DataBridge MCP Server

Overview

The MCP DataBridge Server integrates with DataBridge to enable ingestion and retrieval of contextual information from a local database, supporting persistent storage for AI applications. It implements the Model Context Protocol (MCP), enabling connections to different ML services through a unified interface.

Key Features

Available Tools

Quick Reference

Detailed Usage

use_mcp_tool({
  server_name: "databridge",
  tool_name: "ingest_data",
  arguments: {
    data: {
      "document_id": "doc123",
      "content": "This is the content of the document."
    },
    collection: "documents"
  }
});

use_mcp_tool({
  server_name: "databridge",
  tool_name: "retrieve_data",
  arguments: {
    query: "document_id = 'doc123'",
    collection: "documents"
  }
});

use_mcp_tool({
  server_name: "databridge",
  tool_name: "list_ml_services",
  arguments: {}
});

Installation

{
  "mcpServers": {
    "databridge": {
      "command": "python",
      "args": [
        "-m",
        "databridge.mcp"
      ]
    }
  }
}

When to Use This Server

DataBridge is a fit when an assistant needs a small, persistent store for contextual data rather than a full database connection. Because it ingests into a local database and exposes retrieval through one interface, it's well suited to:

• Session memory — Store documents, notes, or facts (ingest_data) during a session and retrieve them later (retrieve_data), giving the assistant durable context without external storage.
• Unified ML access — Route to different ML services through a single interface via list_ml_services, so the assistant can discover what's available instead of assuming.
• Lightweight RAG — Keep a document collection locally and retrieve relevant context on demand for grounded answers.

For file-level persistence see Filesystem MCP; for vector search over local embeddings see Chroma MCP and Milvus MCP.

Frequently Asked Questions

Sources