TL;DR: Built an MCP server that lets Claude Desktop analyze OpenAPI specs through natural language queries. Ask "What authentication methods does this API use?" instead of manually parsing JSON schemas.
What is it?
OpenAPI Analyzer MCP bridges the gap between human questions and API documentation.
Instead of:
- Manually reading through 50+ page OpenAPI specs
- Writing custom scripts to parse API schemas
- Switching between multiple tools to understand APIs
You can now ask Claude directly:
- "Which endpoints require authentication?"
- "Find all pagination patterns across these APIs"
- "What's inconsistent between v1 and v2 specs?"
Key Features
Smart Discovery System
- Point it at an API registry (apis.json format) and it loads dozens of specs
automatically
- Fallback chain: API registry → individual URLs → local files
Natural Language Analysis
- Query endpoints semantically, not just by exact matches
- Find patterns across multiple API specifications
- Detect inconsistencies and suggest improvements
Built for Scale
- Handles YAML, JSON, and YML OpenAPI formats
- Memory-efficient parsing with swagger-parser
Why I Built This
Working with microservices architectures, I was constantly jumping between API docs trying to understand:
- Which services handle user authentication
- How different teams implement pagination
- What's the schema for shared data models
Configure in Claude Desktop, point it at any OpenAPI spec (or API registry - see the example config), and start asking questions about your APIs in plain English.
I'd love feedback from the API tooling community. What other analysis patterns would be useful to add next?
GitHub: https://github.com/sureshkumars/openapi-analyzer-mcp