Connect FeatQ to ChatGPT
ChatGPT can connect to FeatQ in two ways, depending on what your ChatGPT plan supports. MCP (Model Context Protocol) is an open standard that lets an AI assistant call tools on an external server, in this case your FeatQ board. ChatGPT supports MCP through connectors on some plans and modes, and support for remote MCP servers has been rolling out gradually. If connectors are not available to you, the widget route works everywhere.
Setup: connector if you have it, widget either way
Option A (MCP connector): if your ChatGPT plan shows a connectors or custom connector option in settings, add a remote MCP server with URL https://featq.com/api/mcp and an Authorization header of Bearer fq_YOUR_KEY. Availability and write-tool support vary by plan and mode, so check what your plan offers.
Option B (widget, works regardless of plan): if you build your product with any AI tool, paste this prompt to add the FeatQ feedback widget to your app.
Add this script tag just before the closing body tag of my app: <script src="https://featq.com/widget.js" data-board="YOUR_BOARD_SLUG" async></script>You can also paste board data into any ChatGPT conversation manually: copy the output of this call and ask ChatGPT to plan the work.
curl -H "Authorization: Bearer fq_YOUR_KEY" https://featq.com/api/v1/top-specWhat it looks like in practice
With a connector set up, a session looks like this:
You: Look at my FeatQ board and tell me what to prioritize. ChatGPT: [calls get_ranked_requests] Your top three by demand: "API access" (56 votes), "Mobile app" (43 votes), "SSO" (12 votes). API access also has 20 recent comments, so interest is current. You: Draft a spec for API access. ChatGPT: [calls generate_spec] Here is the generated spec: endpoints users asked for, auth model, and rate limits flagged in the comments. Paste this into your coding agent, or I can refine it here.
One quirk to know
ChatGPT connector support is the moving part here: which plans get custom MCP connectors, and whether write tools like update_request_status are allowed, has changed over time. Read tools are the safe assumption; verify the rest in your own account before relying on it.
Frequently asked questions
Practical answers about agents, voting, embeds, and pricing.
Custom MCP connectors are generally a paid-plan feature. But you can always paste board data into any conversation: curl the top-spec endpoint and drop the result into the chat.
Only if your plan and mode allow write actions through connectors. Read tools (listing and ranking requests, generating specs) are the reliable part today.
A one-line script tag that adds a floating feedback button to your app, so your users can submit and vote without leaving the product. Any AI builder can place it for you with the one-prompt install.
Yes. One server, one key per board, any MCP client. The tools and ranking behave identically everywhere.
Still have questions? Contact us