ControlAPI-MCP

MCP server that exposes any OpenAPI/REST API as MCP tools with dynamic server switching and variable substitution.

🚀 One-Click Install

Click to install directly in your editor - no configuration needed!

Linux / macOS:

💡 After Installation: The AI assistant will guide you to connect to an API server. Simply provide the OpenAPI URL when asked, or use the set_server_config tool to connect to your API.

Quick Start (Auto-Download)

Zero installation - automatically downloads and runs the latest release:

Download the auto-run script:

curl -O https://raw.githubusercontent.com/fellowabhi/ControlAPI-openapi-to-mcp/main/auto-run.sh
chmod +x auto-run.sh

Use it in your MCP config:

{
  "servers": {
    "controlapi-mcp": {
      "type": "stdio",
      "command": "/path/to/auto-run.sh"
    }
  }
}

Note: You can optionally set OPENAPI_URL, BASE_URL, and SERVER_NICKNAME in env vars, or configure dynamically using the set_server_config tool.

Download from releases or build:

export OPENAPI_URL='http://your-api.com/openapi.json'
export BASE_URL='http://your-api.com'  # optional
./dist/controlapi-mcp

Setup (Development)

pip install -e .

Building Binary

./build.sh

Creates a standalone executable at dist/controlapi-mcp (16MB)

MCP Configuration

Using Binary

{
  "servers": {
    "controlapi-mcp": {
      "type": "stdio",
      "command": "/path/to/openapi-mcp-adapter/dist/controlapi-mcp",
      "env": {
        "OPENAPI_URL": "http://localhost:8000/openapi.json",
        "BASE_URL": "http://localhost:8000",
        "DEBUG_PORT": "45133"
      }
    }
  }
}

Using Python (Development)

{
  "servers": {
    "controlapi-mcp": {
      "type": "stdio",
      "command": "/path/to/project/.venv/bin/python",
      "args": ["-m", "src.main"],
      "cwd": "/path/to/project",
      "env": {
        "OPENAPI_URL": "http://localhost:8000/openapi.json",
        "REFRESH_INTERVAL": "300",
        "PYTHONPATH": "/path/to/project",
        "BASE_URL": "http://localhost:8000",
        "DEBUG_PORT": "45133"
      }
    }
  }
}

Optional: OPENAPI_URL, BASE_URL, SERVER_NICKNAME, DEBUG_PORT (default: 45133)

💡 No Configuration Needed: You can start with no environment variables and configure the server dynamically using the set_server_config tool. The AI assistant will guide you through the setup.

🔍 Debug UI: Access browser-based request/response monitor at http://localhost:45133 (or custom DEBUG_PORT). URL available in get_server_info response.

Features

Cross-Platform Support

Linux (x86_64) - Full support
macOS (Intel & Apple Silicon) - Full support
Windows - Use Python or WSL

Dynamic Server Switching

Connect to any OpenAPI server at runtime
Switch between multiple APIs (dev, staging, production)
Server context tracking with history
Automatic schema reloading

Available Tools

set_server_config - Connect to an OpenAPI server (use this first if not configured)
get_server_info - Check current server and connection status
get_server_history - View recent server switches
health_check - Test server connectivity
list_endpoints - List all API endpoints
search_schema - Search endpoints by keyword
execute_request - Make HTTP requests with variable substitution
set_variable - Store variable (e.g., auth token)
get_variables - View all stored variables
reload_schema - Reload current server's schema

Variable Substitution

Use {{variable_name}} in headers, body, or path:

{
  "headers": {
    "Authorization": "{{token}}"
  }
}

Example Workflow

First time: set_server_config with openapi_url
execute_request to /auth/login → get token
set_variable key="token" value="Bearer xyz..."
execute_request with Authorization: {{token}}
Switch servers: set_server_config to test on different environment