Pinecone Agentic Search — MCP Server

A custom MCP (Model Context Protocol) server that exposes Pinecone vector search as a standardized tool for AI agents. Built to replace the n8n MCP dependency in the GenAI Concepts chat app.

What It Does

Exposes a single tool called agentic-search that:

• Accepts a natural language query
• Embeds it using OpenAI text-embedding-3-small via OpenRouter
• Searches the Pinecone knowledge base (4,128 ArXiv research papers)
• Returns the most relevant excerpts with relevance scores

Architecture

AI Agent → MCP Client → This Server → Pinecone (mcp-server-v1) → results

Tools Exposed

Tool	Description
agentic-search	Search the GenAI knowledge base covering AI Agents, RAG, MCP, and Prompt Engineering

Prerequisites

• Node.js 18+
• OpenRouter API key
• Pinecone API key with index mcp-server-v1 (namespace: arxiv-papers)

Setup

# 1. Install dependencies
npm install

# 2. Create .env file
cp .env.example .env

# 3. Add your keys
PINECONE_API_KEY=your_key
OPENROUTER_API_KEY=your_key
PINECONE_INDEX=mcp-server-v1
PINECONE_NAMESPACE=arxiv-papers
PORT=3001

Run Locally

npm run dev

Endpoints

• GET /health — health check
• POST /mcp — MCP endpoint (Streamable HTTP transport)
• GET /sse — SSE transport for legacy clients

Deploy to Railway

1. Push to GitHub
2. New Project → Deploy from GitHub
3. Add environment variables
4. Railway auto-deploys

Data Source

ArXiv research papers covering AI Agents, RAG, MCP, and Prompt Engineering. Used for non-commercial demonstration purposes only. Papers are subject to their respective authors' licenses (CC BY 4.0).

Evolution

This server replaces the n8n MCP server used in the original architecture, giving full ownership of the MCP layer with no subscription dependencies.