Multi-MCP
Multi-model AI orchestration MCP server for automated code review, chat, compare, and debate in Claude Code.
Add to your client
Copy the config for your MCP client and paste it into its config file.
git clone https://github.com/religa/multi_mcp.git && cd multi_mcp && make installPaste into ~/Library/Application Support/Claude/claude_desktop_config.json
{
"mcpServers": {
"multi-mcp": {
"command": "/path/to/multi_mcp/.venv/bin/python",
"args": [
"-m",
"multi_mcp.server"
]
}
}
}Step-by-step guides: Add to Claude Desktop · Add to Cursor · Add to Windsurf
Before you start
- Python 3.11+
- uv (for dependency install / make install)
- jq (for automatic Claude Code / OpenCode config via make install)
- API key for at least one provider (OpenAI, Anthropic, Google, or OpenRouter)
- Optional: respective CLI tools installed for CLI models (Gemini CLI, Codex CLI, Claude CLI)
About Multi-MCP
An MCP server that Claude Code or OpenCode connects to in order to orchestrate multiple AI models in parallel for code review, chat, comparison, and debate. It supports OpenAI GPT, Anthropic Claude, Google Gemini, and OpenRouter (plus optional Azure OpenAI and AWS Bedrock), mixes API and CLI-based models, and uses async execution so total latency equals the slowest model rather than the sum of all models.
Tools & capabilities (5)
chatInteractive development assistance with repository context awareness; ask a model a question directly.
codereviewSystematic code review workflow with OWASP Top 10 security checks and performance analysis, optionally targeting specific models.
compareParallel multi-model analysis that gathers multiple perspectives (e.g., for architectural decisions) using the configured default models.
debateMulti-agent consensus workflow where models give independent answers and then critique each other for deeper analysis.
modelsList all available models and their aliases.
When to use it
- Automated code review with OWASP Top 10 security and performance checks directly in Claude Code
- Getting multiple AI models' perspectives on an architectural or design decision in parallel
- Running multi-agent debates to reach consensus on a tricky technical question
- Mixing API and local CLI models (e.g., Ollama) in a single review or comparison workflow
- Experimental standalone CLI code review for CI/pipelines via JSON output
Security notes
Requires API keys for at least one provider (OpenAI, Anthropic, Google, or OpenRouter), stored in a local .env file or ~/.multi_mcp/.env. Your code and prompts are sent to the configured third-party AI model providers for analysis. CLI models run external CLIs as subprocesses with elevated permissions (auto-edit / full-auto / acceptEdits modes), so only enable CLI models you trust.
Multi-MCP FAQ
Do I need all three AI providers?
No. Just one API key (OpenAI, Anthropic, or Google) is enough to get started.
Does it truly run in parallel?
Yes. The codereview, compare, and debate tools execute all models concurrently with Python's asyncio.gather(), so you get results in the time of the slowest model, not the sum of all response times.
How many models can I run at the same time?
There is no hard limit; in practice 2-5 models work well for most use cases, and tools default to your configured default models (typically 2-3).
How do I add custom or local models?
Define them in ~/.multi_mcp/config.yaml. You can add API models via litellm_model or CLI models via provider: cli with cli_command/cli_args, and assign short aliases.
Alternatives to Multi-MCP
Compare all alternatives →
Microsoft's official browser-automation MCP using Playwright's accessibility tree (no vision model).
Up-to-date, version-specific library documentation injected into your coding agent.
Code intelligence engine that indexes repos into a persistent knowledge graph for AI coding agents.
Compare Multi-MCP with: