Super Agent Party is a 3D AI desktop companion designed for extensive customization and integration, targeting developers and power users seeking to build sophisticated AI agents. It offers a pluggable platform for enhancing LLM capabilities with features like RAG, web search, memory, code execution, and multimodal support, while enabling deployment across various terminals including WeChat, QQ, and VRM avatars.

How It Works

The project acts as an LLM enhancement platform, allowing seamless integration of diverse functionalities without code modification. It supports local LLM engines and cloud vendor interfaces, providing an OpenAI API-compatible interface for easy integration. Its architecture facilitates cross-platform capability aggregation and acts as a tool chain for systems like ComfyUI, MCP, and A2A, enabling asynchronous tool calls without blocking agent replies.

Quick Start & Requirements

• Installation: Desktop installers for Windows, macOS (beta), and Linux (.AppImage, .deb). Docker image available (ailm32442/super-agent-party:latest). Source code deployment via git clone and uv sync, npm install.
• Access: Localhost:3456 for web UI, or via OpenAI-compatible API calls.
• Prerequisites: Docker for containerized deployment. Python environment for source builds.
• Documentation: 中文文档, English doc.

Highlighted Details

• Supports RAG, web search, long-term memory, code interpreters, MCP, A2A, Comfyui, WeChat, QQ, and Bilibili live bots.
• Offers an OpenAI API-compatible interface for easy integration and secondary development.
• Enables VRM virtual desktop pets and WeChat/QQ bot functionalities.
• Facilitates ComfyUI workflow integration as agent tools with load balancing.

Maintenance & Community

• Community: QQ Group: 931057213, WeChat Group: we_glm, Discord link available.
• Support: Encourages stars and donations.

Licensing & Compatibility

• License: Dual licensing: AGPLv3 by default. Commercial license required for closed-source commercial use.
• Compatibility: Compatible with Windows, macOS, Linux, and Docker.

Limitations & Caveats

macOS desktop installation is in beta. Commercial use in closed-source applications requires a separate license, and the default AGPLv3 license may impose copyleft restrictions on derivative works.