Pinecone

Pinecone connects Claude Code to your Pinecone vector database so you can create, inspect, and modify indexes without leaving the editor. It lists available indexes, reads index metadata, creates and deletes indexes, and uploads (upserts) or deletes vectors. You can fetch vectors by id, run similarity queries with text or precomputed vectors, and retrieve raw match scores and metadata for each hit.

The plugin sends queries in plain language or vector form, maps query results back to your data, and returns the original payloads and namespace info. It supports batching for bulk upserts, paginated fetches for large result sets, and optional score thresholds to filter matches. You can watch operations and get status responses for long-running jobs.

Use it when you want to prototype a semantic search feature, iterate on index schemas, or debug retrieval quality without switching to another console. For example, while building a customer support search you can upsert transcripts, run natural-language queries, inspect top matches and adjust namespaces or metadata on the spot — saving the time of toggling between Claude Code and the Pinecone dashboard.