Skip to main content

Prerequisites

  • Node.js 18+
  • FalkorDB instance (running locally or remotely)
  • Claude Desktop app (for AI integration)

Running from npm

Add to your Claude Desktop config (~/Library/Application Support/Claude/claude_desktop_config.json on macOS):
Restart Claude Desktop and you’ll see the FalkorDB tools available!

Running with npx

You can run the server directly from the command line:

Docker Compose

Run FalkorDB and the MCP server together:
This starts both FalkorDB and the MCP server with health checks and persistent volumes.

Available MCP Tools

Once connected to Claude or another MCP-compatible AI assistant, you can:

Query Graphs

The query_graph tool supports a readOnly parameter to execute queries using GRAPH.RO_QUERY, ideal for:
  • Running queries on replica instances
  • Preventing accidental write operations
  • Ensuring data integrity in production environments

Manage Data

Explore Structure

Example Usage

Once connected, Claude can help you write and execute queries:

Frequently Asked Questions

You need Node.js 18+, a running FalkorDB instance (locally via Docker or FalkorDB Cloud), and an MCP-compatible AI client like Claude Desktop. No additional dependencies are required since the server runs via npx.
Add the MCP server configuration to your Claude Desktop config file (claude_desktop_config.json). Set the command to npx with args ['-y', '@falkordb/mcpserver@latest'] and provide your FalkorDB connection details as environment variables.
Yes. Run it directly with npx -y @falkordb/mcpserver using stdio transport, or set MCP_TRANSPORT=http and MCP_PORT=3005 for HTTP transport that any MCP-compatible client can connect to.
Clone the FalkorDB-MCPServer repository, copy .env.example to .env, configure your settings, and run docker compose up -d. This starts both FalkorDB and the MCP server with health checks and persistent volumes.
You can ask Claude to query graphs (‘Show me all people who know each other’), create data (‘Create a person named Alice who knows Bob’), explore structure (‘List all available graphs’), and run read-only queries on replica instances.