Neo4j Tools

The Neo4j provider executes Cypher statements against a Neo4j database through the HTTP Query API (/db/{database}/query/v2). Use it in a workflow to read from or write to a graph database — for example matching nodes, creating relationships, or running parameterized queries with values pulled from upstream blocks.

Overview

Operations

Configuration

neo4j_run_query

Runs a single Cypher statement (optionally parameterized) against the target database. It sends a POST to {dbUrl}/db/{database}/query/v2 with a Basic Auth header built from the username and password.

Notes:

• password is a secret credential — mark it as a secret/environment variable rather than hardcoding it.
• dbUrl, username, password, and database are user-only, so the workflow-builder agent cannot fill them from the LLM — they must be supplied as block inputs.
• statement and parameters are user-or-llm, so they may be generated by the agent or set directly.

Outputs

neo4j_run_query

• data (json) — The Neo4j Query API result (the parsed JSON body returned by the /query/v2 endpoint).
• metadata (json) — Response metadata (currently returned as an empty object).

YAML Example

neo4j_1:
  type: neo4j
  name: "Neo4j"
  inputs:
    operation: "neo4j_run_query"
    dbUrl: "https://your-id.databases.neo4j.io"
    username: "neo4j"
    password: "{{NEO4J_PASSWORD}}"
    database: "neo4j"
    statement: "MATCH (p:Person {name: $name}) RETURN p LIMIT 10"
    parameters:
      name: "Jane"
  connections:
    outgoing:
      - target: next-block-id

Tips

• Use parameterized Cypher ($name) with the parameters field instead of string-concatenating values into statement — this avoids injection issues and lets you wire upstream values like {{previousBlock.field}} into parameters.
• Authentication is Basic Auth: the tool Base64-encodes username:password into the Authorization header. Store the password as an environment variable such as {{NEO4J_PASSWORD}} rather than inline.
• Point dbUrl at your HTTP Query API host (e.g. an AuraDB instance https://<id>.databases.neo4j.io). The tool always targets the /db/{database}/query/v2 path, so set database (default neo4j) to match the database you want to query.
• The result graph/records live under the data output — reference them downstream with {{neo4j_1.data}}.