment
- Remove unnecessary refresh button and related logic
Replace recursive directory traversal with first-level-only symlink scanning
to fix severe performance issues on large model collections (220K+ files).
- Rename _scan_directory_links to _scan_first_level_symlinks
- Only scan symlinks directly under each root directory
- Skip traversal of normal subdirectories entirely
- Update tests to reflect first-level behavior
- Add test_deep_symlink_not_scanned to document intentional limitation
Startup time reduced from 15+ minutes to seconds for affected users.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add `/ac` and `/noac` commands to toggle prompt tag autocomplete on/off
- Commands only appear when relevant (e.g., `/ac` shows when autocomplete is off)
- Show toast notification when toggling setting
- Use ComfyUI's setting API with fallback to legacy API
- Clear autocomplete token after toggling to provide clean UX
- Add `hasQueuedPrompts` reactive flag to track queued executions
- Pass `is-pause-disabled` prop to settings view to disable pause button
- Update pause button title to indicate why it's disabled
- Remove server queue clearing logic from pause toggle handler
- Clear `hasQueuedPrompts` flag when manually changing index or resetting
- Set `hasQueuedPrompts` to true when adding prompts to execution queue
- Update flag when processing queued executions to reflect current queue state
- Update package.json test script to run both JS and Vue tests
- Simplify LoraCyclerLM output by removing redundant lora name fallback
- Extend Vitest config to include TypeScript test files
- Add Vue testing dependencies and setup for component testing
- Implement comprehensive test suite for BatchQueueSimulator component
- Add test setup file with global mocks for ComfyUI modules
- Change `model_type` field to `sub_type` for checkpoint models to improve naming consistency
- Add `sub_type="embedding"` for embedding models to properly categorize model subtypes
- Maintain backward compatibility with existing metadata structure
- Add optional `media_type_hint` parameter to `_get_file_extension_from_content_or_headers` method
- When `media_type_hint` is "video" and no extension can be determined from content/headers/URL, default to `.mp4`
- Pass image metadata type as hint in both `process_example_images` and `process_example_images_batch` methods
- Add unit tests to verify media type hint behavior and priority
Add CSS rules to hide the model sub-type and separator elements when the compact-density class is applied. This change saves visual space in compact mode by removing less critical information, improving the layout for dense interfaces.
Reset execution state when user manually changes LoRA index to ensure next execution starts from the user-set index. This prevents stale execution state from interfering with user-initiated index changes.
- Add Flux.2 Klein 9B, 9B-base, 4B, and 4B-base models to BASE_MODELS, BASE_MODEL_ABBREVIATIONS, and Flux Models category
- Include ZImageBase model and its abbreviation
- Add LTXV2 video model to BASE_MODELS, BASE_MODEL_ABBREVIATIONS, and Video Models category
- Update model categories to reflect new additions
The no-presets placeholder element has been removed from the filter presets section in the header component. This change likely indicates that the application now handles empty presets states differently, possibly through dynamic content rendering or alternative UI patterns.
- Add gradient overlay to card header for better icon readability
- Update base model label to display sub-type abbreviation alongside base model
- Add separator between sub-type and base model for visual clarity
- Improve label styling with flex layout, adjusted padding, and enhanced backdrop filter
- Add helper functions for sub-type abbreviation retrieval and display names
- Remove backward compatibility code for `model_type` in `ModelScanner._build_cache_entry()`
- Update `CheckpointScanner` to only handle `sub_type` in `adjust_metadata()` and `adjust_cached_entry()`
- Delete deprecated aliases `resolve_civitai_model_type` and `normalize_civitai_model_type` from `model_query.py`
- Update frontend components (`RecipeModal.js`, `ModelCard.js`, etc.) to use `sub_type` instead of `model_type`
- Update API response format to return only `sub_type`, removing `model_type` from service responses
- Revise technical documentation to mark Phase 5 as completed and remove outdated TODO items
All cleanup tasks for the model type refactoring are now complete, ensuring consistent use of `sub_type` across the codebase.
This commit resolves the semantic confusion around the model_type field by
clearly distinguishing between:
- scanner_type: architecture-level (lora/checkpoint/embedding)
- sub_type: business-level subtype (lora/locon/dora/checkpoint/diffusion_model/embedding)
Backend Changes:
- Rename model_type to sub_type in CheckpointMetadata and EmbeddingMetadata
- Add resolve_sub_type() and normalize_sub_type() in model_query.py
- Update checkpoint_scanner to use _resolve_sub_type()
- Update service format_response to include both sub_type and model_type
- Add VALID_*_SUB_TYPES constants with backward compatible aliases
Frontend Changes:
- Add MODEL_SUBTYPE_DISPLAY_NAMES constants
- Keep MODEL_TYPE_DISPLAY_NAMES as backward compatible alias
Testing:
- Add 43 new tests covering sub_type resolution and API response
Documentation:
- Add refactoring todo document to docs/technical/
BREAKING CHANGE: None - full backward compatibility maintained
Move filter preset creation, deletion, application, and storage logic
from FilterManager into a dedicated FilterPresetManager class to
improve separation of concerns and maintainability.
- Add FilterPresetManager with preset CRUD operations
- Update FilterManager to use preset manager via composition
- Handle EMPTY_WILDCARD_MARKER for wildcard base model filters
- Add preset-related translations to all locale files
- Update filter preset UI styling and interactions
Use union type "AUTOCOMPLETE_TEXT_PROMPT,STRING" to enable input mode
compatibility with STRING outputs while preserving autocomplete widget
functionality via widgetType option.
