Connect FeatQ to Cursor

Cursor supports MCP natively. MCP (Model Context Protocol) is an open standard that lets an AI agent call tools on an external server, in this case your FeatQ board. Once connected, Cursor Agent can pull your top-voted requests and build from a generated spec without leaving the editor.

Setup: one JSON snippet

Open Cursor settings, find the MCP section, and add a new server with this config (or add it to your mcp.json). Replace fq_YOUR_KEY with the key from your board admin page under Agent access.

{
  "mcpServers": {
    "featq": {
      "url": "https://featq.com/api/mcp",
      "headers": {
        "Authorization": "Bearer fq_YOUR_KEY"
      }
    }
  }
}

Save and make sure the featq server shows as enabled in the MCP list. Then ask the agent about your feature requests.

What it looks like in practice

A typical session in Cursor Agent looks like this:

You: Check my FeatQ board. What is most requested?

Cursor: [calls get_ranked_requests]
  1. "Slack notifications" (41 votes, 9 comments)
  2. "Bulk edit" (25 votes)
  Ranked by votes plus comments, recent activity weighted higher.

You: Get me a spec for the Slack one and implement it.

Cursor: [calls generate_spec]
  Spec covers the webhook setting, message format, and failure
  handling mentioned in the comments. Starting with the settings UI...

Cursor: Implemented and passing lint. Want me to mark the request
  as in progress on the board?

One quirk to know

Cursor asks you to approve each tool call the first time by default. That is sensible for update_request_status, but for read-only tools like list_requests you can allow-list them to skip the prompts.

Frequently asked questions

Practical answers about agents, voting, embeds, and pricing.

Agent mode is where MCP tools run. Plain autocomplete and inline edits do not call tools.

Votes plus half-weight comments, with the score fading as a request ages. In practice: what many users want, weighted toward recent demand. No setup or scoring config needed.

No. The connection only pulls board data (requests, votes, comments, specs) into Cursor. FeatQ never sees your repository.

Yes. Add a second server entry with a different name (for example featq_beta) and that board’s key.

Still have questions? Contact us

Keep going

Other clients