Skip to main content
The MCP Server acts as the intelligent bridge between AI Agents (like Claude, Cursor, Windsurf) and the DataDot backend. It implements the Model Context Protocol to expose your data and capabilities as standardized tools.

​
Primary Responsibilities

Tool Exposure

Provides callable functions for document retrieval (query_content), file management, and citation verification.

Agent Integration

Allows external AI agents to seamlessly interact with your workspace contexts.

Secure Auth

Validates agent credentials against the main backend to ensure secure access.

State Sync

Keeps the agent’s understanding of the file system in sync with the actual workspace state.

​
Project Structure

The MCP server is a standalone Python application that communicates with the main backend. 
mcp/
├── app/
│   ├── tools/              # MCP Tools implementations (RAG, File, System)
│   ├── prompts/            # Pre-configured templates for Agents
│   ├── middleware/         # Logging and Integrity checks
│   ├── auth.py             # API Key validation logic
│   ├── server.py           # FastMCP server instance setup
│   └── config.py           # Configuration settings
├── main.py                 # Application entry point
└── Dockerfile              # Container definition

​
Getting Started

To configure and extend the MCP server, explore the detailed guides:

Architecture

Understand how the MCP server connects Agents to the Backend.

Tools Reference

List of available tools and their capabilities.

Prompts

Available system prompts for guiding agent behavior.