* docs(AGENTS.md): add guidance on using Chrome DevTools MCP tool for UI changes to improve development process and ensure accuracy in modifications * feat(docs): add Resource Monitor documentation and initial implementation details for the restart button feature docs(docs): update README.md to include Resource Monitor link and overview feat(resource_monitor.js): add resource monitor extension with restart button to ComfyUI top bar This extension enhances the user interface by integrating a resource monitor similar to Crystools, allowing for future CPU/RAM/GPU/VRAM monitoring capabilities. * feat: remove restart button extension and associated styles from Swiss Army Knife * feat(resource_monitor.js): implement restart functionality for ComfyUI server with user notification and error handling * fix(resource_monitor.js): change reload timeout from 2 seconds to 5 seconds to allow more time for reconnection feat(resource_monitor.js): use hard reload to bypass cache when reloading the page for better resource updates * feat(monitor): add resource monitor service and API endpoints for hardware monitoring feat(monitor): initialize resource monitor service in __init__.py and register API routes chore(monitor): create utility modules for hardware and GPU information handling feat(gpu_info.py): add GPU information collector to monitor VRAM, utilization, and temperature This new module collects GPU metrics using pynvml and torch, providing detailed information about available GPUs and their performance. It enhances monitoring capabilities for applications utilizing GPU resources. feat(monitoring): add hardware information collector and monitoring service to gather and broadcast system metrics - Introduce `hardware_info.py` to collect CPU, RAM, and disk usage metrics. - Implement `monitor_service.py` to periodically collect and broadcast hardware metrics. - Update `requirements.txt` to include necessary dependencies: psutil, pynvml, and py-cpuinfo. feat(tests): add test script for resource monitor backend to verify hardware monitoring functionality * docs(Resource Monitor Testing): add comprehensive testing guide for resource monitor implementation to ensure proper functionality and user experience feat(resource_monitor.js): enhance resource monitoring with dynamic display and color coding for CPU, RAM, and GPU metrics - Add styles for resource monitor display and labels. - Create functions to format and update monitor values. - Implement dynamic creation of monitor elements based on available hardware. - Fetch initial status and listen for updates via WebSocket to keep metrics current. * refactor(resource_monitor.js): move restart button to be added after monitors for better UI flow * feat(resource_monitor.js): enhance server restart process with health checks and user feedback style(resource_monitor.js): improve button styles and layout for better UX refactor(resource_monitor.js): extract GPU model name for clearer display in UI fix(resource_monitor.js): update formatting for memory and CPU display values * feat(resource_monitor): update UI to a floating HUD with live telemetry and inline restart workflow * docs(RESOURCE_MONITOR): update roadmap section with near-term enhancements and ProfilerX parity plan * style(resource_monitor.js): enhance restart button styles for improved UI aesthetics and usability * style(resource_monitor.js): update font size and line height for better readability in the restart button styles * style(resource_monitor.js): update CSS styles for swissarmyknife-monitor class to enhance layout and readability * docs(RESOURCE_MONITOR.md): update the ProfilerX Parity Plan section to include detailed implementation hooks and phases for better clarity and guidance style(resource_monitor.js): improve CSS styles for monitor content and labels to enhance readability and layout consistency * feat(resource_monitor.js): add background progress bar to monitor display for better visual feedback refactor(resource_monitor.js): update percentage thresholds for color changes in progress bar for improved clarity * style(resource_monitor.js): adjust divider styles for better alignment and visibility in resource monitor UI * style(resource_monitor.js): update styles for the restart button to enhance appearance and usability * style(resource_monitor.js): update styles for restart button to use flexbox for better alignment refactor(resource_monitor.js): simplify GPU and VRAM display labels for consistency and clarity * style(resource_monitor.js): update button