Zettelkasten MCP Server

Example Use Cases

• 1

  Persistent, cross-session conversational memory combining semantic search with explicit relationship graphs.

• 2

  Perform semantic similarity retrieval (via Pinecone) alongside relationship-aware queries (via Neo4j) to provide richer context.

• 3

  Easy integration options (stdio for local integrations or SSE for web-based clients) to receive MCP events and responses.

Description

Zettelkasten is an open-source MCP server designed to persist and retrieve cross-session chat context using a hybrid approach: semantic vectors in Pinecone and relationship storage in Neo4j. It offers two transport modes (stdio and Server-Sent Events) for flexible integration with agents and frontends. The architecture is intended to provide fast, context-aware conversational interactions and scale to large chatbot workloads. Developers supply their own Pinecone, Neo4j, and OpenAI credentials to run the server locally or in production.