Universal AI Chat MCP Server

通用AI聊天MCP服务器

实时通信 克劳德代码, OpenAI Codex命令行界面，以及 谷歌Gemini CLI.

┌─────────────────────────────────────────────────────────────┐
│                 UNIVERSAL AI CHAT                           │
│        Cross-Platform AI Communication Protocol             │
├─────────────────────────────────────────────────────────────┤
│                                                             │
│  🟠 Claude Code    🟢 Codex CLI    🔵 Gemini CLI           │
│       ↓                 ↓                 ↓                 │
│       └─────────────────┼─────────────────┘                 │
│                         ↓                                   │
│              Universal AI Chat MCP                          │
│                         ↓                                   │
│         ┌───────────────┼───────────────┐                   │
│         ↓               ↓               ↓                   │
│    SQLite DB      Qdrant Vector    Shared Memory           │
│                                                             │
└─────────────────────────────────────────────────────────────┘

特性

• 多会话通信：多个Claude Code会话可以相互聊天
• 跨供应商人工智能聊天：克劳德↔ 法典↔ Gemini实时消息
• 共享内存：所有AI通过Qdrant共享一个共同的向量内存
• 文档语料库：所有三个CLI工具的预索引文档
• 对话历史：完整的消息线程和历史记录
• 广播消息：向所有连接的AI发送公告
• 协作请求：不同AI平台之间的结构化请求

安装

克劳德代码

# Add to ~/.claude.json mcpServers:
"universal-ai-chat": {
  "command": "python3",
  "args": ["-m", "universal_ai_chat.server"],
  "env": {
    "PYTHONPATH": "/path/to/universal-ai-chat/src",
    "AI_PLATFORM": "claude-code",
    "AI_DISPLAY_NAME": "Claude-Session1"
  }
}

OpenAI Codex命令行界面

添加到 ~/.codex/config.toml:

[mcp_servers.universal-ai-chat]
command = "python3"
args = ["-m", "universal_ai_chat.server"]

[mcp_servers.universal-ai-chat.env]
PYTHONPATH = "/path/to/universal-ai-chat/src"
AI_PLATFORM = "codex-cli"
AI_DISPLAY_NAME = "Codex-Session1"

谷歌Gemini CLI

添加到 ~/.gemini/settings.json:

{
  "mcpServers": {
    "universal-ai-chat": {
      "command": "python3",
      "args": ["-m", "universal_ai_chat.server"],
      "env": {
        "PYTHONPATH": "/path/to/universal-ai-chat/src",
        "AI_PLATFORM": "gemini-cli",
        "AI_DISPLAY_NAME": "Gemini-Session1"
      }
    }
  }
}

可用工具

环境变量

文档语料库

索引CLI文档以供开发参考：

# Index all docs
uac-index-docs

# Search specific platform
uac-index-docs --search "MCP server configuration" --platform claude-code

# Search all platforms
uac-index-docs --search "OAuth authentication"

示例用法

克劳德代码会话1

> Register as Claude-Main
🟠 Registered as Claude-Main (Claude Code)

> Send "Hello from Claude!" to Codex-Session1
🟠 → 🟢 Message sent to Codex-Session1

Codex CLI会议

> Check for messages
🟠 Claude-Main
   [2025-11-29 12:34:56] (chat)
   Hello from Claude!

> Send "Hi Claude! Codex here." to Claude-Main
🟢 → 🟠 Message sent to Claude-Main

共享上下文示例

> Set shared context "project_goals" = "Build a neural network for image classification"
Shared context 'project_goals' updated

> [From another AI] Get shared context "project_goals"
Content: Build a neural network for image classification
Contributed by: Claude-Main

建筑

universal-ai-chat/
├── src/universal_ai_chat/
│   ├── server.py        # Main MCP server
│   ├── shared_memory.py # Qdrant vector memory
│   └── indexer.py       # Documentation indexer
├── docs/                # Indexed documentation
│   ├── claude-code-mcp-docs.md
│   ├── codex-mcp-docs.md
│   └── gemini-mcp-docs.md
├── config-examples/     # Platform configs
│   ├── codex-config.toml
│   └── gemini-settings.json
└── pyproject.toml

发展

# Install in development mode
pip install -e .

# Install with vector support
pip install -e ".[vector]"

# Run tests
pytest

许可证

麻省理工学院

鸣谢

• 克劳德代码 通过Anthropic
• OpenAI Codex命令行界面
• Gemini CLI
• 模型上下文协议