API Docs MCP
@EliFuzz
MCP server for API documentation, supporting GraphQL, OpenAPI/Swagger, and gRPC from local files or remote URLs
Overview
What is API Docs MCP?
API Docs MCP is an MCP server that provides tools for interacting with API documentation. It supports GraphQL, OpenAPI/Swagger, and gRPC specifications, fetching schema definitions from local files or remote URLs, caching them, and exposing them through a set of tools.
How to use API Docs MCP?
Configure API sources via the API_SOURCES environment variable, then run the MCP server. The server automatically registers tools from a specified directory. Use the api_docs tool to list available API methods and api_search to retrieve detailed documentation for specific methods.
Key features of API Docs MCP
- Dynamic tool registration from a specified directory
- Provides
api_docsandapi_searchtools - Schema caching with automatic periodic refresh
- Supports GraphQL, OpenAPI, and gRPC schemas
- Loads schemas from local files or remote URLs
- Environment-based configuration via
API_SOURCES
Use cases of API Docs MCP
- Quickly explore GraphQL schema methods and their arguments without leaving the chat
- Look up OpenAPI endpoint details during API integration development
- Browse gRPC service definitions and RPC methods interactively
- Cache remote API documentation locally for offline or faster access
FAQ from API Docs MCP
How do I configure API sources?
Set the API_SOURCES environment variable to specify the schema files or URLs. The format supports local file paths and remote URLs for GraphQL, OpenAPI, and gRPC.
What schema formats are supported?
GraphQL schemas can be loaded from .graphql, .gql, or JSON introspection results. OpenAPI/Swagger supports YAML and JSON. gRPC supports .proto files or reflection from remote URLs.
How does caching work?
The server caches schema data after fetching to reduce redundant requests. It automatically refreshes the cache periodically to keep documentation up-to-date.
Does the server require any runtime dependencies?
The server runs as an MCP server and is configured entirely through the API_SOURCES environment variable. No additional authentication or transport setup is required.
