baryhuang_mcp-server-any-openapi
1
This project is a scalable OpenAPI endpoint discovery and API request tool intended to support large OpenAPI documents through in-memory semantic search and endpoint-based chunking. It provides a fast and flexible server solution that integrates with Claude Desktop to process API requests efficiently.
MCP Server: Scalable OpenAPI Endpoint Discovery and API Request Tool
Features
- Use remote OpenAPI JSON file as source, no need for local file system access.
- Implements semantic search using optimized MiniLM-L3 model.
- FastAPI-based server with async support.
- Endpoint based chunking of OpenAPI specs.
- In-memory FAISS vector search for instant endpoint discovery.
Limitations
- Not supporting linux/arm/v7.
- Cold start penalty if not using docker image.
- Docker image issues involving model downloading and size.
Installation
Via Smithery
To install use: npx -y @smithery/cli install @baryhuang/mcp-server-any-openapi --client claude
Using pip
Install with: pip install mcp-server-any-openapi
Docker Support
- Supports multi-platform builds for linux/amd64 and linux/arm64.
Configuration
Customize through environment variables like OPENAPI_JSON_DOCS_URL
and MCP_API_PREFIX
.
Available Tools
- {prefix}_api_request_schema: Get API endpoint schemas based on query.
- {prefix}_make_request: Essential for executing complex API requests reliably.