n8n-nodes-neo4j-fixed

A fixed and fully-featured n8n community node for Neo4j — the leading graph database.

This is a clean rewrite of n8n-nodes-neo4j-extended that fixes the logWrapper import error caused by internal @n8n/n8n-nodes-langchain subpath imports. This package uses only the public n8n-workflow API and the official neo4j-driver.

Features

| Feature | Details | |---------|---------| | 🗄️ Graph DB | Execute Cypher, Create/Update/Delete nodes, Create relationships, Get schema | | 🔍 Vector Store | Similarity search, hybrid search, add texts, add documents, delete by ID | | 🤖 AI Tool | Connects to LangChain agents via the Embedding sub-input | | 🚀 Auto-Index | Automatically creates vector indexes when missing | | 🔒 Security | Input validation prevents Cypher injection | | 📋 Dynamic Dropdowns | Lists existing vector indexes |

Installation

In n8n → Settings → Community Nodes → Install:

n8n-nodes-neo4j-fixed

Or via npm:

npm install n8n-nodes-neo4j-fixed

Why This Fix Works

The original n8n-nodes-neo4j-extended imported:

import { logWrapper } from '@n8n/n8n-nodes-langchain/dist/utils/logWrapper';

This is an internal path that newer versions of @n8n/n8n-nodes-langchain removed from their exports map, causing the load failure.

This package achieves the same AI embedding integration using only the public NodeConnectionType.AiEmbedding and getInputConnectionData() API from n8n-workflow.

Credentials

| Field | Description | Default | |-------|-------------|---------| | Connection URI | Neo4j connection string | neo4j://localhost:7687 | | Username | Database username | neo4j | | Password | Database password | — | | Database | Database name | neo4j |

Supports neo4j://, bolt://, neo4j+s://, bolt+s:// protocols.

Operations

Graph Database

• Execute Query — Run any Cypher query with optional parameters
• Create Node — Create a node with label and properties
• Update Node — Update properties of an existing node by match key
• Delete Node — Detach-delete a node by match key
• Create Relationship — Create a relationship between two nodes
• Get Schema — Retrieve labels, relationship types, and property keys

Vector Store

Requires an Embedding model connected to the Embedding sub-input

• Similarity Search — Vector (cosine) or Hybrid (vector + fulltext) search
• Add Texts — Embed and store plain text strings
• Add Documents — Embed and store { pageContent, metadata } objects
• Delete by ID — Remove vectors by their element ID

Index Management

• List Indexes — Show all vector indexes in the database
• Create Index — Create a vector index with dimension and similarity function
• Delete Index — Drop a vector index
• Get Index Info — Get details about a specific index

Vector Dimensions

Neo4j supports dimensions 1–2048.

| Model | Dimension | |-------|-----------| | OpenAI text-embedding-ada-002 | 1536 | | OpenAI text-embedding-3-small | 1536 | | sentence-transformers/all-MiniLM-L6-v2 | 384 | | sentence-transformers/all-mpnet-base-v2 | 768 | | Cohere Embed v3 | 1024 | | GigaChat Embeddings | 2048 |

Note: OpenAI text-embedding-3-large (3072D) exceeds Neo4j's limit. Use text-embedding-3-small instead or reduce dimensions.

Security

• Identifier validation: ^[a-zA-Z_$][a-zA-Z0-9_$]*$
• 255-character limit on all identifier fields
• Parameterized queries for all user-supplied values
• No eval or dynamic code execution

Compatibility

• n8n ≥ 1.0.0
• Neo4j ≥ 4.4 (5.x recommended for full vector support)
• Node.js ≥ 18

License

MIT