SkillAgentSearch skills...

Teamclaw

Teamclaw adopts the Auto-OASIS core: an automated, programmable multi‑Agent collaboration engine. Through simple YAML configuration under a main agent, users can define expert collaboration flows — supporting sequential, parallel, and complex programmable structures — running in a backend‑separated execution mode.

GenerateConvertImprove

Install / Use

/learn @Avalon-467/Teamclaw
About this skill

Quality Score

0/100

Supported Platforms

Universal

README

English | 中文

<a id="english"></a>

TeamClaw

TeamClaw Poster

An OpenAI-compatible local AI workspace with Teams, visual multi-agent orchestration, OASIS Town live mode, multimodal I/O, bots, scheduled tasks, and one-click public access.

Quick Start

Install via AI Code CLI

Open any AI coding assistant such as Codex, Cursor, Claude Code, CodeBuddy, or Trae, and say:

Clone https://github.com/Avalon-467/Teamclaw.git, read SKILL.md, and install TeamClaw.

That agent should then:

  1. Clone the repository
  2. Read SKILL.md
  3. Use docs/index.md to find the right docs
  4. Configure the environment and LLM settings
  5. Start the services

Manual Setup

<details> <summary>Click to expand manual setup</summary>

Linux / macOS

bash selfskill/scripts/run.sh setup
bash selfskill/scripts/run.sh configure --init

# If you already know the model:
bash selfskill/scripts/run.sh configure --batch \
  LLM_API_KEY=sk-xxx \
  LLM_BASE_URL=https://api.example.com \
  LLM_MODEL=<model>

# If you need model discovery:
bash selfskill/scripts/run.sh configure LLM_API_KEY sk-xxx
bash selfskill/scripts/run.sh configure LLM_BASE_URL https://api.example.com
bash selfskill/scripts/run.sh auto-model
bash selfskill/scripts/run.sh configure LLM_MODEL <model>

bash selfskill/scripts/run.sh start

For managed terminals, CI, or agent runners that reap child processes after the command exits, use bash selfskill/scripts/run.sh start-foreground and keep that session open instead of start.

Windows PowerShell

powershell -ExecutionPolicy Bypass -File .\selfskill\scripts\run.ps1 setup
powershell -ExecutionPolicy Bypass -File .\selfskill\scripts\run.ps1 configure --init

# If you already know the model:
powershell -ExecutionPolicy Bypass -File .\selfskill\scripts\run.ps1 configure --batch LLM_API_KEY=sk-xxx LLM_BASE_URL=https://api.example.com LLM_MODEL=<model>

# If you need model discovery:
powershell -ExecutionPolicy Bypass -File .\selfskill\scripts\run.ps1 configure LLM_API_KEY sk-xxx
powershell -ExecutionPolicy Bypass -File .\selfskill\scripts\run.ps1 configure LLM_BASE_URL https://api.example.com
powershell -ExecutionPolicy Bypass -File .\selfskill\scripts\run.ps1 auto-model
powershell -ExecutionPolicy Bypass -File .\selfskill\scripts\run.ps1 configure LLM_MODEL <model>

powershell -ExecutionPolicy Bypass -File .\selfskill\scripts\run.ps1 start

For managed terminals or automation that reap child processes when the command returns, use powershell -ExecutionPolicy Bypass -File .\selfskill\scripts\run.ps1 start-foreground and keep that session attached.

Open the UI at http://127.0.0.1:<PORT_FRONTEND>. On Windows, ports may be auto-remapped; trust config/.env or run.ps1 status.

</details>

Optional: Public Access

Use Cloudflare Tunnel when you explicitly want remote access:

python scripts/tunnel.py

Or start it via the TeamClaw run scripts / frontend settings panel.

TeamClaw combines a local /v1/chat/completions endpoint, a built-in multi-expert orchestration engine called OASIS, an optional OASIS Town live view in the chat tab, a full Web UI, and integrations such as OpenClaw, Telegram, QQ, audio I/O, scheduled tasks, and Cloudflare Tunnel. It supports any OpenAI-compatible provider — including Antigravity-Manager, a local reverse proxy that gives free access to 67+ models (Claude, Gemini, GPT) for users with a Google One Pro membership (e.g. via student verification), and MiniMax with its 1M-context M2.7 model.

