Vendor Search MCP Server (Node.js TypeScript)

A Model Context Protocol (MCP) server for vendor search functionality, converted from Python to Node.js TypeScript. This server provides advanced vendor search capabilities with AI-powered features including synonym expansion, relevance ranking, and document parsing.

Features

• Vendor Search: Advanced search with synonym expansion and relevance-based ranking
• Contact Information: Retrieve vendor contact details including address, email, and phone
• Document Parsing: Parse documents using LlamaParse integration
• Google Search: Perform web searches for additional information
• Database Integration: MongoDB and Typesense support
• AI-Powered: Cohere reranking, OpenAI, and Perplexity integrations

Tools Available

Core Search Tools

• find_relevant_vendors: Search for vendors by service, location, and/or name
• get_vendor_contact_details: Get contact information for specific vendors
• typesense_query: Direct Typesense collection queries
• get_table_schema: Retrieve Typesense schema information

Additional Tools

• google_search: Perform Google searches
• parse_document_link: Parse documents using LlamaParse

Installation

1. Install dependencies:

   npm install

2. Set up environment variables: Create a .env file with the following variables:

   # Typesense Configuration
   TYPESENSE_HOST=localhost
   TYPESENSE_PORT=8108
   TYPESENSE_PROTOCOL=http
   TYPESENSE_API_KEY=your_typesense_api_key_here
   
   # MongoDB Configuration
   MONGODB_URI=mongodb://localhost:27017
   MONGODB_DB_NAME=vendor_search
   
   # Cohere Configuration
   COHERE_API_KEY=your_cohere_api_key_here
   
   # LlamaParse Configuration
   LLAMA_API_KEY=your_llama_api_key_here
   VENDOR_MODEL=your_vendor_model_here
   
   # OpenAI Configuration
   OPENAI_API_KEY=your_openai_api_key_here
   
   # Perplexity Configuration
   PERPLEXITY_API_KEY=your_perplexity_api_key_here

Usage

Development Mode

npm run dev

Production Mode

npm start

Using npx

npx vendor-search-mcp-server [options]

Command Line Options

• --typesense-host: Typesense server host (default: localhost)
• --typesense-port: Typesense server port (default: 8108)
• --typesense-protocol: Typesense protocol (http/https)
• --mongodb-uri: MongoDB connection URI
• --mongodb-db-name: MongoDB database name
• --debug: Enable debug output
• --non-interactive, -n: Run in non-interactive mode
• --help, -h: Show help message

As an MCP Server

Configure your MCP client to use this server:

{
  "mcpServers": {
    "vendor-search": {
      "command": "npx",
      "args": ["vendor-search-mcp-server"]
    }
  }
}

Database Setup

MongoDB Collections

• typesense_schema: Schema information for Typesense collections
• vendor_services_synonyms_updated: Service synonyms for enhanced search
• users: User information (for resources)

Typesense Collections

• vendor4: Main vendor data collection with fields:
  • vendorName: Vendor name
  • servicesOffered: Services provided
  • locationRegion: Vendor location
  • address: Physical address
  • contactEmail: Email contact
  • contactNumber: Phone contact
  • embedding: Vector embeddings for search
  • gpt_chunked_services: AI-processed service descriptions

AI Features

Synonym Expansion

The server uses MongoDB-stored synonym mappings to expand search queries, improving match accuracy for vendor services.

Relevance Ranking

Cohere's reranking API is used to improve search result relevance based on semantic similarity.

Document Processing

LlamaParse integration allows parsing of various document formats for vendor information extraction.

Project Structure

src/
├── config/
│   └── constants.ts          # Configuration management
├── database/
│   └── clients.ts            # Database client wrappers
├── server/
│   ├── index.ts             # Main server implementation
│   ├── resources.ts         # MCP resources handler
│   └── prompts.ts           # MCP prompts handler
├── tools/
│   ├── schemas.ts           # Tool schema definitions
│   └── handlers.ts          # Tool implementation handlers
├── utils/
│   └── logger.ts            # Logging utilities
└── index.ts                 # Entry point

Error Handling

The server includes comprehensive error handling with:

• Graceful degradation when services are unavailable
• Detailed logging for debugging
• Client-friendly error messages
• Fallback mechanisms for AI services

Dependencies

Core Dependencies

• @modelcontextprotocol/sdk: MCP SDK for server implementation
• mongodb: MongoDB client
• typesense: Typesense search client
• cohere-ai: Cohere API client for reranking
• openai: OpenAI API client
• winston: Logging framework

Development Dependencies

• typescript: TypeScript compiler
• tsx: TypeScript execution environment
• @types/node: Node.js type definitions

Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Submit a pull request

License

MIT License - see LICENSE file for details.

Support

For issues and questions:

1. Check the debug logs (enable with --debug flag)
2. Verify all required environment variables are set
3. Ensure database connections are working
4. Check API key validity for external services