styles to center content and adjust padding for better layout * style(resource_monitor.js): update restart button class for consistent styling * feat(profiler): implement profiling hooks and API for workflow performance tracking * feat(profiler): add workflow profiling functionality with UI integration - Implemented NodeProfile and WorkflowProfile classes for tracking execution time, VRAM/RAM usage, and cache hits. - Created ProfilerManager as a singleton to manage profiling sessions and history. - Added methods for starting and ending workflows and nodes, including VRAM and RAM tracking. - Introduced a new button in the resource monitor UI to access the profiler. - Developed a popup interface to display profiling statistics and node performance. - Implemented fetching and displaying of profiler data with retry logic. - Styled the profiler UI components for better user experience. - Added functionality to clear profiler history and view full history in a modal. * feat(prompt_builder): add Prompt Builder node for custom prompt composition feat(resource_monitor): enhance resource monitor UI with action button group and improved button styles * feat(profiler): enhance workflow profiling with console logging and data fetching improvements * feat(profiler): add execution time logging for node profiling and adjust popup positioning * feat(prompt_builder): add Prompt Builder node documentation and quick reference guide feat(media_describe): update override fields for backward compatibility and clarity * feat(media_describe): rename 'override_movement' to 'override_action' for consistency * feat(prompt_builder): update MediaDescribeOverrides documentation and add Prompt Builder node details * feat(prompt_builder): import Prompt Builder node and update mappings * feat(profiler): enhance profiling with VRAM tracking and execution timing for nodes * feat(prompt_builder): update Prompt Builder node to streamline input parameters and enhance documentation * feat(lora_info_extractor): add LoRA Info Extractor node for enriched metadata extraction and CivitAI lookup * feat(lora_info_extractor): integrate LoRA Info Extractor node and update documentation * refactor(civit_metadata_helper.py): improve code readability and structure by reformatting input types and updating docstrings feat(civit_metadata_helper.py): enhance metadata collection with structured outputs for better CivitAI submissions refactor(civit_metadata_helper.py): restructure collect_metadata method for better readability and maintainability feat(civit_metadata_helper.py): add type hints to method signatures for improved clarity and type safety chore(civit_metadata_helper.py): clean up code by removing commented-out sections and improving prompt handling * feat(civit_metadata_helper): add input normalization for metadata collection to handle various data types * feat(config_api.py): add profiler_enabled setting to configuration for enhanced performance tracking feat(prestartup.py): integrate profiler_enabled check to conditionally enable profiling functionality feat(swiss-army-knife.js): add UI option for enabling/disabling workflow profiler and sync with backend settings * feat(resource_monitor.js): enhance hard reload functionality with multiple methods for better browser compatibility style(resource_monitor.js): adjust popup positioning and styles for improved layout and user experience style(resource_monitor.js): update action button group to a vertical layout for better accessibility * feat(resource_monitor.js): refactor style injection to load external CSS for resource monitor and update monitor display logic * feat(resource_monitor): enhance documentation with known limitations and troubleshooting tips for profiling features * feat(profiler): add GPU model tracking and tooltip display for loaded models * feat(resource_monitor): add execution glow effect to monitor during workflow execution * feat(resource_monitor): enhance workflow execution glow effect with dual pseudo-elements for improved visual feedback * feat(resource_monitor): enhance duration formatting for total execution time in profiler stats * feat(resource_monitor): enhance monitor layout with inner wrapper for improved telemetry display * feat: Implement GPU memory scanning and enhanced model detection - Added direct GPU memory scanning in `profiler_core.py` to track loaded models using Python's garbage collector. - Removed fragile node-name tracking and implemented a caching mechanism for model scans. - Introduced comprehensive debug