Integration· 4 min read

Using IntentDocs with Cursor

Cursor's agent is fast at turning instructions into code — the quality ceiling is the instructions. IntentDocs gives Cursor a real spec to build against: an interview extracts what you're building into a story map, data model, and build order, and Cursor reads it over MCP (Model Context Protocol) instead of you re-explaining the product in every chat.

To use IntentDocs with Cursor: create your spec via the AI interview, run the setup shown in your project dashboard to register the IntentDocs MCP server, then build story by story — Cursor reads the spec over MCP and syncs progress back into a built-versus-planned audit trail.

What is Cursor?

Cursor is the AI code editor from Anysphere — a VS Code-style IDE with an agent mode that plans and executes multi-file changes. (Anysphere)

How the integration works

1. Create your spec in IntentDocs

Run the AI interview (or import an existing GitHub repo) to generate the story map, data model, and build order.

2. Register the MCP server in .cursor/mcp.json

Your project's Start Building panel provides a ready-to-paste mcp.json block (plus the repo-link init command with your project's sync token). Create or edit .cursor/mcp.json in your project root, paste it in, and Cursor picks up the IntentDocs MCP server — the dashboard confirms once the agent can reach your project.

3. Build against the spec

In agent mode, ask Cursor for the next story. It reads the story's implementation plan and surrounding context over MCP, builds, and reports progress back — the story map tracks built versus planned as you go.

What Cursor gets from the spec

  • Query the spec over MCP — stories, data model, build order — instead of relying on pasted context that goes stale.
  • Work story by story with per-story implementation plans.
  • Sync progress back so the story map reflects what actually shipped.
  • Share one spec across tools — the same project serves Claude Code, Cursor, and any MCP client.

What IntentDocs is

IntentDocs is a spec-driven development platform: an AI interview turns your idea into a structured spec — a visual story map, data model, and build order — and your coding agent (Claude Code, Cursor, or any MCP client) builds from it over MCP, with an audit trail of what was built versus what was planned. 7-day free trial, no card required. Pro is £12/month (or £120/year, equivalent to £10/month). Team is £15 per seat/month (or £144/seat/year), minimum 2 seats. Enterprise is custom. See the pricing page for details.

Frequently asked questions

Does this work with Cursor's agent mode? Yes — that's the intended use. Once the MCP server is registered in .cursor/mcp.json, agent mode can call IntentDocs tools to read stories and write progress back.

Can my team use Cursor and Claude Code against the same spec? Yes. The spec lives in IntentDocs, not in either tool — both connect over MCP to the same project, and progress from either is reflected in the same audit trail.

Do I have to keep the IntentDocs app open while Cursor builds? No. The MCP server handles the connection; the app is just where you view and reshape the story map and audit trail.

What does IntentDocs cost? There's a 7-day free trial (no card required), then Pro at £12/month or £120/year. Full details at intentdocs.com/pricing.

Keep reading

Ready to build what you intended?

Turn your idea into a structured spec and hand it to Claude Code or Cursor over MCP — with an audit trail of built versus planned.

Start free →

7-day free trial · no card required