home / mcp / pinecone developer mcp server

Pinecone Developer MCP Server

Connect your Pinecone projects to Cursor, Claude, and other AI assistants

Installation
Add the following to your MCP client configuration file.

Configuration

View docs
{
  "mcpServers": {
    "pinecone-io-pinecone-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "@pinecone-database/mcp"
      ],
      "env": {
        "PINECONE_API_KEY": "YOUR_API_KEY"
      }
    }
  }
}

The Pinecone Developer MCP Server lets your coding assistants interact with Pinecone projects by connecting to Pinecone through the Model Context Protocol (MCP). It enables you to search Pinecone documentation, help configure indexes, generate code based on your index setup and Pinecone examples, and upsert or query data within your indexes from your development environment.

How to use

You configure an MCP client to talk to the Pinecone MCP Server, then your AI tools can search Pinecone docs, list and describe your indexes, create new ones using embedded models, and operate on your data. Use this to accelerate development, testing, and integration work within your Pinecone projects.

How to install

Prerequisites you need before starting: Node.js v18 or later and a shell where you can run commands. Ensure that node and npx are available in your PATH.

{
  "mcpServers": {
    "pinecone": {
      "command": "npx",
      "args": [
        "-y", "@pinecone-database/mcp"
      ],
      "env": {
        "PINECONE_API_KEY": "<your pinecone api key>"
      }
    }
  }
}

Additional setup steps

To use the MCP server with Cursor, Claude desktop, or Gemini CLI, add the same MCP configuration to the respective client. The common configuration uses the MCP server executable via npx and injects your Pinecone API key through an environment variable.

Set the API key in your environment as shown in the configuration snippet and restart the client to enable MCP-based features.

Security and credentials

Keep your Pinecone API key secure. Do not expose it in logs or shared configurations. Use environment variables to keep secrets out of your code. If you revoke or rotate keys, update the MCP configuration accordingly.

Troubleshooting

If the MCP server does not appear in your AI tool, verify Node.js v18+ is installed and that npx is available in your PATH. Check that the MCP configuration file is in the correct location and uses valid JSON syntax, then restart your AI tool.

If you encounter authentication issues, confirm the Pinecone API key matches the one in the Pinecone console and ensure PINECONE_API_KEY is set in the MCP configuration. Look for extra spaces or quotes around the key.

For tool-related problems, remember the MCP server focuses on indexes with integrated inference. If you run into issues with non-integrated indexes, create or switch to an index that supports embedding. Review logs in your client’s MCP settings to diagnose errors.

Connection issues on restricted networks may require allowing traffic to api.pinecone.io. If needed, run the MCP server manually to view detailed errors using: PINECONE_API_KEY=<your-key> npx @pinecone-database/mcp

Notes

This server focuses on developer workflows for Pinecone, including documentation search, index configuration assistance, and data operations to test and verify behavior within your dev environment.

Available tools

search-docs

Search the official Pinecone documentation to retrieve information and examples relevant to your questions.

list-indexes

List all Pinecone indexes in your project along with their configurations.

describe-index

Describe the configuration of a specific Pinecone index.

describe-index-stats

Provide statistics about the data in an index, including record counts and namespaces.

create-index-for-model

Create a new index that uses an integrated inference model to embed text as vectors.

upsert-records

Insert or update records in an index using the integrated inference model.

search-records

Search for records in an index by text queries, with embedding-based relevance and optional metadata filters.

cascading-search

Search across multiple indexes with deduplication and reranking of results.

rerank-documents

Rerank a collection of records or documents using a specialized reranking model.