TeamClaw

分类: 工具与效率 | 上传者: Avalon-467 | 下载: 0 | 版本: v1.0（最新）

A multi-agent orchestration platform with visual workflow (OASIS). Create and configure agents (OpenClaw/external API), orchestrate them into Teams, design workflows via visual canvas. Supports Team conversations, scheduled tasks, Telegram/QQ bots, and Cloudflare Tunnel for remote access.

更新日志: Source: GitHub https://github.com/Avalon-467/Teamclaw

目录结构

当前层级: 根目录

📁 chatbot/
- 📄 QQbot.py 8.0 KB
- 📄 setup.py 8.8 KB
- 📄 telegrambot.py 9.8 KB
📁 config/
- 📄 .env.example 3.5 KB
- 📄 .gitkeep 0 B
- 📄 requirements.txt 364 B
- 📄 users.json.example 88 B
📁 data/
- 📁 prompts/
  - 📁 agency_agents/
    
    📁 design/
    
    📄 design-brand-guardian.md 11.3 KB
    
    📄 design-image-prompt-engineer.md 10.8 KB
    
    📄 design-inclusive-visuals-specialist.md 6.5 KB
    
    📄 design-ui-designer.md 12.9 KB
    
    📄 design-ux-architect.md 15.3 KB
    
    📄 design-ux-researcher.md 12.6 KB
    
    📄 design-visual-storyteller.md 7.2 KB
    
    📄 design-whimsy-injector.md 15.4 KB
    
    📁 engineering/
    
    📄 engineering-ai-engineer.md 7.1 KB
    
    📄 engineering-autonomous-optimization-architect.md 7.7 KB
    
    📄 engineering-backend-architect.md 9.2 KB
    
    📄 engineering-data-engineer.md 14.2 KB
    
    📄 engineering-devops-automator.md 12.5 KB
    
    📄 engineering-frontend-developer.md 9.0 KB
    
    📄 engineering-mobile-app-builder.md 16.6 KB
    
    📄 engineering-rapid-prototyper.md 15.0 KB
    
    📄 engineering-security-engineer.md 11.5 KB
    
    📄 engineering-senior-developer.md 6.0 KB
    
    📄 engineering-technical-writer.md 13.8 KB
    
    📁 marketing/
    
    📄 marketing-app-store-optimizer.md 12.7 KB
    
    📄 marketing-content-creator.md 3.0 KB
    
    📄 marketing-growth-hacker.md 2.9 KB
    
    📄 marketing-instagram-curator.md 6.3 KB
    
    📄 marketing-reddit-community-builder.md 7.2 KB
    
    📄 marketing-social-media-strategist.md 7.1 KB
    
    📄 marketing-tiktok-strategist.md 7.3 KB
    
    📄 marketing-twitter-engager.md 7.5 KB
    
    📄 marketing-wechat-official-account.md 9.5 KB
    
    📄 marketing-xiaohongshu-specialist.md 9.3 KB
    
    📄 marketing-zhihu-strategist.md 11.6 KB
    
    📁 product/
    
    📄 product-behavioral-nudge-engine.md 5.1 KB
    
    📄 product-feedback-synthesizer.md 7.0 KB
    
    📄 product-sprint-prioritizer.md 9.1 KB
    
    📄 product-trend-researcher.md 10.2 KB
    
    📁 project-management/
    
    📄 project-management-experiment-tracker.md 9.2 KB
    
    📄 project-management-project-shepherd.md 9.5 KB
    
    📄 project-management-studio-operations.md 9.6 KB
    
    📄 project-management-studio-producer.md 10.6 KB
    
    📄 project-manager-senior.md 5.1 KB
    
    📁 spatial-computing/
    
    📄 macos-spatial-metal-engineer.md 11.8 KB
    
    📄 terminal-integration-specialist.md 3.9 KB
    
    📄 visionos-spatial-engineer.md 3.2 KB
    
    📄 xr-cockpit-interaction-specialist.md 1.7 KB
    