logging in `resource_monitor.js` for better troubleshooting. - Enhanced model grouping by extracting metadata from safetensors files, allowing for user-friendly model names and accurate VRAM reporting. - Implemented a five-phase approach to identify top-level models, calculate VRAM, and match with metadata. - Improved performance with caching strategies and optimized scanning logic to limit results to top 10 models per GPU. - Added tests and validation steps to ensure accurate model detection and reporting. * feat(resource_monitor): enhance logging with color-coded messages and improve VRAM tooltip functionality * feat(resource_monitor): implement OOM diagnosis and prevention features with enhanced logging and tooltip updates * feat(prestartup.py): implement OOM detection and logging for PyTorch errors to enhance error handling and debugging capabilities feat(profiler_core.py): add out-of-memory (OOM) detection and tracking fields to NodeProfile and WorkflowProfile for better resource management feat(profiler_core.py): implement OOM statistics update logic in ProfilerManager to track OOM occurrences and their details fix(profiler_core.py): ensure VRAM usage is monitored before node execution to prevent OOM issues feat(profiler_core.py): add OOM statistics and analytics to ProfilerManager for better monitoring and recommendations fix(profiler_core.py): include OOM summary in archived workflows to enhance data insights * fix(profiler_core.py): update OOM event to calculate execution time from start and end times for better performance tracking * feat(resource_monitor): add OOM analysis UI with detailed statistics and recommendations * feat(resource_monitor): implement comprehensive OOM detection and analysis features with frontend UI enhancements * feat(profiler): enhance profiler popup interaction with hover and timeout logic * fix(prompt_builder.py): change join delimiter from single newline to double newline for better prompt formatting * feat(docs): add detailed character analysis system prompt to structured output documentation for clarity and usage guidance feat(llm_studio_structured.py): update system prompt to reflect character analysis focus and improve output consistency docs(LLMStudioStructuredDescribe.md): create new documentation file for structured description node with usage tips and input/output details * feat(docs): add comprehensive guidelines for custom node documentation using Markdown * refactor: Remove retired Media Describe, LLM Studio Describe Nodes * refactor: rename docs to dev-docs * docs: update developer documentation for clarity and structure * refactor: remove filename generator, update video metadata, show text and control panel overview nodes * feat: add NSFW field to schemas and prompts for structured output nodes * ♻️ (llm_studio_structured.py): refactor nsfw field type from boolean to string for better clarity and flexibility in handling NSFW content descriptions. * ♻️ (llm_studio_structured.py): refactor max_tokens configuration to increase max limit and change step size for better granularity in token generation settings * ✨ (prompt_builder.py): add support for NSFW text override in PromptBuilder to enhance prompt customization options * feat(media-selection): enhance media source options to include Redgifs and Gfycat URLs for better media retrieval fix(media-selection): update error messages to reflect new media source options for clarity fix(llm_studio_structured.py): increase max_tokens default value to accommodate larger inputs test(redgifs): add unit tests for downloading media from Redgifs to ensure functionality and error handling test: add tests for direct download functionality from Redgifs to ensure reliability and correctness of media selection docs: update MediaSelection documentation to include support for direct Redgifs URLs in media sourcing options * feat: add script to fix ComfyUI base64 payloads by locating JPEG start and writing preview * feat(nodes): add ParseStringToBBox node to parse JSON-style bbox strings into float coordinates feat(nodes): update NODE_CLASS_MAPPINGS and NODE_DISPLAY_NAME_MAPPINGS to include ParseStringToBBox test: add unit tests for ParseStringToBBox to validate correct parsing and error handling * feat(parse_string_to_bbox): update bbox parsing to return integer coordinates and dimensions for better usability test(parse_string_to_bbox): update tests to validate new integer output format and add new test for rounding behavior test(parse_string_to_bbox_ints): add new tests for parsing