It is designed for both:

  • people who want a powerful local AI control center
  • AI coding agents that can clone the repo, read SKILL.md, and install / operate it autonomously

Why TeamClaw

  • Team: unified multi-agent orchestration: combine internal agents, OpenClaw agents, and external API agents into a single Team — with one-click import/export of complete Team configurations
  • OpenAI-compatible from day one: expose a local /v1/chat/completions endpoint that works with standard clients and custom tools
  • Visual orchestration included: design workflows in OASIS, or save / run YAML workflows directly
  • Live observability built in: switch active discussions into OASIS Town and watch / nudge them in real time from the chat tab
  • Real operator features: settings UI, group chat, scheduled tasks, voice input, TTS, login tokens, and public tunnel support
  • Agent-first operations: SKILL.md + docs/index.md + docs/repo-index.md let other coding agents install and manage TeamClaw with progressive disclosure

What You Can Do Today

| Capability | What It Gives You | |---|---| | OpenAI-compatible API | Local chat completions endpoint for apps, tools, and clients | | Web UI | Chat, settings, OASIS panel, group chat, tunnel control | | OASIS workflows | Sequential, parallel, branching, and DAG-style expert orchestration | | OASIS Town | Turn a live OASIS topic into a pixel-town view in chat, with live residents, nudges, and ambient audio | | Team system | Public/private agents, personas, workflows, and Team snapshots | | OpenClaw + external agents | Bring in external runtimes and API-based agents | | Multimodal I/O | Images, files, voice input, TTS, provider-aware audio defaults | | Bots | Telegram and QQ integrations | | Automation | Scheduled tasks and long-running workflow execution | | Remote access | Cloudflare Tunnel plus login-token / password flows | | Import / export | Share or restore Teams and related assets |

Typical Use Cases

  • Local AI workspace: run a private AI assistant with a browser UI and OpenAI-compatible API
  • Team debate and execution: let multiple experts challenge, refine, and conclude on the same task
  • Live debate observability: watch an OASIS discussion as a pixel town in the chat tab and inject nudges while it is running
  • AI integration hub: connect bots, external agent runtimes, and other OpenAI-compatible clients
  • Operational cockpit: manage settings, ports, audio, workflows, public access, and users from one place

Product Highlights

OASIS Orchestration

OASIS is the engine that turns TeamClaw from a chatbot into a programmable multi-expert system.

  • combine stateless experts, stateful sessions, OpenClaw agents, and external API agents
  • run sequential, parallel, selector-based, or DAG-style workflows
  • support Team-level personas and reusable saved workflows
  • switch the current discussion into OASIS Town for a live pixel-town view inside the chat tab
  • monitor topics, conclusions, and session state from CLI or UI

Teams and Personas

Each Team can combine:

  • built-in lightweight internal agents
  • OpenClaw agents
  • external API agents
  • public and private expert personas
  • reusable workflows and Team snapshots

Bots, Audio, and Operations

TeamClaw is no longer just chat + orchestration. It also includes:

  • Telegram and QQ bot integration
  • voice input and text-to-speech
  • provider-aware audio defaults for OpenAI / Gemini-style setups
  • settings UI and restart flow
  • login tokens and password-based remote access
  • scheduled tasks and system-triggered execution

Acknowledgements

TeamClaw also benefited from several open-source projects:

Documentation Paths

Start with the level that matches your task:

Deep dives:

License

MIT License

<a id="中文"></a>

TeamClaw

一个 OpenAI 兼容的本地 AI 工作台：带 Team、多专家可视化编排、OASIS Town 实时模式、多模态输入输出、Bot、定时任务，以及一键公网访问。

快速开始

通过 AI Code CLI 安装

在 Codex、Cursor、Claude Code、CodeBuddy、Trae 之类的 AI 编码助手里输入：

Clone https://github.com/Avalon-467/Teamclaw.git，读取 SKILL.md，然后安装 TeamClaw。

正常情况下，这个 Agent 会自动：

  1. 克隆仓库
  2. 阅读 SKILL.md
  3. 通过 docs/index.md 找到需要的文档
  4. 配置环境和 LLM
  5. 启动服务

手动安装

<details> <summary>点击展开手动安装步骤</summary>

Linux / macOS

bash selfskill/scripts/run.sh setup
bash selfskill/scripts/run.sh configure --init

# 如果已经知道模型：
bash selfskill/scripts/run.sh configure --batch \
  LLM_API_KEY=sk-xxx \
  LLM_BASE_URL=https://api.example.com \
  LLM_MODEL=<model>

# 如果还不知道模型：
bash selfskill/scripts/run.sh configure LLM_API_KEY sk-xxx
bash selfskill/scripts/run.sh configure LLM_BASE_URL https://api.example.com
bash selfskill/scripts/run.sh auto-model
bash selfskill/scripts/run.sh configure LLM_MODEL <model>

bash selfskill/scripts/run.sh start

如果你所在的受管终端、CI 或 agent runner 会在命令返回后清理子进程，请改用 `bash selfski

Avalon-467

View profile

View on GitHub
GitHub Stars8
CategoryCustomer
Updated4d ago
Forks3
Avalon-467/Teamclaw

Languages

JavaScript

Security Score

85/100

Audited on Mar 31, 2026

No findings