MCP/Pinecone Developer MCP
Pinecone Developer MCP logo

Pinecone Developer MCP

Connect AI coding assistants to Pinecone vector databases for documentation search, index management, and vector data operations.

ai-ml•Vector Database & Embeddings
GitHub72
License:Apache-2.0
VERIFIED
ByPinecone@Pinecone

About this Server

Develop with Pinecone, the vector database built for knowledgeable AI. Connects AI coding assistants to Pinecone vector databases for documentation search, index management, and data operations. Enables AI tools to help configure serverless vector indexes, generate integration code, upsert vector embeddings, and test semantic queries directly in your development environment.

Real-time Interaction

Direct low-latency RPC protocol communication with structured JSON schemas.

Scoped Sandboxing

Restricted capability bounds to prevent accidental modifications or leakage.

Security & Best Practices

API Key Isolation: Store PINECONE_API_KEY inside local environment variables or client configuration; never commit API keys to version control.
Namespace Partitioning: Use dedicated namespaces to partition development or evaluation vectors from production datasets.
Metadata Privacy: Avoid storing sensitive unencrypted personally identifiable information (PII) directly in vector metadata attributes.
Encrypted Communication: All index management calls and vector retrieval requests communicate over secure TLS endpoints.

When to Use (and When Not To)

Best For
  • •Developers building AI applications with vector search, RAG pipelines, and semantic retrieval
  • •Teams working with dense or sparse embeddings and high-performance similarity matching
  • •Engineers testing vector queries, top-k ranking, and metadata filter expressions in development
  • •AI coding assistant users developing Pinecone client workflows in TypeScript, Python, or Go
Not For / Limitations
  • •Relational SQL queries or complex multi-table joins (use PostgreSQL, Supabase, or Neon instead)
  • •Environments lacking a valid Pinecone API key or internet access to Pinecone cloud endpoints
  • •Ultra-high-throughput streaming pipelines ingesting millions of vectors simultaneously (use direct gRPC SDK clients)

Install & Setup

terminal command
$claude mcp add -e PINECONE_API_KEY='<your pinecone api key>' pinecone -- npx -y @pinecone-database/mcp

Exposed Tools & Endpoints

7 tools

Functions your AI agent can invoke via this Model Context Protocol server

search_docs

Search official Pinecone documentation, SDK guides, and vector search patterns for accurate implementation guidance.

list_indexes

Retrieve all serverless and pod-based vector indexes, status details, and compute endpoints in your Pinecone account.

create_index

Create a new serverless vector index with custom dimension, distance metric (cosine, euclidean, dotproduct), and cloud region.

describe_index

Fetch detailed index specification, host URL, status, and vector count statistics for a target index.

upsert_vectors

Upsert vector embeddings with unique record IDs, values, and rich metadata payloads into a specified index namespace.

query_vectors

Execute vector similarity search returning top-k nearest neighbors with similarity scores and matched metadata.

delete_vectors

Delete specific vectors by ID list or delete all vectors contained within a designated namespace.