Home › Development › Multi-MCP

Multi-MCP

Multi-model AI orchestration MCP server with code review, compare, and debate tools.

Python macOSWindowsLinux

★ 36 stars Last pushed 2026-09-05 License: MIT

View repository → Website →

Connect

Review any command before running it. Package names and URLs come from the server's own registry entry.

Package (pypi 0.1.1)

uvx multi-mcp

Or add to your MCP client config:

{
  "mcpServers": {
    "multi_mcp": {
      "command": "uvx",
      "args": [
        "multi-mcp"
      ],
      "env": {
        "OPENAI_API_KEY": "<YOUR_VALUE>",
        "ANTHROPIC_API_KEY": "<YOUR_VALUE>",
        "GEMINI_API_KEY": "<YOUR_VALUE>",
        "OPENROUTER_API_KEY": "<YOUR_VALUE>",
        "DEFAULT_MODEL": "<YOUR_VALUE>",
        "DEFAULT_MODEL_LIST": "<YOUR_VALUE>"
      }
    }
  }
}

OPENAI_API_KEY secret — OpenAI API key for GPT models (optional - at least one provider required)
ANTHROPIC_API_KEY secret — Anthropic API key for Claude models (optional - at least one provider required)
GEMINI_API_KEY secret — Google API key for Gemini models (optional - at least one provider required)
OPENROUTER_API_KEY secret — OpenRouter API key for additional models (optional)
DEFAULT_MODEL — Default model to use (defaults to gpt-5-mini)
DEFAULT_MODEL_LIST — Default models for multi-model comparison (comma-separated, defaults to gpt-5-mini,gemini-3-flash)