bbox strings to integer coordinates and validate error handling * feat(docs): add detailed documentation for ParseStringToBBox node including inputs, outputs, and usage tips * ✨ (llm_studio_structured.py): introduce new structured description features for subject appearance and video frames 📝 (llm_studio_structured.py): update documentation for new input types and prompts ♻️ (llm_studio_structured.py): refactor node registration to include new combined structured describe class * ✨ (llm_studio_structured.py): introduce new SUBJECT_APPEARANCE_SCHEMA and VIDEO_FRAME_ARCHITECT_SCHEMA to enhance media description capabilities with detailed attributes for subject appearance and video frame analysis. * ✨ (media_describe): add LMStudioCombinedStructuredDescribe to exports for enhanced functionality ✨ (nodes): include LMStudioCombinedStructuredDescribe in mappings for improved node support * 📝 (llm_studio_structured.py): update prompts for clarity and detail to enhance user guidance and output structure ♻️ (llm_studio_structured.py): refactor combined text generation to improve readability and maintainability by using list comprehension * ✨ (llm_studio_structured.py): update schema presets and default values to enhance functionality ♻️ (llm_studio_structured.py): refactor frame extraction to return metadata instead of paths for better data handling 💡 (llm_studio_structured.py): improve content building logic to allow custom content for structured output 📝 (llm_studio_structured.py): update comments for clarity and maintainability of the code * ✨ (llm_studio_structured.py): add 'subject' field to VIDEO_FRAME_ARCHITECT_SCHEMA for better video description ♻️ (llm_studio_structured.py): update required fields in schema to include 'subject' and 'summary' for improved validation ♻️ (llm_studio_structured.py): refactor conditionals to handle new schema preset 'video_frame_architect' for enhanced functionality * ♻️ (llm_studio_structured.py): remove subject field from VIDEO_FRAME_ARCHITECT_SCHEMA to simplify schema requirements ♻️ (llm_studio_structured.py): update field assignments in LLMStudioStructuredDescribe to reflect schema changes * ✨ (llm_studio_structured.py): enhance NSFW summary handling by adding support for structured notes and summaries to improve content description clarity * ✨ (llm_studio_structured.py): add formatted_response to return types and names for improved output structure ♻️ (llm_studio_structured.py): refactor NSFW handling to separate plain and formatted responses for better clarity 💡 (llm_studio_structured.py): enhance response formatting for better readability and organization of output sections * ✨ (llm_studio_structured.py): add minItems and maxItems constraints to frames array to enforce validation rules on video frame data * 📝 (llm_studio_structured.py): add 'subject' property to VIDEO_FRAME_ARCHITECT_SCHEMA for better metadata representation 📝 (llm_studio_structured.py): update descriptions for NSFW analysis and frame metadata for clarity 📝 (llm_studio_structured.py): modify required fields in schema to include 'subject' and adjust 'has_nsfw' to 'summary' * ✨ (llm_studio_structured.py): add 'timestamp_seconds' and 'frames' to required fields to ensure necessary data is provided for video frame descriptions * ♻️ (llm_studio_structured.py): refactor frame index calculation to improve handling of invalid fps and ensure proper frame extraction ✨ (llm_studio_structured.py): add fallback for fps and enhance frame index tracking for better frame management * 💡 (llm_studio_structured.py): add debug logging for frame indices and file details to assist in troubleshooting 🐛 (llm_studio_structured.py): improve verbose logging for base64 encoding to catch potential issues with short encodings * ✨ (azure_queue_publisher.py, .env.example, config_api.py, nodes.py, swiss-army-knife.js): introduce Azure Queue Publisher node to publish job events to Azure Storage Queue for event-driven architecture 📝 (requirements.txt): add azure-storage-queue dependency to support Azure Queue operations 🔧 (config_api.py): update config API to include Azure job events queue in settings 💡 (swiss-army-knife.js): add function to retrieve Azure job events queue from settings for better integration ♻️ (various files): refactor code for improved readability and maintainability, including consistent formatting and logging updates * ✨ (routes): add event routes for emitting events and health check to enhance API functionality 🐛 (azure_queue_publisher.py): add prompt_id as an