This project provides a comprehensive framework for building Large Language Model (LLM) workflows within ComfyUI, targeting users who want to integrate advanced AI capabilities into their existing image generation pipelines. It offers a wide array of nodes for multi-agent interactions, RAG (Retrieval-Augmented Generation), and social app integration, enabling the creation of sophisticated AI assistants and specialized workflows.

How It Works

The framework extends ComfyUI with custom nodes that abstract complex LLM interactions. It supports various LLM backends, including OpenAI-compatible APIs, Ollama for local models, and direct loading of Hugging Face or GGUF formats. Key features include agent-to-agent communication patterns (radial, ring), integration with TTS (Text-to-Speech) and OCR, and support for multimodal models (VLMs). This approach allows for flexible and modular construction of LLM-powered applications directly within a visual programming environment.

Quick Start & Requirements

• Installation: Install via ComfyUI Manager or by cloning the repository into ComfyUI's custom_nodes folder.
• Dependencies: Run pip install -r requirements.txt within the project directory. Specific models may require updated libraries (e.g., pip install -U transformers).
• Configuration: API keys, base URLs, and model names can be set in config.ini or directly within ComfyUI nodes.
• Resources: A Windows portable package is available for users new to ComfyUI.
• Documentation: Tutorials and example workflows are provided.

Highlighted Details

• Supports a broad range of LLMs including OpenAI, Ollama, Gemini, Grok, Qwen, GLM, DeepSeek, Kimi, and Doubao.
• Includes nodes for industry-specific RAG and GraphRAG for localized knowledge base management.
• Features integration with social platforms like Feishu, Discord, and QQ.
• Offers streaming output for LLM API nodes and a reasoning_content output for separating model reasoning from responses.

Maintenance & Community

• Active development with regular updates, including streaming API output and new VLM support.
• Community support available via QQ group (931057213) and Discord.
• Project updates are shared on Bilibili and YouTube.

Licensing & Compatibility

• The project appears to be open-source, with some nodes borrowing from other projects. Specific licensing details for the core project are not explicitly stated in the README, but it is generally compatible with ComfyUI's open-source nature.

Limitations & Caveats

• The Windows portable package is exclusively for Windows OS.
• Users are advised to read the project homepage thoroughly due to high usage thresholds.
• Some image hosting services mentioned may have had API changes or discontinuations.