Agile Luminary MCP Server

This is a Model Context Protocol (MCP) server that connects AI clients (like Cursor or Claude Desktop) to the Agile Luminary project management system hosted at https://agileluminary.com.

What is MCP?

The Model Context Protocol (MCP) is a standardized way for AI applications to connect to external data sources and services. This server acts as a bridge between your AI client and the Agile Luminary REST API, allowing you to retrieve project details, work assignments, and documentation directly within your AI conversations.

Architecture

AI Client (Cursor/Claude) → Local MCP Server → Agile Luminary API (localhost:5006)

Features

This MCP server provides three main tools that fetch data from the Agile Luminary API:

📚 getRelatedDocuments

• Purpose: Retrieve related documents based on a search string
• Parameters: searchString (required) - The text to search for in documents
• Use Case: Find relevant documentation and specifications related to your current work

📋 getCurrentWork

• Purpose: Fetches work currently assigned to the user
• Parameters: None
• Use Case: See what tasks and user stories are actively assigned to you

🔍 getPastWork

• Purpose: Retrieves completed work related to user stories based on search criteria
• Parameters: searchString (required) - Search term to find relevant past work
• Use Case: Review historical context and completed tasks for better decision-making

API Endpoints

The server connects to the following Agile Luminary API endpoints:

• POST http://localhost:5006/bend/mcp/documents/search - Search documents
• GET http://localhost:5006/bend/mcp/userstories/current - Get current work
• POST http://localhost:5006/bend/mcp/userstories/search - Search past work

Benefits

• Real-time Data: Always get the latest project information from your deployed system
• Context-Aware AI: Your AI assistant has full context of your project status and documentation
• Seamless Integration: Works directly within your development environment
• Secure: Server handles authentication and API communication locally
• Search Capabilities: Find relevant documents and past work through intelligent search

Authentication

The server uses the LUMINARY_API_KEY environment variable for authentication. This key is passed in the Authorization header for all API requests. Make sure to set this environment variable before running the server.

Error Handling

The server includes comprehensive error handling for:

• Network connectivity issues
• API authentication failures
• Malformed responses
• Service unavailability
• Missing API keys

Installation & Setup

Prerequisites

• Node.js (v16 or higher)
• An AI client that supports MCP (Cursor IDE, Claude Desktop, etc.)

Install Dependencies

npm install @modelcontextprotocol/sdk zod

Environment Variables

Set your API key as an environment variable:

export LUMINARY_API_KEY=your_api_key_here

Run the Server

node server.js

The server will start and listen for connections from your AI client via stdio transport.

Configuration

For Cursor IDE

Add this to your MCP configuration:

{
  "mcpServers": {
    "agile-luminary": {
      "command": "node",
      "args": ["path/to/your/server.js"],
      "env": {
        "LUMINARY_API_KEY": "your_api_key_here"
      }
    }
  }
}

For Claude Desktop

Add this to your claude_desktop_config.json:

{
  "mcpServers": {
    "agile-luminary": {
      "command": "node",
      "args": ["path/to/your/server.js"],
      "env": {
        "LUMINARY_API_KEY": "your_api_key_here"
      }
    }
  }
}

Usage Examples

Once connected, you can use natural language in your AI client:

• "Search for documents about authentication"
• "Show me my current work assignments"
• "Find past work related to user login functionality"
• "What documentation exists about the payment system?"
• "Based on my current work, what similar tasks were completed before?"