ComfyUI LLM Party
NormalDescription
LLM Agent framework within ComfyUI integrating MCP server, TTS, OCR, GraphRAG, and other AI tool nodes for visual workflow building
Key Features
- Complete set of LLM workflow nodes for ComfyUI including API, local, GGUF, and VLM loaders
- MCP server integration for connecting external tools via modelcontextprotocol.json configuration
- Multi-agent pipeline support with agent-agent radial and ring interaction patterns
- Industry-specific RAG capabilities including word vector RAG and GraphRAG for knowledge bases
- TTS integration, OCR, and image hosting nodes for end-to-end multimedia workflows
- Streaming output mode with real-time console display and R1 model reasoning_content separation
Use Cases
💡 Building custom AI assistants with role settings and multi-tool calls
💡 Creating industry-specific knowledge bases using GraphRAG for enterprise document management
💡 Generating SD prompts and images through local VLM or API-based vision models
💡 Connecting social apps (QQ, Feishu, Discord) for individual users
💡 Integrating LLM + TTS + ComfyUI workflows for streaming content creation
Categories
Quick Start
1. Download the Windows portable package (includes LLM Party + Manager plugins) or install into existing ComfyUI via ComfyUI-Manager
2. Drag workflow JSON files into ComfyUI: start_with_LLM_api.json (API), start_with_Ollama.json (local Ollama), or start_with_LLM_local.json (local model)
3. Fill in base_url and api_key in the API LLM loader node (for API) or enable is_ollama option (for Ollama)
4. For local models, specify the model path or HuggingFace repo ID in the local model loader node