Summary
Graph queries with the schema one call away.
Writing Cypher against an unfamiliar graph is guesswork until the node types and their relationships are visible, which is exactly what the schema tool returns. Reads and writes sit on separate tools, so a mutation is always a deliberate choice.
What it is
A Python MCP server for Neo4j that runs read and write Cypher against a database selected on the command line, over Server-Sent Events or stdio.
What you get
- Read queries executed, with results returned as an array of objects
- Write queries on a separate tool, returning a summary counter of nodes updated and relationships created
- The whole schema listed: every node type with its attributes, their names and types, and its relationships to other node types
- Remote databases selectable at launch, with SSE bound to a default host and port
- A demo prompt that walks through generating a schema and sample data
Requirements
Python 3.12+ and pip install -r requirements.txt, then launch it with the database's address, the account to connect as, that account's password, and `--database` for which database to open. Neo4j must be running and reachable for remote connections.