heshengtao/super-agent-party

⭐ All-in-one AI companion! Super Agent Party = Self hosted neuro sama + openclaw! ⭐ 全能AI伴侣！超级智能体派对 = 自托管neuro sama + openclaw!

Super Agent Party – An AI‑powered desktop companion

What it is

A cross‑platform desktop application that lets large‑language‑model agents interact with you through a visual avatar (VRM 3‑D models or Live2D THA heads).
The agents can control the computer (mouse, keyboard, terminal), browse the web, send messages on many IM platforms, and even stream live video.
It ships as ready‑to‑run binaries for Windows, macOS (M‑chip), Linux AppImage, Docker images, and a source‑code version for developers.

Key capabilities

Feature	What you get
Custom avatars	Load any VRM 3‑D model or Live2D THA head, define actions, expressions and voice.
Computer vision & control	The agent can see the screen, recognise UI elements and issue mouse/keyboard commands (e.g., playing Wordle).
Multi‑window chat UI	Phone‑size, capsule‑dock, or full‑screen windows; custom backgrounds, emoji packs, persona cards.
Group chat & memory	Talk with several characters at once, each with long‑term memory and “tavern” cards.
Task Center & Skills	Define reusable skills (MCP extensions) that let the agent perform complex background jobs.
Instant‑messenger bots	One‑click deployment to QQ, WeChat, Feishu, DingTalk, Telegram, Discord, Slack.
Live‑stream bots	Stream the avatar to Bilibili, YouTube, Twitch, with optional 360° panoramic mode.
AI‑driven browser	An embedded browser that the agent can navigate and control automatically.
Extension system	Install or create plugins (e.g., story adventure, AI editor, galgame, RSS reader, code server).
OpenAI‑compatible API	Expose a local `/v1` endpoint that works with the official OpenAI Python client, enabling any existing code to talk to the local agent.
MCP interface	Machine‑Control‑Protocol endpoint for other tools to invoke the agent’s actions.

How to get it

Windows: portable zip (no install) or an installer (.exe).
macOS (M‑chip): portable zip or a signed .dmg bundle.
Linux: AppImage for x86_64 or arm64.
Docker: pull ailm32442/super-agent-party (or the Alibaba Cloud mirror) and run with a mounted data folder; a browser‑only UI is served at http://localhost:3456.
Source: git clone … && uv sync && npm install && npm run dev for developers who want to modify the code.

First‑time setup

Remove macOS quarantine (xattr) if needed.
Make the helper scripts executable (chmod +x 一键更新(update).sh 一键启动(start).sh).
Run the update script to fetch the latest models and dependencies, then start the app.

Typical workflows

Personal assistant – talk to a 3‑D avatar, ask it to schedule meetings, browse the web, or control your PC.
Content creator – use the live‑stream bot to broadcast a talking‑head avatar on YouTube/Twitch while the AI narrates.
Developer – call the local OpenAI‑compatible endpoint from your own code, or write custom extensions (plugins are listed on the built‑in marketplace).
Team collaboration – deploy the Docker version on a server, expose the UI to teammates, and let the AI act as a shared “assistant” in Slack or Discord.

Hardware requirements

Minimum: 2 CPU cores, 2 GB RAM.
All AI models are optional; you can point the app at a local inference engine (e.g., Ollama) or a cloud provider, keeping the resource footprint low.

Limitations / things to watch

The desktop UI currently runs only on Windows 10/11, macOS M‑chip, and Linux AppImage; older OS versions may not be supported.
Full computer‑control features (/yolo, /cowork, etc.) require the CLI Agent component and appropriate permissions.
Some integrations (VTS, certain IM platforms) depend on third‑party software that must be installed separately.
The Docker variant provides a browser‑only view; the native desktop pet experience needs the native binaries.

Where to learn more

All information above is taken directly from the repository’s README; no external assumptions have been added.