MCP Directory

Pinecone Developer MCP Server

Official

Official Pinecone server: manage indexes, upsert/search records, rerank, and search Pinecone docs.

Verified
stdio (local)
API key
TypeScript

Add to your client

Copy the config for your MCP client and paste it into its config file.

Install / run
npx -y @pinecone-database/mcp

Paste into ~/Library/Application Support/Claude/claude_desktop_config.json

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

Step-by-step guides: Add to Claude Desktop · Add to Cursor · Add to Windsurf

Before you start

  • Node.js v18 or later with npx available on PATH
  • A Pinecone API key, generated in the Pinecone console at https://app.pinecone.io (set as PINECONE_API_KEY)
  • A Pinecone index created with integrated inference (required for the upsert/search record tools)
  • An MCP-compatible client such as Cursor, Claude Desktop, or the Gemini CLI

About Pinecone Developer MCP Server

The Pinecone Developer MCP Server is the official Model Context Protocol server from Pinecone, letting AI coding assistants and agents work directly against your Pinecone projects. It bundles documentation search, index management, and vector data operations (upsert, search, rerank) behind a single stdio MCP connection.

Because Pinecone's integrated inference handles embedding generation server-side, you can upsert and query records as plain text — the server creates indexes, writes records, and runs semantic search with metadata filtering and reranking without you managing an embedding model yourself. It also exposes a search-docs tool so an assistant can ground answers in current Pinecone documentation.

It is designed for development-time use inside clients like Cursor, Claude Desktop, and the Gemini CLI, so you can scaffold and debug a RAG pipeline conversationally. Note that only indexes created with integrated inference are supported — standalone/non-inference indexes will not work with the record tools.

Tools & capabilities (9)

search-docs

Search the official Pinecone documentation.

list-indexes

List all indexes in the Pinecone project.

describe-index

View the configuration of a specific index.

describe-index-stats

Get record/data statistics and namespaces for an index.

create-index-for-model

Create a new index with integrated inference for a chosen model.

upsert-records

Insert or update text records, embedding them via integrated inference.

search-records

Run a semantic text query with metadata filtering and optional reranking.

cascading-search

Search across multiple indexes and merge results.

rerank-documents

Rerank a set of results using a specialized reranking model.

When to use it

  • Use it when you want an AI assistant to scaffold and iterate on a Pinecone-backed RAG pipeline conversationally.
  • Use it when you need to inspect indexes, namespaces, and record stats without leaving your editor.
  • Use it when you want to upsert and semantically search plain-text records without managing your own embedding model.
  • Use it when you want results reranked or queried across multiple indexes in one step.
  • Use it when you want an assistant to answer Pinecone API questions grounded in current docs.

Quick setup

  1. 1Create a Pinecone API key in the console at https://app.pinecone.io.
  2. 2Add the server to your client's MCP config, e.g. command `npx` with args `-y @pinecone-database/mcp`.
  3. 3Set the PINECONE_API_KEY environment variable in the server's env block.
  4. 4Restart your MCP client (Cursor, Claude Desktop) so it picks up the new server.
  5. 5Verify by asking the assistant to list your indexes or search the Pinecone docs.

Security notes

The PINECONE_API_KEY grants full access to your Pinecone project's indexes and data, so store it in env vars and rotate if leaked. Index-mutating tools (create/upsert) can change live data, so review agent actions before granting write access.

Pinecone Developer MCP Server FAQ

Does it work with any Pinecone index?

No. The record tools (upsert-records, search-records) only support indexes created with integrated inference. Standalone or non-inference indexes are not supported.

How does it authenticate?

Via a Pinecone API key passed as the PINECONE_API_KEY environment variable. The key is scoped to the project it was generated in.

Which clients are supported?

The README documents Cursor (.cursor/mcp.json), Claude Desktop (claude_desktop_config.json), and the Gemini CLI (installed as an extension). Any stdio MCP client should work.

Do I need to install anything globally?

No. The recommended setup runs the package via npx (`npx -y @pinecone-database/mcp`), so it is fetched on demand; you only need Node.js 18+.

Alternatives to Pinecone Developer MCP Server

Compare all alternatives →

Official MCP server providing persistent, file-backed knowledge-graph memory across sessions.

Verified
stdio (local)
No auth
TypeScript
9 tools
Updated 5 months agoRepo

Structured step-by-step reasoning tool for breaking problems into revisable thought sequences.

Verified
stdio (local)
No auth
TypeScript
1 tool
Updated 5 months agoRepo

Fully managed remote server for AWS docs, blogs, What's-New and Well-Architected guidance — no key.

Verified
HTTP (remote)
No auth
Hosted
6 tools
Updated 5 months agoRepo