optional parameter to improve event data structure ♻️ (azure_queue_publisher.py): change character_index condition to allow zero value for better logic 📝 (init.py): register new event routes in the application to ensure proper routing * ✨ (events.py): enhance event route registration to prefer PromptServer API 🐛 (events.py): add fallback to aiohttp router and improve error handling 💡 (events.py): add diagnostics logging for registered routes in debug mode * ✨ (events.py): enhance event route registration with best-effort fallbacks to improve reliability and diagnostics in case of registration failures * ✨ (routes/events.py): refactor event route registration to use ComfyUI's recommended pattern for better maintainability and clarity 📝 (init.py): update event route registration to auto-import events module for automatic route initialization and improve error handling * ♻️ (events.py): refactor to remove os dependency and use Swiss Army Knife config system for connection string and queue name retrieval to improve maintainability and flexibility * ✨ (execution_hooks): add execution hooks for automatic event publishing to Azure Queue during workflow execution 📝 (init.py): import execution hooks to register them into ComfyUI execution lifecycle and handle import exceptions gracefully * ✨ (execution_hooks.py): refactor execution hooks to integrate with PromptServer's message bus for improved event handling 🐛 (execution_hooks.py): update event handling methods to manage prompt lifecycle and error messages more effectively 📝 (execution_hooks.py): enhance module documentation for clarity on functionality and usage * 💡 (execution_hooks.py): add comments to clarify the behavior of execution events and their handling * ✨ (events.py): optimize JSON serialization by using custom separators for smaller payloads 💡 (events.py): add logging for published queue payload to improve traceability * ✨ (media_selection.py): add support for Azure Blob Storage as a media source with validation and downloading functionality * ✨ (docs, config_api, media_describe): add support for configurable LM Studio base URL to enhance flexibility and usability 📝 (docs): update documentation to reflect changes in base_url input and its usage ♻️ (config_api, media_describe): refactor code to resolve LM Studio base URL from settings or environment variables for better maintainability * 📝 (STRUCTURED_OUTPUT_IMPLEMENTATION.md): update documentation to clarify LM Studio endpoint usage and remove base_url parameter from input definitions for consistency ♻️ (llm_studio_structured.py): refactor code to remove base_url parameter from describe_image and describe_video methods, relying on global LM Studio endpoint configuration ✨ (swiss-army-knife.js): add debug logging for model dropdown interactions to enhance traceability and debugging capabilities in the UI * ✨ (azure_queue_publisher.py, execution_hooks.py, events.py): add audit logging for event publishing to track success and failure 💡 (execution_hooks.py): refactor logging logic into a separate method for clarity and reuse 📝 (azure_queue_publisher.py, execution_hooks.py, events.py): improve code comments for better understanding of audit logging functionality * ♻️ (init.py, azure_queue_publisher.py, execution_hooks.py, events.py): refactor import paths to ensure package root is discoverable for modules 💡 (init.py, azure_queue_publisher.py, execution_hooks.py, events.py): add comments to clarify the purpose of modifying sys.path for module loading * ♻️ (init.py, azure_queue_publisher.py, execution_hooks.py, events.py): refactor import statements to use relative imports for better module organization and maintainability * ✨ (audit_logger_loader.py): introduce a helper module to load audit_logger 🐛 (azure_queue_publisher.py, execution_hooks.py, events.py): update import paths to use the new audit_logger_loader for better module management * ✨ (init.py): introduce dynamic loading of the lib package to handle hyphenated module names and improve import reliability ♻️ (audit_logger_loader.py): remove redundant audit logger loader module as its functionality is now integrated into __init__.py 🔧 (azure_queue_publisher.py, execution_hooks.py, events.py): update imports to reference the new lib package structure for audit logger * ✨ (audit_logger): introduce AuditLogger class for structured logging of job audit events ♻️ (imports): refactor import statements to use relative imports for consistency 📝 (gitignore): update .gitignore to exclude specific directories