Pinecone Vector Database MCP Server

Knowledge & MemoryPython

Read and write to a Pinecone vector database for RAG applications

Available Tools

semantic-search

Search for records in the Pinecone index using semantic similarity

read-document

Read a document from the Pinecone index by its ID

list-documents

List all documents in the Pinecone index, optionally filtering by namespace or metadata

pinecone-stats

Get statistics about the Pinecone index, including the number of records, dimensions, and namespaces

process-document

Process a document into chunks, generate embeddings, and upsert them into the Pinecone index

Pinecone Vector Database MCP provides seamless integration between Claude and Pinecone's vector database service. This MCP enables retrieval-augmented generation (RAG) capabilities by allowing Claude to store, search, and retrieve documents using semantic similarity. With this integration, Claude can maintain a persistent knowledge base in Pinecone, enhancing its ability to work with domain-specific information and large document collections.

Overview

The Pinecone Vector Database MCP connects Claude to Pinecone's vector database service, enabling powerful retrieval-augmented generation (RAG) capabilities. This integration allows Claude to store documents, search for semantically similar content, and retrieve specific information from your Pinecone index.

Prerequisites

A Pinecone account with an API key
A Pinecone index created for your documents
Python 3.8+ with uv or pip installed

Installation

Using Smithery (Recommended)

The easiest way to install the Pinecone MCP is through Smithery:

npx -y @smithery/cli install mcp-pinecone --client claude

Manual Installation

Install the package using uv (recommended) or pip:

uvx install mcp-pinecone

or

uv pip install mcp-pinecone

Configure Claude Desktop to use the MCP server:

For MacOS: Edit ~/Library/Application\ Support/Claude/claude_desktop_config.json

For Windows: Edit %APPDATA%/Claude/claude_desktop_config.json

Add the following configuration:

"mcpServers": {
  "mcp-pinecone": {
    "command": "uvx",
    "args": [
      "--index-name",
      "your-pinecone-index-name",
      "--api-key",
      "your-pinecone-api-key",
      "mcp-pinecone"
    ]
  }
}

Configuration Options

The MCP supports the following configuration options:

--index-name: The name of your Pinecone index (required)
--api-key: Your Pinecone API key (required)
--environment: Pinecone environment (optional, defaults to "gcp-starter")
--host: Host address for the MCP server (optional, defaults to "127.0.0.1")
--port: Port for the MCP server (optional, defaults to 8000)

Usage

Once installed and configured, you can use the Pinecone MCP in Claude by:

Starting a new conversation in Claude Desktop
Enabling the Pinecone MCP from the context panel
Using the available tools to interact with your Pinecone index

Example Workflows

Storing Documents: Ask Claude to store a document in Pinecone by providing the content and optionally specifying metadata.

Searching for Information: Ask Claude to search for information related to a specific query, and it will retrieve semantically similar content from your Pinecone index.

Reading Documents: Request Claude to retrieve and read a specific document from your Pinecone index by its ID.

Listing Documents: Ask Claude to list all documents stored in your Pinecone index, optionally filtering by namespace or metadata.

Troubleshooting

If you encounter issues:

Verify your Pinecone API key and index name are correct
Check that your Pinecone index is properly configured and accessible
Ensure Claude Desktop is properly configured to use the MCP server
Check the logs for any error messages

For more detailed information, visit the GitHub repository.