Summary
A large specification stops being something you paste into context.
The catalog carries the metadata, so an assistant searches for the operation it needs and loads that one alone — by operationId, or by the path and method read off a log line. When the files on disk change, refreshing the catalog is the entire update procedure.
What it is
A catalog over the OpenAPI specifications in a directory: metadata about every spec, its operations and its schemas, with search and targeted loads on top.
What you get
- A catalog holding metadata for every specification in the directory, its operations and its schemas
- Operations searched across all specifications at once, or narrowed to one by specId
- Schemas searched the same way, across specifications or within one
- A single operation loaded by operationId, or by path and method when that is what you have in front of you
- A schema loaded by name
- The catalog refreshed after the specs on disk change
Requirements
A directory of OpenAPI specification files, and npx on your PATH. No account and no key.
Setup effort
One command — npx -y @reapi/mcp-openapi@latest --dir ./specs