and files 🔧 (pyproject.toml): update package configuration to include "routes" package 📝 (init.py): remove unused code and clean up __init__.py for better readability * ✨ (app_insights.py): add Application Insights logging and tracing support to monitor application performance and errors ♻️ (events.py): integrate Application Insights logger into event emission process for better observability of events and errors * ♻️ (azure_queue_publisher.py, execution_hooks.py, audit_logger.py, events.py): refactor timestamp formatting to use 'Z' instead of '+00:00' for better compatibility with ISO 8601 ♻️ (azure_queue_publisher.py, execution_hooks.py, audit_logger.py): refactor JSON serialization to use custom separators for more compact output * ♻️ (execution_hooks.py, events.py): refactor queue name retrieval to use a fallback with or operator for better clarity and consistency * ✨ (azure_queue_publisher.py, execution_hooks.py, audit_logger.py, events.py): add TextBase64EncodePolicy to QueueClient initialization for better message encoding 💡 (azure_queue_publisher.py, execution_hooks.py, audit_logger.py, events.py): update timestamp formatting to include seconds for improved precision * ✨ (civit_metadata_helper.py): add optional JSON fields for subject and video prompts to enhance metadata collection * ✨ (civit_metadata_helper.py): add JSON parsing for subject and video prompts to handle structured data inputs 📝 (civit_metadata_helper.py): update return structure to include parsed subject and video prompts for better data handling * ✨ (init.py, config_api.py, upload_job_to_azure.py): introduce Upload Job to Azure node to upload files to Azure Blob Storage and create job documents in CosmosDB ♻️ (init.py): rename Azure Storage Upload node to Upload Job to Azure for clarity 📝 (upload-job-to-azure.md): add documentation for Upload Job to Azure node detailing its purpose, inputs, outputs, and configuration 🔧 (config_api.py): add configuration settings for Azure Cosmos DB connection and job container 🐛 (azure_storage_upload.py): remove obsolete Azure Storage Upload node code as it is replaced by the new Upload Job to Azure node * ✨ (execution_hooks.py, llm_studio_structured.py, cosmos_utils.py): add functionality to push structured analysis results to CosmosDB for job tracking 📝 (execution_hooks.py): add get_current_prompt_id function to retrieve the current job ID ♻️ (llm_studio_structured.py): refactor code to utilize current job ID for metadata updates in CosmosDB 📝 (cosmos_utils.py): add utility functions for interacting with Azure CosmosDB and Blob Storage * ♻️ (cosmos_utils.py): refactor container name from "media" to "media-files" for better clarity and organization * ✨ (llm_studio_structured.py): update keys in cosmos_meta to use snake_case for consistency 📝 (cosmos_utils.py): replace logger.debug with print statements for better visibility of errors and status messages * ✨ (Dockerfile, requirements.txt, swiss-army-knife.js): add Azure Cosmos DB support to enhance data storage capabilities and improve integration with Azure services. * 💡 (cosmos_utils.py): add comments to clarify the use of 'id' as partition key ✅ (cosmos_utils.py): update replace_item method to include partition_key for consistency * 💡 (cosmos_utils.py): update comments to clarify the use of jobId field as partition key 🐛 (cosmos_utils.py): fix partition key usage in replace_item to use jobId field value for consistency * ✨ (cosmos_utils.py): add support for PartitionKey to handle undefined partition keys in CosmosDB 🐛 (cosmos_utils.py): improve error handling and logging for document queries and updates 💡 (cosmos_utils.py): clarify comments regarding partition key discrepancies and operations
11 KiB
DEBUG Mode Implementation
Overview
The Swiss Army Knife extension supports a centralized debug mode setting that controls logging across both frontend (JavaScript) and backend (Python) code. Debug mode can be enabled via:
- Frontend Settings: ComfyUI Settings → Swiss Army Knife → Debug Mode (recommended)
- Environment Variable:
DEBUG=truein.env(backwards compatibility)
How It Works
Frontend Settings (Primary Method)
Setting Registration
The debug mode setting is registered in web/js/swiss-army-knife.js:
{
id: "SwissArmyKnife.debug_mode",
name: "Debug Mode",
type: "boolean",
defaultValue: false,
tooltip: "Enable debug mode for detailed logging",
onChange: (newVal, oldVal) => {
debugLog(`Debug mode changed from ${oldVal} to ${newVal}`);
syncApiKeysToBackend(); // Sync debug mode to backend
},
}
