MCPilot

MCPilot is an MCP server and gateway built on FastAPI that federates multiple Model Context Protocol servers and converts existing REST APIs into MCP-compliant tools. Designed for AI engineers, developers, and platform teams building complex agent architectures, the project unifies distributed AI toolchains behind a single control plane. MCPilot connects to local and remote services via HTTP, JSON-RPC, WebSocket, SSE, and stdio transports, centralizing tools, prompts, and resources with full JSON-Schema validation. It provides an administrative web interface, automated OpenAPI/Swagger documentation, and configuration controls for managing server registrations and wrapper behaviors. By acting as a gateway, it enables AI clients like Claude Desktop to discover and execute operations across several independent systems through a single unified endpoint, while allowing non-MCP REST endpoints to be queried as standard model context tools without custom agent code.

Category: AI & LLM Tooling

Tags: fastapi, gateway, infrastructure, orchestration

Visit MCPilot

How to install and configure MCPilot

To set up and connect MCPilot, follow these steps: 1. Ensure Python 3.10 or higher is installed, along with uv or pip. 2. Install the package dependencies: bash uv sync Or install with pip: bash pip install -e . 3. Start the gateway server: bash uv run python -m mcpilot.main 4. For client configuration in Claude Desktop, edit claude_desktop_config.json (located at ~/Library/Application Support/Claude/claude_desktop_config.json on macOS or %APPDATA%/Claude/claude_desktop_config.json on Windows): json { "mcpServers": { "MCPilot": { "command": "uvx", "args": ["MCPilot"] } } }

What you can do with MCPilot

• Expose legacy REST API endpoints to AI assistants as native MCP tools using configurable path and parameter wrappers. * Unify multiple internal MCP servers into a single centralized gateway running behind an admin dashboard. * Provide real-time interactive JSON-RPC communication for AI agents over WebSocket and SSE transports. * Standardize schema validation and payload handling across disparate toolchains without modifying upstream backend code. * Store, manage, and summarize notes and contextual prompts using the gateway's built-in reference resources.

Key facts

• https://github.com/ferrary7/MCPilot
• AI & LLM Tooling
• fastapi, gateway, infrastructure, orchestration

Part of MCP Servers

How do I install MCPilot?

MCPilot requires Python 3.10 or higher. You can install its dependencies using uv by running uv sync, or with pip via pip install -e . Once installed, launch the gateway locally by executing uv run python -m mcpilot.main.

What can MCPilot do?

MCPilot serves as a central gateway for the Model Context Protocol. It wraps REST APIs into MCP-compliant tools, federates multiple MCP servers into one endpoint, provides an admin UI, and supports transports like HTTP, WebSocket, SSE, and stdio.

Which MCP clients work with MCPilot?

MCPilot works with any client supporting the Model Context Protocol, including Claude Desktop, Cursor, or custom clients connecting via HTTP REST endpoints, WebSocket connections, or standard I/O commands.

Is MCPilot open source?

Yes, MCPilot is an open-source project distributed under the MIT License on GitHub.