Fixes issue where text inputs could not receive connections from
STRING-type outputs after changing from built-in STRING to custom
AUTOCOMPLETE_TEXT_PROMPT type.
Affected nodes:
- Prompt (LoraManager)
- Text (LoraManager)
Moves onboarding_completed and dismissed_banners from localStorage
to backend settings (settings.json) to survive incognito/private
browser modes.
Fixes#786
- Change badge from text label to icon-only for cleaner UI
- Adjust CSS for smaller circular badge with centered icon
- Maintain tooltip functionality for accessibility
- Update badge styling to be more compact and visually consistent
When users try to import custom example images without configuring the
download location, show a helpful guidance interface instead of failing
silently or showing an error after the fact.
Changes:
- ShowcaseView.js: Check if example_images_path is configured before
showing import interface; display setup guidance with open settings button
- showcase.css: Add styles for the setup guidance state
- locales: Add translation keys for all 10 supported languages
Clicking 'Open Settings' will:
1. Open the settings modal
2. Scroll to the Example Images section
3. Highlight the section with a brief animation
4. Focus the input field
Fixes#785
- Auto-commit input value when clicking save button
- Auto-commit on blur to handle users clicking outside input
- Fixes issue where users would type a trigger word and click save,
but the word wasn't added because they didn't press Enter first
- Maintains backward compatibility with existing comma-based workflows
Introduce a new TextLM node to the Lora Manager extension, providing a simple text input with autocomplete functionality for tags and styles. The node is integrated into the module's import system and node class mappings, enabling users to utilize autocomplete features for efficient prompt creation.
- Add generic type parameter to ComponentWidget<T> for type-safe callbacks
- Remove LegacyLoraPoolConfig interface and migrateConfig function
- Update LoraPoolWidget to use ComponentWidget<LoraPoolConfig>
- Clean up type imports across widget files
- Restructure document to clearly separate simple vs complex widget patterns
- Add detailed explanation of ComfyUI's built-in callback mechanism
- Provide complete implementation examples for both patterns
- Remove outdated sync chain diagrams and replace with practical guidance
- Emphasize using DOM element as source of truth for simple widgets
- Document proper use of internal state with widget.callback for complex widgets
- Add aliases column to tags table to store comma-separated alias lists
- Update FTS schema to version 2 with searchable_text field containing tag names and aliases
- Implement schema migration to rebuild index when upgrading from old schema
- Modify search logic to match aliases and return canonical tag with matched alias info
- Update index building to include aliases in searchable text for FTS matching
This enables users to search for tag aliases (e.g., "miku") and get results for the canonical tag (e.g., "hatsune_miku") with indication of which alias was matched.
Remove pointer event .stop modifiers from textarea to allow events
to propagate to container where forwardMiddleMouseToCanvas forwards them
to ComfyUI canvas for pan functionality
Remove multiple sources of truth and async sync chains that caused
values to be lost during load/switch workflow or reload page.
Changes:
- Remove internalValue state variable from main.ts
- Update getValue/setValue to read/write DOM directly via widget.inputEl
- Remove textValue reactive ref and v-model from Vue component
- Remove serializeValue, onSetValue, and watch callbacks
- Register textarea reference on mount, clean up on unmount
- Simplify AutocompleteTextWidgetInterface
Follows ComfyUI built-in addMultilineWidget pattern:
- Single source of truth (DOM element value only)
- Direct sync (no intermediate variables or async chains)
Also adds documentation:
- docs/dom-widgets/value-persistence-best-practices.md
- docs/dom-widgets/README.md
- Update docs/dom_widget_dev_guide.md with reference
Remove all autocomplete.txt parsing logic and fallback code, simplifying
the service to use only TagFTSIndex for Danbooru/e621 tag search
with category filtering.
- Remove WordEntry dataclass and _words_cache, _file_path attributes
- Remove _determine_file_path(), get_file_path(), load_words(), save_words(),
get_content(), _parse_csv_content() methods
- Simplify search_words() to only use TagFTSIndex, always returning
enriched results with {tag_name, category, post_count}
- Remove GET/POST /api/lm/custom-words endpoints (unused)
- Keep GET /api/lm/custom-words/search for frontend autocomplete
- Rewrite tests to focus on TagFTSIndex integration
This reduces code by 446 lines and removes untested pysssss plugin
integration. Feature is unreleased so no backward compatibility needed.
- Centralize cache path resolution in new py/utils/cache_paths.py module
- Migrate legacy cache files to organized structure: {settings_dir}/cache/{model|recipe|fts|symlink}/
- Automatically clean up legacy files after successful migration with integrity verification
- Update Config symlink cache to use new path and migrate from old location
- Simplify service classes (PersistentModelCache, PersistentRecipeCache, RecipeFTSIndex, TagFTSIndex) to use centralized migration logic
- Add comprehensive test coverage for cache paths and automatic cleanup
- Change path separators from backslashes to forward slashes in embedding autocomplete
- Extend embedding detection to also trigger when searchType is 'embeddings'
- Improves cross-platform compatibility and makes embedding autocomplete more reliable
Update the placeholder text in the PromptLM class to include guidance for quick tag search functionality. The new placeholder now reads "Enter prompt... /char, /artist for quick tag search", providing users with immediate cues on how to utilize tag search features directly within the input field. This improves usability by making advanced functionality more discoverable.
Remove searchType check from prompt behavior's hidePreview method.
When an embedding was selected, the input event dispatched by
insertSelection caused searchType to change before hide() was called,
preventing the preview tooltip from being hidden.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Will Miao