Feedback → shipped features, on autopilot

Feedback boards that
write your next sprint.

Collect ideas and bug reports from a site widget or a hosted board, let votes rank what matters, then hand the top task straight to an AI coding agent over MCP.

How it works

From a stray idea to a merged fix.

A crow perched on a branch watching a flock of smaller sparrows below, representing feedback captured wherever users already are
01

Capture feedback where users already are.

Drop the toggle widget into your own site or point people at a hosted board. No account required to submit or vote, so the signal stays honest.

Two boxes drawn on the floor, sparrows crowded thickly into one and barely present in the other, with a crow observing, representing votes ranking the backlog
02

Let votes rank the kanban for you.

Every idea lands on a real board: Backlog, In Progress, Done. Vote counts and status moves are attributed, so priority reflects actual demand, not the loudest voice in standup.

A crow watching a small gear mechanism spin on its own, representing an autonomous coding agent
03

Ship it with an agent, not a standup.

An MCP server exposes your top-voted tasks to coding agents. They pick a task, open the implementation, and mark it complete, closing the loop without a human triage meeting.

See it in action

A real board, not a spreadsheet.

Every idea lands here, ranked by votes, moved by your team (or an agent). Try the vote arrows.

Backlog 2
24
Dark mode support
A lot of users asked for this in the widget too
Feature
9
Slack notifications
Ping a channel when a post hits Done
Doing 2
41
Bulk CSV export
Picked up by an agent over MCP
claude-code
15
SSO for teams
Blocking one enterprise customer
jordan
Done 1
33
Public roadmap page
Shipped and linked from the footer
Shipped
Features

Everything a feedback loop needs. Nothing it doesn't.

On-site toggle widget

One script tag drops a feedback panel into any product, no separate app for users to find.

Anonymous, attributed voting

Guests vote without an account; logged-in votes stay attributed for accurate ranking.

Multi-tenant boards

Run one board per product or team, each with its own roles and members.

MCP task server

Coding agents query top tasks and mark them complete through a first-class MCP interface.

Every change has a name on it

Each post keeps a timeline of status moves, reassignments, and tag edits, attributed to the person or agent who made them. Nothing shifts on your board without a trail.

Built for loop engineering

Your backlog, queryable by machines.

Cawcus's MCP server hands agents a ranked queue instead of a triage meeting: fetch the highest-voted task in the backlog, implement it, report back. The board updates itself.

  • Top-voted task first, always
  • Status flips the moment an agent completes work
  • Internal notes stay hidden from public voters
mcp: cawcus
# agent requests the next task
tools/call list_posts { board: "growth", status: "backlog", sort_by: "score" }
→ #482 "Export votes as CSV" · 37 votes · backlog

# agent ships the fix, then reports back
tools/call update_post_status { board: "growth", post_id: 482, status: "done" }
→ status: done
Pricing

Simple plans. Scale when you do.

Free

$0

Try it on one board

  • 1 board (owner-only)
  • 10 posts/month
  • Guest voting only (no comments)
  • No MCP integration
Get started
Most popular

Starter

$6/mo

One board, fully wired up

  • 1 board (owner-only)
  • 200 posts/month
  • MCP integration included
  • Guest voting + commenting
Start 3-day trial

Pro

$19/mo

For teams running several boards

  • Up to 3 boards
  • 2,000 posts/month per board
  • Team roles & access
  • Customizable board columns
  • AI generation features
  • Guest voting + commenting
Start 3-day trial
FAQ

Questions, answered.

Do users need an account to submit feedback? +

No. Guests can submit posts, vote, and leave public comments without registering. Voting is still one-per-person, tracked per session.

How do AI agents actually pick up a task? +

Cawcus exposes an MCP server. Any MCP-compatible agent can call board.next_task to fetch the highest-voted open item and board.mark_complete once it ships.

Can I run more than one board? +

Yes. Boards are fully multi-tenant, each product or team gets its own board, roles, and members.

Can I see who changed what? +

Yes. Every post has an activity timeline recording status changes, assignee changes, tag edits, and comments, each tagged with who did it and when. Changes made by an AI agent are marked as such, so you can always tell human work from automated work.

Can I move between plans later? +

Yes. Upgrade or downgrade between Free, Starter, and Pro at any time. Your boards, votes, and history carry over.

Stop triaging feedback by hand.

Set up a board, wire up an agent, and let the queue run itself.

Start a board free