Dynamic Debug Checking
All extension JavaScript files check the setting dynamically:
// Check debug mode from settings
const isDebugEnabled = () => {
try {
return (
app.extensionManager?.setting?.get('SwissArmyKnife.debug_mode') ||
false
);
} catch (error) {
return false;
}
};
// Conditional logging wrapper
const debugLog = (...args) => {
if (isDebugEnabled()) {
console.log('[SwissArmyKnife]', ...args);
}
};
Backend (Python)
-
Config API Endpoint (
nodes/config_api.py):- Stores debug mode setting in
_cached_settingsglobal - Exposes
get_debug_mode()function for backend code - Syncs setting from frontend via
/swissarmyknife/set_api_keysendpoint
- Stores debug mode setting in
-
Debug Utilities (
nodes/debug_utils.py):def is_debug_enabled(): """ Check if debug mode is enabled. Checks both environment variable and frontend settings. """ # Check environment variable first (for backwards compatibility) env_debug = os.environ.get("DEBUG", "false").lower() in ("true", "1", "yes") if env_debug: return True # Check frontend settings try: from .config_api import get_debug_mode return get_debug_mode() except (ImportError, Exception): return False def debug_print(*args, **kwargs): """Print debug messages only when debug mode is enabled""" if is_debug_enabled(): print(*args, **kwargs)
Environment Variable (Legacy Support)
The DEBUG environment variable is still supported for backwards compatibility:
debug = os.environ.get("DEBUG", "false").lower() in ("true", "1", "yes")
- Supports:
true,1,yes(case-insensitive) for enabling debug mode - Defaults to
falseif not set
Frontend (JavaScript)
- Log Classification:
- Debug logs (controlled by debug mode setting):
debugLog()- Implementation details, state tracking, diagnostic info- All
[DEBUG],[SERIALIZE],[CONFIGURE],[LOADED],[API]tagged messages
- Always-visible logs (not controlled by debug mode):
console.log()- User-facing messages, version infoconsole.warn()- Warningsconsole.error()- Errors- Extension loading message
- Debug logs (controlled by debug mode setting):
Configuration
Enable Debug Mode (Recommended: Frontend Settings)
- Open ComfyUI Settings (gear icon)
- Navigate to "Swiss Army Knife" section
- Enable "Debug Mode" checkbox
- Debug logging activates immediately (no restart required)
Enable Debug Mode (Legacy: Environment Variable)
In .env:
DEBUG=true
Or any of:
DEBUG=1
DEBUG=yes
DEBUG=True
DEBUG=YES
Disable Debug Mode
Via Settings (Recommended):
- Open ComfyUI Settings
- Uncheck "Debug Mode"
- Debug logging stops immediately
Via Environment Variable:
In .env:
DEBUG=false
Or simply omit the DEBUG setting entirely (defaults to false).
Testing
-
Test Frontend Settings Method (Recommended):
- Enable debug mode via ComfyUI Settings
- Open browser console
- Should see debug logs with
[SwissArmyKnife]and[CivitAI]prefixes - Disable debug mode
- Debug logs should stop immediately
-
Verify Backend Sync:
# Check server console logs for: # [Swiss Army Knife] Settings cached: ... Debug=True -
Test Environment Variable Method (Legacy):
- Set
DEBUG=truein.env - Restart ComfyUI server
- Refresh browser
- Open browser console
- Should see: "Swiss Army Knife Debug Mode: ENABLED"
- Should see debug logs with
[DEBUG]tags
- Set
-
Test Debug Mode Disabled:
- Set
DEBUG=falsein.env - Restart ComfyUI server
- Refresh browser
- Open browser console
- Should see: "Swiss Army Knife Debug Mode: DISABLED"
- Should NOT see debug logs with
[DEBUG]tags
- Set
Implementation Details
Files Modified
-
__init__.py:- Added config API route registration
- Added
DEBUGto exports
-
nodes/config_api.py(new file):- Created config API endpoint
- Reads DEBUG from environment
-
web/js/swiss-army-knife.js:- Added
loadDebugConfig()async function - Added
debugLog()wrapper function - Replaced ~50+
console.log()calls withdebugLog()for debug-level messages - Kept user-facing logs as
console.log()
- Added
Logging Categories Changed to debugLog()
- Node registration messages
- Widget state serialization/deserialization
- File upload handlers
- Workflow loading
- API event processing
- Dimensions display updates
- Media widget management
- LoRA info extraction
Logging Categories Kept as console.log()
- Extension version and load timestamp
- Debug mode status (enabled/disabled)
- Development utility functions
- Version update notifications
- Cache control messages
- User-facing warnings and errors
Benefits
- Cleaner Console: Users can disable debug logs for cleaner console output in production
- Debugging: Developers can enable detailed logging when troubleshooting
