Triage your product feedback from your AI coding tool
Hoot now ships an MCP server. Connect Claude Code, Cursor, or Codex to a project and let your agent read, search, triage, and ship feedback — without ever leaving your editor.
You already live in your AI coding tool. Now your product feedback can live there too.
Today we're shipping the Hoot MCP server. Point Claude Code, Cursor, or Codex at one of your projects and your agent can read, search, triage, and ship feedback — right alongside the code it's already writing. No tab-switching, no copy-paste, no context lost between "what users asked for" and "what you're building."
What your agent can do
Once connected, your agent has a small, focused toolbox scoped to a single project:
- See what matters — pull the top-voted feedback and this week's numbers before deciding what to build.
- Search before you file — find related cards or duplicates so the same request doesn't get logged twice.
- Triage — categorize cards into your existing categories and move them across the board (Pending → In progress → Done).
- Clean up reports — rewrite a vague "it's broken" into a clear, actionable card.
- Close the loop — mark something shipped and it lands on your public changelog.
It's the same triage you'd do by hand on the board — just driven by the agent that already has the full context of what you shipped.
One prompt, done
"Triage this week's feedback: surface the top-voted items, categorize anything that's uncategorized, and move the ones we just shipped to Done."
Your agent reads each card, picks the best-fitting existing category, and applies the changes — pausing to confirm before anything destructive.
Connect in one command
Open your project, head to the MCP panel, and generate a token. Then drop one line into your tool of choice:
claude mcp add hoot --transport http \
--header "Authorization: Bearer hoot_xxx" https://mcp.usehoot.app/mcp
Cursor and Codex get their own ready-to-paste config in the same panel — including an option to reference the token from an environment variable, so the secret never lands in a committed config file.
Your data stays yours
Each token is scoped to a single project — an agent only ever sees and acts on that project's feedback. Mint a separate token per tool or machine, check when each was last used, and revoke any of them in one click.
Available on Pro
The MCP server is part of Hoot Pro. Open a project → MCP → generate a token, and your roadmap is a claude mcp add away.
Ship the thing people asked for, then tell them it shipped — without leaving your editor. That's the whole loop, and now it runs where you already work. 🦉