Package Exports

hybrid-search-mcp
hybrid-search-mcp/build/index.js

This package does not declare an exports field, so the exports above have been automatically detected and optimized by JSPM instead. If any package subpath is missing, it is recommended to post an issue to the original package (hybrid-search-mcp) to support the "exports" field. If that is not possible, create a JSPM override to customize the exports field for this package.

Readme

Hybrid Search MCP Server (`hybrid-search-mcp`)

License: MIT

MCP server providing hybrid search capabilities using Meilisearch for full-text search and Qdrant for semantic vector search, powered by Cohere embeddings.

Prerequisites

Node.js (Check package.json for version requirements)
Access to a running Qdrant instance
Access to a running Meilisearch instance
A Cohere API key

Configuration

This server requires the following environment variables to be set. You can create a .env file in the root directory to manage these:

COHERE_API_KEY="your_cohere_api_key"
QDRANT_ENDPOINT="your_qdrant_url" # e.g., http://localhost:6333
QDRANT_API_KEY="your_qdrant_api_key" # Optional, if Qdrant requires auth
MEILISEARCH_ENDPOINT="your_meilisearch_url" # e.g., http://localhost:7700
MEILISEARCH_API_KEY="your_meilisearch_api_key" # Optional, if Meilisearch requires auth

# Optional:
# COLLECTION_NAME="your_custom_collection_or_index_name" # Defaults to 'my_documents'

Installation

npm install

Building

npm run build

This compiles the TypeScript source code to JavaScript in the build directory.

Running the Server

Production

npm start

This runs the compiled JavaScript code from the build directory.

Development

npm run dev

This starts the TypeScript compiler in watch mode (tsc -w) and runs the server using node --watch, automatically restarting on changes.

Usage

Once running, the server communicates using the Model Context Protocol (MCP) over standard input/output (stdio). You can interact with it using an MCP client.

The server exposes the following tools:

store_document: Stores text in both Meilisearch and Qdrant.
full_text_search: Performs a keyword search using Meilisearch.
similarity_search: Performs a vector similarity search using Qdrant and Cohere embeddings.