Connects Claude to Pinecone's vector database through five core operations: semantic search, document reading, listing documents, index statistics, and document processing with automatic chunking and embedding. Uses Pinecone's inference API for embeddings and includes a token-based chunker borrowed from LangChain. You'd reach for this when building RAG applications where Claude needs to store, retrieve, and search through document embeddings. Setup requires a Pinecone API key and index name, then you can have Claude ingest documents, perform semantic searches, and pull relevant context from your vector store.
claude mcp add --transport stdio sirmews-mcp-pinecone uvx mcp-pinecone