MCP Server

Context42 implements the Model Context Protocol (MCP) to provide AI assistants with semantic search over your documentation.

Starting the server

c42 serve

The server communicates via stdio, which is the standard transport for MCP servers used with Claude Desktop.

Available tools

The MCP server exposes a search tool that AI assistants can call:

Search indexed documentation by semantic similarity.

Parameters:

  • query (string, required) — The search query

Returns: A list of relevant text chunks, ranked by a combination of semantic similarity and source priority.

How it works

  1. The AI assistant sends a search query via MCP
  2. Context42 converts the query into a vector embedding (locally)
  3. The vector index returns the most similar chunks
  4. Results are re-ranked using source priorities
  5. The top chunks are returned to the AI assistant

Privacy

All processing happens locally:

  • Embeddings are computed on your machine
  • The vector index is stored locally
  • No data is sent to external services
  • The MCP server only communicates with the local AI client via stdio
Configuration