    📄 xr-immersive-developer.md 1.7 KB
    
    📄 xr-interface-architect.md 1.6 KB
    
    📁 specialized/
    
    📄 agentic-identity-trust.md 16.7 KB
    
    📄 agents-orchestrator.md 15.3 KB
    
    📄 data-analytics-reporter.md 2.9 KB
    
    📄 data-consolidation-agent.md 2.3 KB
    
    📄 lsp-index-engineer.md 10.8 KB
    
    📄 report-distribution-agent.md 2.5 KB
    
    📄 sales-data-extraction-agent.md 2.6 KB
    
    📄 specialized-cultural-intelligence-strategist.md 6.4 KB
    
    📄 specialized-developer-advocate.md 14.4 KB
    
    📁 support/
    
    📄 support-analytics-reporter.md 15.3 KB
    
    📄 support-executive-summary-generator.md 8.7 KB
    
    📄 support-finance-tracker.md 18.7 KB
    
    📄 support-infrastructure-maintainer.md 21.8 KB
    
    📄 support-legal-compliance-checker.md 25.8 KB
    
    📄 support-support-responder.md 24.0 KB
    
    📁 testing/
    
    📄 testing-accessibility-auditor.md 14.9 KB
    
    📄 testing-api-tester.md 12.2 KB
    
    📄 testing-evidence-collector.md 8.1 KB
    
    📄 testing-performance-benchmarker.md 11.8 KB
    
    📄 testing-reality-checker.md 9.8 KB
    
    📄 testing-test-results-analyzer.md 14.2 KB
    
    📄 testing-tool-evaluator.md 17.0 KB
    
    📄 testing-workflow-optimizer.md 21.5 KB
  - 📄 agency_experts.json 30.0 KB
  - 📄 base_system.txt 6.3 KB
  - 📄 base_system_subagent.txt 1.5 KB
  - 📄 oasis_expert_discuss.txt 786 B
  - 📄 oasis_experts.json 3.3 KB
  - 📄 oasis_summary.txt 415 B
  - 📄 oasis_trigger.txt 308 B
  - 📄 system_trigger.txt 338 B
  - 📄 tool_status.txt 274 B
- 📄 .gitkeep 0 B
📁 docs/
- 📄 cli.md 16.2 KB
- 📄 ports.md 11.0 KB
📁 logs/
- 📄 .gitkeep 0 B
📁 oasis/
- 📄 __init__.py 46 B
- 📄 engine.py 41.9 KB
- 📄 experts.py 53.5 KB
- 📄 forum.py 9.0 KB
- 📄 models.py 2.4 KB
- 📄 scheduler.py 21.0 KB
- 📄 server.py 70.1 KB
📁 packaging/
- 📄 build.py 2.2 KB
- 📄 build_dmg.sh 11.7 KB
- 📄 launcher.py 1.0 KB
📁 scripts/
- 📄 adduser.sh 212 B
- 📄 cli.py 57.3 KB
- 📄 launcher.py 9.3 KB
- 📄 setup_apikey.sh 5.5 KB
- 📄 setup_env.sh 1.8 KB
- 📄 start.sh 326 B
- 📄 tunnel.py 9.8 KB
- 📄 tunnel.sh 300 B
📁 selfskill/
- 📁 scripts/
  - 📄 adduser.py 1.3 KB
  - 📄 configure.py 16.3 KB
  - 📄 configure_openclaw.py 18.6 KB
  - 📄 run.sh 15.8 KB
📁 src/
- 📁 static/
  - 📁 css/
    
    📄 style.css 64.3 KB
  - 📁 js/
    
    📄 app.js 14.7 KB
    
    📄 main.js 266.0 KB
    
    📄 orchestration.js 182.1 KB
- 📁 templates/
  - 📄 index.html 54.0 KB
- 📄 agent.py 36.3 KB
- 📄 api_patch.py 4.2 KB
- 📄 front.py 107.6 KB
- 📄 llm_factory.py 5.7 KB
- 📄 mainagent.py 78.3 KB
- 📄 mcp_commander.py 13.1 KB
- 📄 mcp_filemanager.py 4.3 KB
- 📄 mcp_llmapi.py 12.6 KB
- 📄 mcp_oasis.py 58.8 KB
- 📄 mcp_scheduler.py 3.6 KB
- 📄 mcp_search.py 2.0 KB
- 📄 mcp_session.py 6.0 KB
- 📄 mcp_telegram.py 8.5 KB
- 📄 time.py 5.5 KB
📁 test/
- 📄 chat.py 2.8 KB
- 📄 view_history.py 3.7 KB
📁 tools/
- 📄 gen_password.py 1.5 KB
📁 visual/
- 📄 main.py 43.2 KB
📄 .gitignore 841 B
📄 LICENSE 1.0 KB
📄 README.md 62.6 KB
📄 run.sh 2.4 KB
📄 SKILL.md 7.1 KB

