c2-contact/c2-mcp

c2-mcp

A Model Context Protocol (MCP) server for contact management, powered by pglite (embedded Postgres/SQLite hybrid) and Ollama for AI-powered embeddings and semantic search.

Features

• MCP server: Implements the Model Context Protocol for tool-based automation and agent integration.
• Contact management: Create, read, update, delete, list, and search contacts.
• Bulk operations: Bulk create, update, and delete contacts.
• Semantic search: Find contacts using vector embeddings and similarity search (via Ollama).
• Configurable: Database location, AI base URL, and embeddings model are all configurable.

Requirements

• Bun (v1.2.18 or later recommended)
• Ollama running locally or accessible via HTTP (default: http://localhost:11434)

Installation

bun install

Running the Server

bun run index.ts

Configuration

You can configure the server using environment variables or command-line arguments:

• DB_PATH or --db-path=... — Path to the database directory (default: ./db in storage dir)
• AI_BASE_URL or --ai-base-url=... — Base URL for the Ollama API (default: http://localhost:11434/v1)
• EMBEDDINGS_MODEL or --embeddings-model=... — Embeddings model to use (default: mxbai-embed-large)

Example:

DB_PATH=./mydb AI_BASE_URL=http://localhost:11434/v1 EMBEDDINGS_MODEL=mxbai-embed-large bun run index.ts

Main Tools/Endpoints

The MCP server exposes the following tools:

• create-contact — Create a new contact
• get-contact — Get a contact by ID
• list-contacts — List all contacts (with pagination)
• search-contacts — Search contacts by name, email, or phone
• update-contact — Update an existing contact
• delete-contact — Delete a contact by ID
• semantic-search-contacts — Semantic search using embeddings
• bulk-create-contacts — Bulk create contacts
• bulk-update-contacts — Bulk update contacts
• bulk-delete-contacts — Bulk delete contacts

Development

• The project uses drizzle-orm for database access and migrations.
• Database migrations are stored in the drizzle/ directory and run automatically on startup.
• Embeddings are generated via the Ollama API and stored in the database for semantic search.

Project Structure

• index.ts — Main entrypoint, server setup
• src/contact-service.ts — Contact management logic
• src/database.ts — Database setup (pglite)
• src/embeddings.ts — Embedding generation (Ollama)
• src/context.ts — Context and configuration helpers
• drizzle/ — Database migrations
• integration/, e2e/, src/*.test.ts — Tests

License

MIT

This project was created using bun init in bun v1.2.18. Bun is a fast all-in-one JavaScript runtime.