# Mcp Neo4j Vectordb

> MCP server that mimics a pure vector database using Neo4j - vector search, fulltext search, NO graph traversal. Designed to compare Vector RAG vs GraphRAG effectiveness. Uses OpenAI embeddings. Built with FastMCP.

- **Type:** MCP server
- **Install:** `agentstack add mcp-neo4j-field-mcp-neo4j-vectordb`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [neo4j-field](https://agentstack.voostack.com/s/neo4j-field)
- **Installs:** 0
- **Category:** [Databases](https://agentstack.voostack.com/c/databases)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [neo4j-field](https://github.com/neo4j-field)
- **Source:** https://github.com/neo4j-field/mcp-neo4j-vectordb

## Install

```sh
agentstack add mcp-neo4j-field-mcp-neo4j-vectordb
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# Neo4j VectorDB MCP Server

An MCP server that uses Neo4j as a **pure vector database** — no graph traversal, no relationships.

## Purpose

This server is designed to **compare Vector RAG vs Graph RAG**. It provides the same search capabilities as `mcp-neo4j-graphrag` but intentionally hides graph features, allowing you to measure the value that graph context adds to LLM responses.

| Feature | `mcp-neo4j-vectordb` (This) | `mcp-neo4j-graphrag` |
|---------|----------------------------|----------------------|
| Vector search | ✅ | ✅ |
| Fulltext search | ✅ | ✅ |
| Graph traversal | ❌ | ✅ |
| Cypher queries | ❌ | ✅ |
| Relationships | ❌ Hidden | ✅ Visible |

## Installation

### Step 1: Download the Repository

```bash
git clone https://github.com/guerinjeanmarc/mcp-neo4j-vectordb.git
cd mcp-neo4j-vectordb
```

### Step 2: Configure Claude Desktop

Edit the configuration file:
- **macOS/Linux:** `~/Library/Application Support/Claude/claude_desktop_config.json`
- **Windows:** `%APPDATA%\Claude\claude_desktop_config.json`

Add this server configuration (update the path to where you cloned the repo):

```json
{
  "mcpServers": {
    "neo4j-vectordb": {
      "command": "uv",
      "args": [
        "--directory",
        "/path/to/mcp-neo4j-vectordb",
        "run",
        "mcp-neo4j-vectordb"
      ],
      "env": {
        "NEO4J_URI": "neo4j+s://demo.neo4jlabs.com",
        "NEO4J_USERNAME": "recommendations",
        "NEO4J_PASSWORD": "recommendations",
        "NEO4J_DATABASE": "recommendations",
        "OPENAI_API_KEY": "sk-...",
        "EMBEDDING_MODEL": "text-embedding-ada-002"
      }
    }
  }
}
```

### Step 3: Reload Configuration

Quit and restart Claude Desktop to load the new configuration.

## Tools

### `get_searchable_content`

Discover available indexes and searchable node properties (no relationships shown).

💡 The agent should automatically call this tool first to understand what can be searched.

**Example prompt:**
> "What can I search in this database?"

### `vector_search`

Semantic similarity search.

**Example prompt:**
> "Find movies about 'a hero's journey in space'"

### `fulltext_search`

Keyword search with Lucene syntax.

**Example prompt:**
> "Where was Tom Hanks born??"

### `get_node_by_id`

Retrieve a node's properties by ID (no relationships).

**Example prompt:**
> "Get the full details of the first movie from my search results"

---

## Comparison Workflow: Vector RAG vs Graph RAG

Use both servers to compare how graph context improves LLM responses.

### Prerequisites

Download both repositories:

```bash
git clone https://github.com/guerinjeanmarc/mcp-neo4j-vectordb.git
git clone https://github.com/guerinjeanmarc/mcp-neo4j-graphrag.git
```

### Step 1: Configure Both Servers in Claude Desktop

Edit `~/Library/Application Support/Claude/claude_desktop_config.json`:

```json
{
  "mcpServers": {
    "neo4j-vectordb": {
      "command": "uv",
      "args": [
        "--directory",
        "/path/to/mcp-neo4j-vectordb",
        "run",
        "mcp-neo4j-vectordb"
      ],
      "env": {
        "NEO4J_URI": "neo4j+s://demo.neo4jlabs.com",
        "NEO4J_USERNAME": "recommendations",
        "NEO4J_PASSWORD": "recommendations",
        "NEO4J_DATABASE": "recommendations",
        "OPENAI_API_KEY": "sk-...",
        "EMBEDDING_MODEL": "text-embedding-ada-002"
      }
    },
    "neo4j-graphrag": {
      "command": "uv",
      "args": [
        "--directory",
        "/path/to/mcp-neo4j-graphrag",
        "run",
        "mcp-neo4j-graphrag"
      ],
      "env": {
        "NEO4J_URI": "neo4j+s://demo.neo4jlabs.com",
        "NEO4J_USERNAME": "recommendations",
        "NEO4J_PASSWORD": "recommendations",
        "NEO4J_DATABASE": "recommendations",
        "OPENAI_API_KEY": "sk-...",
        "EMBEDDING_MODEL": "text-embedding-ada-002"
      }
    }
  }
}
```

### Step 2: Test with VectorDB Only

1. In Claude Desktop, click the **"Search and Tools"** button in the conversation interface
2. **Disable** `neo4j-graphrag` tools (keep only `neo4j-vectordb` active)
3. Ask the test question:

> "Find movies about artificial intelligence, and tell me which directors have made multiple AI films and what other genres they typically work in."

4. Note Claude's answer and what tools it used

### Step 3: Test with GraphRAG Only

1. Click **"Search and Tools"** again
2. **Disable** `neo4j-vectordb` tools and **enable** `neo4j-graphrag` tools
3. Ask:

> "You have now access to the Neo4j tools, please answer the same question."

4. Note Claude's answer and what tools it used

### Step 4: Compare Results

Ask Claude to compare:

> "Compare your two previous answers. What approach gave you better context to answer the question? Please explain what technology is better to answer the question: VectorDB or Neo4j?"

### Example Questions for Comparison

| Question | Vector RAG | Graph RAG |
|----------|------------|-----------|
| "What is The Matrix about?" | ✅ Good | ✅ Good |
| "Find sci-fi movies" | ✅ Good | ✅ Good |
| "Which actors worked with both Spielberg and Nolan?" | ⚠️ Limited | ✅ Excellent |
| "What genres does Tom Hanks typically act in?" | ⚠️ Limited | ✅ Excellent |
| "Find movies similar to Inception and show their directors' other work" | ⚠️ Limited | ✅ Excellent |

---

## Configuration

### Environment Variables

| Variable | Required | Default | Description |
|----------|----------|---------|-------------|
| `NEO4J_URI` | Yes | `bolt://localhost:7687` | Neo4j connection URI |
| `NEO4J_USERNAME` | Yes | `neo4j` | Neo4j username |
| `NEO4J_PASSWORD` | Yes | `password` | Neo4j password |
| `NEO4J_DATABASE` | No | `neo4j` | Database name |
| `EMBEDDING_MODEL` | No | `text-embedding-3-small` | Embedding model |

### Embedding Providers

Supports all [LiteLLM embedding providers](https://docs.litellm.ai/docs/embedding/supported_embedding):
- OpenAI: `text-embedding-ada-002`, `text-embedding-3-small`
- Azure: `azure/deployment-name`
- Bedrock: `bedrock/amazon.titan-embed-text-v1`
- Cohere: `cohere/embed-english-v3.0`
- Ollama: `ollama/nomic-embed-text`

---

## Related

- [mcp-neo4j-graphrag](https://github.com/guerinjeanmarc/mcp-neo4j-graphrag) — Full GraphRAG with graph traversal
- [Neo4j MCP Documentation](https://neo4j.com/developer/genai-ecosystem/model-context-protocol-mcp/)
- [Official Neo4j MCP Server](https://github.com/neo4j/mcp)

## License

MIT License

## Source & license

This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [neo4j-field](https://github.com/neo4j-field)
- **Source:** [neo4j-field/mcp-neo4j-vectordb](https://github.com/neo4j-field/mcp-neo4j-vectordb)
- **License:** MIT

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/mcp-neo4j-field-mcp-neo4j-vectordb
- Seller: https://agentstack.voostack.com/s/neo4j-field
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.