How to Connect Cursor to Your Project Roadmap

MCP (Model Context Protocol) is changing how AI coding agents understand your project. Instead of copying and pasting context into every chat, your AI can pull live information from external tools.

CollabLearn now has MCP support. This means Cursor, Claude Desktop, and VS Code Copilot can connect directly to your project and read your roadmap, tasks, and specs in real time.

What MCP actually does

Think of MCP as a bridge between your AI agent and other tools. When you connect Cursor to CollabLearn via MCP, the AI gets access to new "tools" it can call:

• get_context - Fetches your product specs and technical architecture
• get_active_plan - Gets your current roadmap as markdown
• get_tasks - Lists all tasks with their status
• update_task - Marks tasks as done or in progress

Your AI can call these tools during a conversation. So when you ask "what should I work on next?", it can actually check your roadmap and give you a real answer.

Setting it up

The setup takes about 2 minutes:

1. Get your API key

Go to collablearn.in/docs/mcp and generate an API key. This authenticates your agent with your CollabLearn account.

2. Add the config to Cursor

Open Cursor settings (Cmd+, on Mac) and find the MCP section. Add this config:

{
  "mcpServers": {
    "collablearn": {
      "command": "npx",
      "args": ["mcp-remote", "https://api.collablearn.in/mcp"],
      "env": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}

Replace YOUR_API_KEY with the key you generated.

3. Restart Cursor

After restart, CollabLearn will appear in your available MCP tools. Try asking "What are my current tasks?" and watch it fetch from your actual project.

Claude Desktop and VS Code

The setup is similar for Claude Desktop. Edit your claude_desktop_config.json and add the same config structure.

For VS Code with Copilot, create a .vscode/mcp.json file in your project root.

CollabLearn's /docs/mcp page has copy-paste configs for all three editors.

The two-way sync

The really useful part is that it syncs both ways.

When your AI agent completes a task, it can call update_task to mark it done. That update shows up immediately in CollabLearn's UI. Your teammates see it. Your roadmap updates.

If someone on your team marks a task done in the CollabLearn UI, your AI agent sees that change next time it fetches context.

You don't have to manually keep anything in sync. The AI works with live data.

When to use this vs context files

Use context files when:

• You want a simple setup with no external connections
• Your project specs don't change often
• You're working offline a lot

Use MCP when:

• You want live, always-current context
• You're working with a team and need real-time sync
• You want your AI to actually update your project status

They're not mutually exclusive. You can have context files for static stuff (like coding conventions) and use MCP for dynamic stuff (like current tasks and progress).

Connect your AI agent to CollabLearn via MCP. Go to /docs/mcp to get started.