@flowrag/mcp

MCP (Model Context Protocol) server for FlowRAG. Exposes your knowledge base to AI assistants like Claude, Kiro, and Copilot.

Quick Start

npx @flowrag/mcp --data ./data --docs ./content

Or with a config file:

npx @flowrag/mcp --config ./flowrag.config.json

Configuration

Create flowrag.config.json:

{
  "data": "./data",
  "docs": "./content",
  "schema": {
    "entityTypes": ["SERVICE", "DATABASE", "PROTOCOL"],
    "relationTypes": ["USES", "PRODUCES", "CONSUMES"]
  },
  "embedder": { "provider": "local" },
  "extractor": { "provider": "gemini" }
}

API keys go in .env (auto-loaded from config directory or cwd):

GEMINI_API_KEY=your-key-here

CLI Flags

| Flag | Description | Default | |------|-------------|---------| | --config <path> | Config file path | ./flowrag.config.json | | --data <path> | Data directory | ./data | | --docs <path> | Documents directory | — |

Priority: CLI flags > config file > defaults.

AI Assistant Setup

Claude Desktop / Kiro

Add to mcp.json:

{
  "mcpServers": {
    "flowrag": {
      "command": "npx",
      "args": ["@flowrag/mcp", "--config", "/path/to/flowrag.config.json"]
    }
  }
}

Tools

| Tool | Description | |------|-------------| | flowrag_index | Index documents into the knowledge base | | flowrag_search | Search with dual retrieval (vector + graph) | | flowrag_entities | List or filter entities in the knowledge graph | | flowrag_relations | Get relations for a specific entity | | flowrag_trace | Trace data flow upstream or downstream | | flowrag_path | Find shortest path between two entities | | flowrag_stats | Get index statistics |

Resources

| Resource | Description | |----------|-------------| | flowrag://schema | Current schema definition (JSON) |

Config Change Detection

After indexing, a flowrag.meta.json is saved in the data directory. On startup, the server compares the current config with metadata and warns about:

• Breaking: Embedder changed → re-index required
• Minor: Schema or extractor changed → new types apply on next index

Remote / HTTP Mode

Run as a centralized HTTP server for team use:

{
  "transport": "http",
  "port": 3000,
  "auth": { "token": "${FLOWRAG_AUTH_TOKEN}" },
  "storage": {
    "kv": { "provider": "redis", "url": "redis://redis.internal:6379" },
    "vector": { "provider": "opensearch", "node": "https://os:9200", "dimensions": 1024 },
    "graph": { "provider": "opensearch", "node": "https://os:9200" }
  }
}

Clients connect via URL:

{
  "mcpServers": {
    "flowrag": {
      "url": "https://flowrag.company.com/mcp",
      "headers": { "Authorization": "Bearer ${FLOWRAG_TOKEN}" }
    }
  }
}

A Dockerfile is included for Fargate/ECS deployment. See the deployment guide for details.

License

MIT