At a glance
- Availability: Experimental (how to enable).
- Auth: API key.
- Connection: The key from
PINECONE_API_KEYis sent as theApi-Keyheader. - Docs: https://docs.pinecone.io/reference/api/authentication
Credentials
Set these per environment. See Connect an integration.| Variable | Required | Description |
|---|---|---|
PINECONE_API_KEY | Yes | Pinecone API key (starts with pcsk_) Docs. |
Setup
- Create a Pinecone account: Sign up at https://app.pinecone.io - the free Starter plan includes serverless indexes for testing.
- Create an API key: In the Pinecone console, select your project, open API Keys, and create a key (pcsk_…). Copy it immediately.
- Store the key: Add it to your .env file as PINECONE_API_KEY=pcsk_…
- Verify access: Run the List Indexes tool. To read or write vectors, first run Describe Index and use the returned ‘host’ value as the indexHost parameter for Query Vectors / Upsert Vectors.
- Pinecone is two-step: control-plane tools (list/describe/create index) hit api.pinecone.io, while query/upsert hit the per-index host returned by Describe Index
- All requests send the X-Pinecone-API-Version header (default 2025-01)
- Embeddings must be generated separately (e.g. with OpenAI) - vectors you query or upsert must match the index dimension
Tools
| Tool | Access | Description |
|---|---|---|
| List Indexes | Read | List all indexes in the Pinecone project |
| Describe Index | Read | Get an index’s configuration and its data-plane host (needed for Query Vectors and Upsert Vectors) |
| Create Index | Write | Create a new serverless or pod-based index |
| Query Vectors | Read | Query an index for the nearest vectors (run Describe Index first to get the index host) |
| Upsert Vectors | Write | Write vectors into an index (run Describe Index first to get the index host) |
Example prompts
- List my Pinecone indexes with their dimensions, metrics, and status.
- Describe my Pinecone index to get its host, then query it with an embedding vector and show the top matches with metadata.