laurus-mcp 0.3.0

MCP (Model Context Protocol) server for the Laurus search engine
docs.rs failed to build laurus-mcp-0.3.0
Please check the build logs for more information.
See Builds for ideas on how to fix a failed build, or Metadata for how to configure docs.rs builds.
If you believe this is docs.rs' fault, open an issue.

laurus-mcp

Crates.io License: MIT

Model Context Protocol (MCP) server for the Laurus search engine. Enables AI assistants such as Claude to index documents and perform searches through the standard MCP stdio transport.

Features

  • MCP stdio transport -- Runs as a subprocess; communicates with the AI client via stdin/stdout
  • gRPC client -- Proxies all tool calls to a running laurus-server instance
  • All search modes -- Lexical (BM25), vector (HNSW/Flat/IVF), and hybrid search
  • Dynamic connection -- Connect to any laurus-server endpoint via the connect tool
  • Document lifecycle -- Put, add, get, delete, and search documents through MCP tools

Architecture

AI Client (Claude, etc.)
  └─ stdio (JSON-RPC) ─→ laurus-mcp
                            └─ gRPC ─→ laurus-server
                                         └─ Index on Disk

Quick Start

# Start laurus-server
laurus serve --port 50051

# Configure Claude Code
claude mcp add laurus -- laurus mcp --endpoint http://localhost:50051

MCP Tools

Tool Description
connect Connect to a laurus-server gRPC endpoint
create_index Create a new index with a schema
get_stats Get index statistics (document count, vector fields)
get_schema Get the current index schema
add_field Dynamically add a field to the index
delete_field Remove a field from the schema
put_document Put (upsert) a document
add_document Add a document as a new chunk (append)
get_documents Retrieve all documents by ID
delete_documents Delete all documents by ID
commit Commit pending changes to disk
search Search documents using the Laurus query DSL

Documentation

License

This project is licensed under the MIT License - see the LICENSE file for details.