comfyui_LLM_party

Why we included this project

ComfyUI users who want LLM and agent steps alongside their image work get a practical fit here. The project provides a large set of nodes for model calls, tool use, multi-agent loops, MCP connections, and RAG or GraphRAG knowledge bases, letting you assemble an assistant by dragging blocks instead of writing glue code. One graph can reach hosted models through OpenAI-compatible or aisuite endpoints, run local and GGUF models, and add vision and speech steps like OCR and TTS. It can also push results out to chat apps such as Feishu and Discord. The real value is keeping models, tools, memory, and media generation in a single reusable workflow that plugs into the image pipelines you already have.