SKILL.md

---
name: "TeamClaw"
description: "A multi-agent orchestration platform with visual workflow (OASIS). Create and configure agents (OpenClaw/external API), orchestrate them into Teams, design workflows via visual canvas. Supports Team conversations, scheduled tasks, Telegram/QQ bots, and Cloudflare Tunnel for remote access."
user-invokable: true
compatibility:
  - "deepseek"
  - "openai"
  - "gemini"
  - "claude"
  - "anthropic"
  - "ollama"

argument-hint: "[REQUIRED] LLM_API_KEY, LLM_BASE_URL, LLM_MODEL. [OPTIONAL] TTS_MODEL/TTS_VOICE, OPENCLAW_*, TELEGRAM_BOT_TOKEN/QQ_APP_ID, PORT_*. [TUNNEL] PUBLIC_DOMAIN (user must explicitly request). Agent MUST NOT auto-download or start tunnel."

metadata:
  version: "1.0.2"
  github: "https://github.com/Avalon-467/Teamclaw"
  ports:
    agent: 51200
    scheduler: 51201
    oasis: 51202
    frontend: 51209
  auth_methods:
    - "user_password"
    - "internal_token"
    - "chatbot_whitelist"
  integrations:
    - "openclaw"
    - "telegram"
    - "qq"
    - "cloudflare_tunnel"
---

# TeamClaw

## 是什么

TeamClaw 是一个**多 Agent 编排平台**，核心能力：

1. **可视化编排团队 Agent**
   - 将 OpenClaw Agent、TeamClaw 内置 Agent 或任意外部 API Agent 编排为"团队"
   - 通过 Web UI 画布（OASIS）拖拽添加专家，设置协作关系
   - **Expert（专家）**：人设，是特殊的 prompt，定义 Agent 的角色和能力
   - **Agent（智能体）**：具有工具、技能和提示词的实体，可执行具体任务
   - **OpenClaw Agent**：可添加、配置可用工具、可用 Skill、提示词等
   
2. **团队工作流**
   - 多AGENT并行讨论/执行，汇总结论
   - **状态图编排**：支持串行、并行、选择、循环

3. **便携分享**
   - 将团队配置导出为压缩包，一键分享给他人
   - 导入他人分享的团队配置，快速复用

## 内置轻量级 Agent

TeamClaw 内置了**轻量级 Agent**（类似简化版 OpenClaw）：

| 能力 | 说明 |
|------|------|
| **文件管理** | 读取、写入、搜索文件 |
| **指令执行** | 执行 shell 命令 |
| **社交媒体** | 与 Telegram/QQ 用户沟通 |

- **更轻量**：相比 OpenClaw，prompt 更简洁、工具集更精简
- **快速启动**：无需安装 OpenClaw，开箱即用
- **可扩展**：可添加自定义工具和 Skill

| 功能 | 说明 |
|------|------|
| **Team 对话** | 多专家并行讨论/执行，支持 4 种专家类型 |
| **OASIS 工作流** | 可视化画布编排，支持驱动并行 AGENT TEAM|
| **定时任务** | APScheduler 任务调度中心 |
| **Web UI** | 完整聊天界面，支持 127.0.0.1 免密登录 |
| **Bot 集成** | Telegram / QQ Bot（可选配置） |
| **公网访问** | Cloudflare Tunnel（用户明确要求时启用） |

---

## 快速启动

```bash
# 1. 安装依赖
bash selfskill/scripts/run.sh setup

# 2. 初始化配置
bash selfskill/scripts/run.sh configure --init

# 3. 配置 LLM（必填）
bash selfskill/scripts/run.sh configure --batch \
  LLM_API_KEY=sk-xxx \
  LLM_BASE_URL=https://api.deepseek.com \
  LLM_MODEL=deepseek-chat

# 4. 启动服务
bash selfskill/scripts/run.sh start
```

**启动后访问**：`http://127.0.0.1:51209`

---

## 用户账号

### 免密登录（默认，首次运行推荐）

- 无需运行 `add-user`
- 询问用户设置用户名，或者用户可以使用默认 `admin`，"请设置您统领agent team的身份"
- 通过 **127.0.0.1** 访问 Web UI → 自动免密登录
- 密码登录不可用

### 密码登录（可选）

```bash
bash selfskill/scripts/run.sh add-user <用户名> <密码>
```

> ⚠️ 执行前必须先询问用户想要的用户名和密码

---

## 高权限功能

### 新建用户

```bash
bash selfskill/scripts/run.sh add-user <username> <password>
# 可创建多个用户
```

### OpenClaw 集成（可视化工作流）

```bash
# 检测/安装 OpenClaw
bash selfskill/scripts/run.sh check-openclaw

# 如果未安装，会提示安装（需要用户确认）
# 安装后自动配置 OPENCLAW_API_URL
```

