MCP Meilisearch API Server

A comprehensive Model Context Protocol (MCP) server implementation that provides a bridge between AI models and the Meilisearch search engine using the StreamableHTTP transport. This project enables seamless integration of Meilisearch's powerful search capabilities within AI workflows.

Project Overview

This project implements an MCP (Model Context Protocol) server that provides AI models with direct access to Meilisearch's functionalities. The implementation follows a client-server architecture with these key components:

• MCP Server: Implements the Model Context Protocol to expose Meilisearch APIs as tools
• Web Client: Simple demo interface for testing the search functionality
• Command Line Client: Utility client for testing and development

Architecture

┌──────────────┐      ┌──────────────┐     ┌───────────────┐
│  Web Client  │      │  MCP Server  │     │  Meilisearch  │
│  (Browser)   │ <--> │   (Node.js)  │ <-> │   Instance    │
└──────────────┘      └──────────────┘     └───────────────┘
       ^                                            ^
       │                                            │
┌──────────────┐                          ┌───────────────┐
│ Command Line │                          │ Document Data │
│    Client    │                          │   Sources     │
└──────────────┘                          └───────────────┘

Key Features

• StreamableHTTP Transport: Implements the StreamableHTTP transport for MCP, enabling real-time communication between clients and server
• Full Meilisearch API Support: Exposes all Meilisearch functionalities as MCP tools
• Category-based Organization: Tools are organized by functional categories
• Error Handling: Comprehensive error handling for API requests
• Web Client Demo: Simple web interface to demonstrate search capabilities
• Command Line Client: For testing and development

Available Tool Categories

The MCP server exposes Meilisearch APIs organized into these functional categories:

1. System Tools: Health checks, version information, server stats
2. Index Tools: Managing indexes (create, update, delete, list)
3. Document Tools: Document operations (add, update, delete, retrieve)
4. Search Tools: Advanced search capabilities including vector search
5. Settings Tools: Configuration management for indexes
6. Task Tools: Asynchronous task management
7. Vector Tools: Vector search capabilities (experimental feature)

Getting Started

Prerequisites

• Node.js v20 or higher
• Meilisearch instance running locally or remotely
• API key for Meilisearch (if required by your Meilisearch configuration)

Setup

1. Clone the repository
2. Install dependencies:

npm install

3. Create a .env file in the server directory with your Meilisearch configuration:

MEILISEARCH_HOST=http://localhost:7700
MEILISEARCH_API_KEY=your_master_key_here
MEILISEARCH_TIMEOUT=5000

Running the Server

Build and start the server:

npm run dev:cmd  # For command line testing
# OR
npm run dev:web  # For web interface testing

Accessing the Web Interface

Once running, the web demo is available at:

http://localhost:8000

Development

This project uses:

• TypeScript for type safety
• Lerna for workspace management
• Express for the web server
• Model Context Protocol SDK for AI integration

Project Structure

• server/: MCP server implementation
  • src/tools/: Implementation of Meilisearch API tools
  • src/utils/: Utility functions for API communication and error handling
  • src/server.ts: StreamableHTTP MCP server implementation
• client_web/: Web demo client
• client_cmd/: Command line client for testing