# Pinecone

> Managed vector database for semantic search and retrieval in AI agents.

[Canonical HTML page](https://top-mcps.com/mcp/pinecone) · [server.json](https://top-mcps.com/mcp/pinecone.json) · [methodology](https://top-mcps.com/about/methodology)

## Install

### Claude Desktop — `claude_desktop_config.json`

Paste under mcpServers. Fully quit and reopen Claude after editing.

```json
{
  "mcpServers": {
    "pinecone": {
      "command": "npx",
      "args": [
        "-y",
        "@pinecone-database/mcp"
      ],
      "env": {
        "PINECONE_API_KEY": "${PINECONE_API_KEY}"
      }
    }
  }
}
```

### Claude Code — `CLI or .mcp.json`

Run from your repo. Commit .mcp.json to share with your team.

```shell
# export PINECONE_API_KEY=YOUR_API_KEY
claude mcp add pinecone -- npx -y @pinecone-database/mcp
```

### Cursor — `.cursor/mcp.json`

Global path: ~/.cursor/mcp.json. Reload window after editing.

```json
{
  "mcpServers": {
    "pinecone": {
      "command": "npx",
      "args": [
        "-y",
        "@pinecone-database/mcp"
      ],
      "env": {
        "PINECONE_API_KEY": "${PINECONE_API_KEY}"
      }
    }
  }
}
```

### VS Code — `.vscode/mcp.json`

VS Code uses the "servers" key (not "mcpServers").

```jsonc
{
  "servers": {
    "pinecone": {
      "command": "npx",
      "args": [
        "-y",
        "@pinecone-database/mcp"
      ],
      "env": {
        "PINECONE_API_KEY": "${PINECONE_API_KEY}"
      }
    }
  }
}
```

### Windsurf — `~/.codeium/windsurf/mcp_config.json`

Open via Cascade → hammer icon → Configure.

```json
{
  "mcpServers": {
    "pinecone": {
      "command": "npx",
      "args": [
        "-y",
        "@pinecone-database/mcp"
      ],
      "env": {
        "PINECONE_API_KEY": "${PINECONE_API_KEY}"
      }
    }
  }
}
```

### Cline — `cline_mcp_settings.json`

Open via the Cline sidebar → MCP Servers → Edit.

```json
{
  "mcpServers": {
    "pinecone": {
      "command": "npx",
      "args": [
        "-y",
        "@pinecone-database/mcp"
      ],
      "env": {
        "PINECONE_API_KEY": "${PINECONE_API_KEY}"
      }
    }
  }
}
```

### Continue — `~/.continue/config.json`

Continue uses modelContextProtocolServers with a transport block.

```json
{
  "experimental": {
    "modelContextProtocolServers": [
      {
        "transport": {
          "type": "stdio",
          "command": "npx",
          "args": [
            "-y",
            "@pinecone-database/mcp"
          ],
          "env": {
            "PINECONE_API_KEY": "${PINECONE_API_KEY}"
          }
        }
      }
    ]
  }
}
```

### Codex CLI — `~/.codex/config.toml`

Codex uses TOML. Each server is a [mcp_servers.<name>] subtable.

```shell
# ~/.codex/config.toml
[mcp_servers.pinecone]
command = "npx"
args = [
  "-y",
  "@pinecone-database/mcp",
]
env = { PINECONE_API_KEY = "${PINECONE_API_KEY}" }
```

### Zed — `~/.config/zed/settings.json`

Zed calls them "context_servers". Settings live-reload on save.

```jsonc
{
  "context_servers": {
    "pinecone": {
      "command": {
        "path": "npx",
        "args": [
          "-y",
          "@pinecone-database/mcp"
        ]
      },
      "env": {
        "PINECONE_API_KEY": "${PINECONE_API_KEY}"
      }
    }
  }
}
```

### ChatGPT — `ChatGPT → Apps directory`

Pinecone doesn't ship a hosted HTTPS endpoint today. ChatGPT supports remote MCP servers only — to use this server in ChatGPT you'll need to deploy it to a public HTTPS URL first (e.g. via Cloudflare Workers or Vercel) or wait for an official remote build.

```none

```

## At a glance

- **Maintainer:** Pinecone
- **Transport:** stdio
- **Auth model:** API key
- **Required secrets:** PINECONE_API_KEY
- **Supported clients:** Claude, Cursor, Any MCP-compatible client
- **License:** Apache-2.0
- **Language:** TypeScript
- **Latest version:** latest
- **Last verified:** 2026-05-26
- **GitHub stars:** 68 (fetched 2026-05-26T10:50:23.074Z)
- **Score:** 80/100 (rubric 2026-04 — see https://top-mcps.com/about/methodology)
- **Source:** https://github.com/pinecone-io/pinecone-mcp

## Security & scope

- **Access scope:** read-write
- **Sandbox:** API key grants full project access. Use separate keys per environment.
- **Gotchas:**
  - No row-level security — anyone with the key can read/write any vector in the project.
  - Embeddings cost money on every upsert; idempotent upserts matter.

## Quick answer

**What it does.** Provides MCP tools for index management, vector upserts, similarity queries, namespace operations, and metadata filtering against a Pinecone serverless or dedicated cluster.

**Best for:**
- Semantic search over private corpora
- Production-grade retrieval
- Multi-tenant agent memory
- Hybrid metadata + vector queries

**Not for:**
- Self-hosted-only requirements
- Hobby projects (free tier is generous but capped)
- Workflows that need full-text alongside vectors

## Description

The Pinecone MCP exposes Pinecone's vector-database operations: create indexes, upsert vectors with metadata, query by similarity, and filter on metadata. The fastest way to give an agent persistent retrieval over your own corpus.

## Why it matters

Most "give the agent memory" workflows boil down to embedding + vector search. Pinecone is the most-used managed vector DB; the MCP wrapper means an agent can store and retrieve without bespoke glue code.

## Key features

- Managed vector DB
- Serverless and dedicated tiers
- Metadata filters
- Namespaces for multi-tenancy
- Sparse-dense hybrid
- Live re-ranking

## FAQ

### Do I need to bring my own embeddings?

Yes by default — embed with OpenAI, Anthropic, or open-source models, then upsert the vectors. Pinecone also exposes hosted embedding models in some plans.

### How does pricing scale?

Pinecone serverless charges by read/write units plus storage. For small indexes the free tier is usually enough; large agent corpora can run into hundreds of dollars per month.

### What about hybrid search?

Pinecone supports sparse-dense hybrid where you combine BM25-style sparse vectors with dense embeddings. Useful when keywords still matter (proper names, codes).

## Changelog

- **2026-05-26** — Refreshed install snippets and fact sheet; verified for 2026.
- **2025-02-10** — Initial directory listing.