- Centralized Control: Single environment variable controls all debug logging
- No Code Changes: Toggle debug mode without modifying code
- Server Restart: Changes to DEBUG setting require server restart (not just browser refresh)
Future Enhancements
Potential improvements:
- Add different log levels (DEBUG, INFO, WARN, ERROR)
- Add per-node debug control
- Add runtime debug toggle (no server restart needed)
- Add debug log filtering by category
- Add debug log export functionality
Final Statistics
After complete implementation:
- Total debugLog calls: 167 (converted from console.log)
- Remaining console.log calls: 15 (all user-facing or system-required)
- 1 inside debugLog function itself (required)
- 2 user-facing status messages (debug mode, extension loading)
- 2 error/status messages (filename preview error, video preview status)
- 10 development utility messages (cache busting, localhost only)
- Conversion rate: ~92% of logging now respects DEBUG flag
Log Categories Converted to debugLog
All these message types now respect the DEBUG environment variable:
[DEBUG]- Implementation details and diagnostics[STATE]- Widget state management[WIDGET]- Widget operations[CONFIGURE]- Configuration and restoration[SERIALIZE]- Workflow serialization[LOADED]- Workflow loading[UPLOAD]- File upload operations[API]- API event handling[LLMStudioStructuredDescribe]- Structured describe diagnostics- Widget discovery and manipulation
- File restoration and state management
- Dimensions display updates
User-Facing Messages (Always Visible)
These messages always appear regardless of DEBUG setting:
- Extension version and load timestamp
- Debug mode status (ENABLED/DISABLED)
- Development utilities info (localhost only)
- Critical error messages
- Cache clearing confirmations
Troubleshooting
"Still seeing debug logs after setting DEBUG=false"
-
Verify .env file:
cat /Users/samkumar/Development/dev-lab-hq/ai-image-hub/apps/comfyui-swiss-army-knife/.env | grep DEBUGShould show:
DEBUG=false -
Test config API:
curl http://127.0.0.1:8188/swissarmyknife/configShould return:
{"debug": false}If it returns
{"debug": true}, the .env file isn't being read correctly. -
Clear ALL caches:
- Stop ComfyUI server completely
- Clear browser cache (hard refresh: Cmd+Shift+R / Ctrl+Shift+F5)
- In browser console, run:
localStorage.clear(); sessionStorage.clear() - Or use the built-in utility:
clearSwissArmyKnifeCache() - Restart ComfyUI server
- Reload page
-
Check for cached JavaScript:
- Browser may have cached the old swiss-army-knife.js
- Force cache bypass: Disable browser cache in DevTools Network tab
- Or increment EXTENSION_VERSION in swiss-army-knife.js to force reload
-
Verify server restart:
- Ensure you fully restarted the ComfyUI server (not just refreshed browser)
- Environment variables are only read at server startup
"Debug mode status shows DISABLED but still see logs from other extensions"
The DEBUG flag only controls Swiss Army Knife logs. Other extensions (Node Enhancer, Super LoRA Loader, etc.) have their own logging that isn't affected by this setting.
Logs from other extensions will show different formats:
extension.js:6753 NodeEnhancer: Registering extension...extension.js:6891 Super LoRA Loader: Registering extension...
These are separate extensions and not controlled by Swiss Army Knife's DEBUG setting.
Final Statistics
After complete implementation:
- Total debugLog calls: 167 (converted from console.log)
- Remaining console.log calls: 15 (all user-facing or system-required)
- Conversion rate: ~92% of logging now respects DEBUG flag
User-Facing Messages (Always Visible)
These messages always appear regardless of DEBUG setting:
- Extension version and load timestamp
- Debug mode status (ENABLED/DISABLED)
- Development utilities info (localhost only)
- Critical error messages
Troubleshooting
"Still seeing debug logs after setting DEBUG=false"
The logs you're seeing are from other extensions (Node Enhancer, Super LoRA Loader). Swiss Army Knife debug logs will have tags like [DEBUG], [STATE], [WIDGET], etc.
Verification steps:
- Check browser console shows: "Swiss Army Knife Debug Mode: DISABLED"
- Verify no messages with
[DEBUG],[STATE],[WIDGET]tags appear - Other extensions' logs (extension.js:XXXX) are normal and unrelated
If Swiss Army Knife debug logs still appear:
- Verify
.envhasDEBUG=false - Fully restart ComfyUI server (environment variables load at startup)
- Hard refresh browser (Cmd+Shift+R / Ctrl+Shift+F5)
- Clear browser cache:
clearSwissArmyKnifeCache()in console