4.5 KiB
type, title, description, resource, tags, timestamp
| type | title | description | resource | tags | timestamp | ||||
|---|---|---|---|---|---|---|---|---|---|
| Guide | Project Overview | High-level project summary, environment setup, directory structure, and asset references for Sage Utils. | Unsloth_Project_Docs.md |
|
2026-07-01T00:00:00Z |
This concept documents the overall Sage Utils project, including its purpose, installation layout, environment setup, and the locations of key assets and examples.
Project Summary
Sage Utils is a ComfyUI custom node pack that adds:
- 100+ custom nodes
- LLM integration with Ollama, LM Studio, and OpenAI-compatible backends
- Prompt builder and AI-assisted prompt generation
- Image gallery and metadata extraction
- Model management, scanning, and CivitAI metadata support
- File and notes management
- A rich sidebar interface and cross-tab workflows
Environment & Setup
- ComfyUI is installed at
~/ai/programs/comfyui/. - Sage Utils lives at
~/ai/programs/comfyui/custom_nodes/comfyui_sageutils/. - Use the ComfyUI Python environment and launch via
start.sh. - ComfyUI is typically available at
http://127.0.0.1:8188/. - User data is stored under
~/ai/programs/comfyui/user/default/SageUtils/.
Directory Structure
The repository is organized into the following top-level directories:
assets/— static JSON and markdown configuration filesdocs/— development and architecture documentationexample_workflows/— JSON workflows demonstrating common use casesjs/— frontend UI and sidebar codenodes/— Python node definitions using ComfyUI v3 APIutils/— Python utilities and backend helpersroutes/— aiohttp backend endpointstests/— Python test suite
JavaScript subdirectories
The js/ folder is further organized into:
js/components/— reusable UI components and interface elementsjs/shared/— shared JavaScript utilities and infrastructurejs/sidebar/— sidebar tab orchestration and tab-specific modulesjs/nodes/— node-specific frontend UI helpersjs/promptBuilder/— prompt builder workflow modulesjs/llm/— LLM client, provider management, and conversation statejs/gallery/— gallery UI and image transfer logicjs/file/— file browser and text editor components
Documentation and reference
docs/contains design notes, architecture guides, and implementation plans.docs/ref_docs/holds developer reference materials and walkthroughs.okf/contains the structured, agent-friendly knowledge bundle used as the canonical reference for project organization, architecture, and documentation.js/docs/is the local node documentation source for custom nodes and should be considered when authoring or updating Sage Utils node docs.
Assets & Static Data
Key asset files include:
config.json— legacy settings and provider URLsdefault_tag_library.json— tag library for Prompt Builderllm_integration_profiles.json— provider tool/MCP profilesllm_prompts.json— LLM prompt templatesmetadata_templates.json— metadata output templatessage_styles.json— UI styling overridessystem_prompt.md— system prompt for image analysis
Example Workflows
Example workflows are stored in example_workflows/ and include:
- Ace 1.5 Example
- Chroma Example
- Flux Dev Example
- LLM Example
- Lumina 2 Example
- NetaYume Lumina Example
- NoobXL Example
- Pony Example / Pony Upscaling Example
- Z-Image Example
These examples are intended as starting templates for users and developers.
Testing
The repository includes a Python test suite for backend and provider functionality. Key test modules include:
test_llm_service.pytest_llm_registry.pytest_llm_provider_keys.pytest_llm_routes_helpers.pytest_llm_routes_contract.pytest_llm_compat.pytest_llm_init.pytest_llm_provider_availability.pytest_ollama_capabilities.pytest_ollama_tool_loop.pytest_logger.py
Documentation
Primary documentation lives in docs/ and includes:
ARCHITECTURE.mdLLM_ARCHITECTURE_NOTE.mdLOGGING.mddocs/planning/PHASE1_LOGGING_COMPLETE.mddocs/planning/PHASE2_COMPLETE.mdTAB_MANAGER_GUIDE.mdPROMPT_BUILDER_GUIDE.mdLLM_TAB_GUIDE.md- UI component guides (
BUTTON_COMPONENT_GUIDE.md,FORM_ELEMENTS_GUIDE.md,LAYOUT_COMPONENT_GUIDE.md) okf/docs/civitai_api.md— CivitAI REST API reference for model metadata and integration.
Notes
This concept is intended as a starting point for developers new to the project and provides the high-level information they need to navigate the codebase.