openapi -swagger

This MCP server exposes an intelligent interface for OpenAPI/Swagger specifications by converting a Swagger/OpenAPI document into a searchable, context-aware MCP server. It allows your AI agents and MCP-enabled tools to query API endpoints, schemas, and examples without loading the entire Swagger file into memory or your chat context. Once running, you can instruct your MCP-enabled clients to search for endpoints, retrieve schema details, and generate example requests or client code based on the API specification. The server leverages the OpenAPI document to provide accurate type information, parameter details, and response schemas in a concise, AI-friendly format.

Prerequisites:

• Python 3.11 or newer
• Git

Step-by-step:

1. Clone the repository: git clone https://github.com/salacoste/openapi-mcp-swagger.git cd openapi-mcp-swagger

2. Create and activate a Python virtual environment (optional but recommended): python -m venv venv source venv/bin/activate # Unix/macOS venv\Scripts\activate # Windows

3. Install dependencies (from repository): pip install -r requirements.txt

4. Run the MCP server (as an MCP server backend): python -m openapi_mcp_swagger

5. Provide your Swagger/OpenAPI JSON file as input to the server according to its documentation (the project typically accepts a swagger.json path or URL to load the OpenAPI spec).

6. Connect your MCP-compatible client (Cursor, Claude, VS Code, etc.) to the running server endpoint.

Tip: If the project uses a specific CLI flag or environment variable to pass the OpenAPI file location, consult the README or help output (python -m openapi_mcp_swagger --help) for exact usage.

Notes and tips:

• Ensure your OpenAPI document is valid JSON/YAML and reachable by the server at startup.
• If your API uses authentication schemes in OpenAPI, verify that the MCP client queries include appropriate auth headers as defined by the server's capabilities.
• Monitor for large OpenAPI specs (10MB+); the MCP approach helps AI agents avoid context window overflows by providing targeted queries.
• If you modify the OpenAPI file, restart the MCP server to reload the specification.
• Check for MCP compatibility updates or plugin integrations in the repository for enhanced AI tool support.