Tools Reference

The Jeremy MCP server exposes two tools to AI assistants. Use resolve-library-id first to find the correct library, then query-docs to search its documentation.

resolve-library-id

Searches for libraries matching a given name. Use this to find the correct library ID before querying documentation.

Input

{
  "libraryName": "react"
}

Output

Returns an array of matching libraries:

[
  {
    "id": "react",
    "name": "react",
    "version": "19.0.0",
    "description": "A JavaScript library for building user interfaces"
  }
]

Each result includes:

• id -- the library ID to use with query-docs.
• name -- the library's display name.
• version -- the library version, if available.
• description -- a short description of the library.

query-docs

Queries a library's documentation using semantic search. Returns relevant chunks ranked by relevance.

Input

{
  "libraryId": "react",
  "query": "how to use useEffect cleanup",
  "topic": "hooks"
}

Output

Returns an array of documentation chunks ranked by relevance:

[
  {
    "content": "## Cleaning up an Effect\n\nSome Effects need to specify how to stop, undo, or clean up whatever they were doing...",
    "score": 0.92,
    "title": "useEffect",
    "url": "https://react.dev/reference/react/useEffect"
  }
]

Each result includes:

• content -- the documentation text for this chunk.
• score -- relevance score between 0 and 1.
• title -- the title or heading for this chunk.
• url -- the source URL of the documentation.

How It Works

The query-docs tool generates a vector embedding of your query using Workers AI and performs a cosine similarity search against the library's stored embeddings in Vectorize. The top results are enriched with full content from D1 and returned ranked by relevance score.

The optional topic parameter further filters results by matching against chunk titles and content, narrowing the search to a specific area of the documentation.