**OASIS 画布中配置 OpenClaw Agent：**
- 拖拽 OpenClaw 专家到画布
- model 格式：`agent:<agent_name>:<session_name>`
- 示例：`agent:main:default`

### 配置管理

```bash
# 查看配置
bash selfskill/scripts/run.sh configure --show

# 修改配置
bash selfskill/scripts/run.sh configure <KEY> <VALUE>

# 批量设置
bash selfskill/scripts/run.sh configure --batch KEY1=val1 KEY2=val2
```

#### 必填配置

首次启动必须配置：

| 变量 | 说明 | 默认值 |
|------|------|--------|
| `LLM_API_KEY` | LLM 服务商 API 密钥 | - |
| `LLM_BASE_URL` | LLM API 基础地址 | `https://api.deepseek.com` |
| `LLM_MODEL` | 模型名称（用户未指定时自动检测） | `deepseek-chat` |

#### 选填配置（需要时再配置）

Agent 询问用户"是否需要拓展设置"后，按需配置：

| 变量 | 说明 | 默认值 |
|------|------|--------|
| **TTS 语音** | | |
| `TTS_MODEL` | TTS 模型 | - |
| `TTS_VOICE` | TTS 声音 | - |
| **OpenClaw 集成** | | |
| `OPENCLAW_API_URL` | OpenClaw Gateway 地址 | 自动探测 |
| `OPENCLAW_GATEWAY_TOKEN` | OpenClaw 认证令牌 | 自动探测 |
| `OPENCLAW_SESSIONS_FILE` | OpenClaw sessions 文件路径 | 自动探测 |
| **Bot 集成** | | |
| `TELEGRAM_BOT_TOKEN` | Telegram Bot Token | - |
| `TELEGRAM_ALLOWED_USERS` | Telegram 白名单用户 ID | - |
| `QQ_APP_ID` | QQ Bot App ID | - |
| `QQ_BOT_SECRET` | QQ Bot Secret | - |
| `QQ_BOT_USERNAME` | QQ Bot 用户名 | - |
| `AI_MODEL_TG` | Telegram Bot 使用的模型 | `LLM_MODEL` |
| `AI_MODEL_QQ` | QQ Bot 使用的模型 | `LLM_MODEL` |
| `AI_API_URL` | Bot 调用的 AI API 地址 | `LLM_BASE_URL` |
| **高级选项** | | |
| `PORT_AGENT` | Agent 主服务端口 | `51200` |
| `PORT_SCHEDULER` | 定时任务端口 | `51201` |
| `PORT_OASIS` | OASIS 工作流端口 | `51202` |
| `PORT_FRONTEND` | Web UI 端口 | `51209` |
| `OPENAI_STANDARD_MODE` | OpenAI 兼容模式 | `false` |

> 默认值说明："自动探测"表示运行 `check-openclaw` 等命令时自动配置；"自动生成"表示首次启动时系统生成。

### 启动/停止

```bash
bash selfskill/scripts/run.sh start     # 启动（后台）
bash selfskill/scripts/run.sh status    # 查看状态
bash selfskill/scripts/run.sh stop      # 停止
```

### 公网隧道（用户明确要求时）

```bash
bash selfskill/scripts/run.sh start-tunnel   # 启动隧道
bash selfskill/scripts/run.sh stop-tunnel    # 停止隧道
```

> ⚠️ Agent 禁止主动启动隧道，必须用户明确要求

---

## API 概览

**Base URL**: `http://127.0.0.1:51200`

```bash
# 对话接口（OpenAI 兼容）
curl -X POST http://127.0.0.1:51200/v1/chat/completions \
  -H "Authorization: Bearer <username>:<password>" \
  -d '{"model":"teambot","messages":[{"role":"user","content":"Hello"}],"stream":false}'
```

---

## 使用方式

除了 Web 可视化前端，**用户还可以直接与使用此 Skill 的 Agent 对话**，由 Agent 辅助和指导操作：
- 配置 LLM、创建用户、启动服务
- 创建和管理团队、编排工作流
- 发起讨论、执行任务

> 完整命令列表和细节请参考 [docs/cli.md](./docs/cli.md)

登录后下载/点赞/收藏 ❤ 7 | ★ 0

TeamClaw

目录结构

SKILL.md

举报内容

提示