lar-mcp

Legal Assistant RAG MCP Tools

MCP tools for Legal Assistant RAG - enables AI assistants to handle clients, cases and documents.

Overview

This repository contains Model Context Protocol (MCP) tools that integrate with the Legal Assistant RAG application. These tools allow AI assistants to interact with the Legal Assistant RAG backend, performing operations on clients, cases, and documents through a standardized interface.

Features

• Document Management: Upload, list, and delete legal documents for RAG processing
• Client Management: Create, edit, list, and delete client records
• Case Management: Create, edit, list, and delete legal cases
• RAG Capabilities: Ask questions about your legal documents using the RAG system
• Secure Authentication: JWT-based authentication with the backend

Requirements

• Node.js v18+ (v22+ recommended)
• TypeScript
• Access to a running Legal Assistant RAG backend

Installation

# Clone the repository
git clone https://github.com/kevinxe/lar-mcp.git
cd lar-mcp

# Install dependencies
npm install

Setup

Usage with Claude Desktop

To use this with Claude Desktop, add the following to your claude_desktop_config.json:

{
  "mcpServers": {
    "legal-assistant": {
      "command": "npx",
      "args": [
        "-y",
        "tsx",
        "path/to/your/src/main.ts"
      ],
      "env": {
        "API_EMAIL": "your-email@example.com",
        "API_PASSWORD": "your-password",
        "API_URL": "http://localhost:3000",
        "MCP_TRANSPORT": "stdio"
      }
    }
  }
}

Where:

• API_EMAIL: The email you use to log in to the Legal Assistant RAG system
• API_PASSWORD: The password for your Legal Assistant RAG account
• API_URL: The URL where the Legal Assistant RAG backend is deployed (default is http://localhost:3000 for local development)
• MCP_TRANSPORT: The transport method for the MCP protocol, can be "stdio" or "http"

The MCP_TRANSPORT value can be set to either "stdio" or "http", depending on your needs:

• stdio: Simpler for local development, runs on your local machine and communicates directly via standard input/output
• http: Offers more flexibility for distributed teams, can run locally or remotely and communicates over the network

Usage

These MCP tools are designed to be used with AI assistants that support the Model Context Protocol, such as Claude via Cursor or Claude Desktop.

Available Tools

Document Management

• lar-ask: Ask questions about your legal documents
• lar-upload-document: Upload a document from a URL
• lar-list-documents: List all available documents
• lar-delete-document: Delete a document by name

Client Management

• lar-list-clients: List all clients
• lar-create-client: Create a new client
• lar-edit-client: Edit an existing client
• lar-delete-client: Delete a client

Case Management

• lar-list-cases: List all cases
• lar-create-case: Create a new legal case
• lar-edit-case: Edit an existing case
• lar-delete-case: Delete a case

Development

# Run the MCP server locally
npm start

# Build for production
npm run build

Integration

This MCP server integrates with the Legal Assistant RAG application, a comprehensive web application built with C# (.NET 8.0) backend and Angular frontend. The MCP tools communicate with the backend API to perform operations on the system's data.

License

GNU General Public License (GPL)

Related Projects

• Legal Assistant RAG - The main web application that this MCP server integrates with