Turn feedback intocode
changes

Record user paths. Surface issues. Generate code. One loader, no npm updates.

  • Free to start
  • No credit card
  • 2 min setup

Running on this page. Click the bubble in the corner to try it live.

Order summaryTotal$48.00Complete purchase

“This button should be blue.”

component
<CheckoutButton>
source
src/checkout/Button.tsx:42
change
change-style
- bg-gray-400+ bg-blue-600
PR opened
Sample: one click, one pull request
  • fix-bug
  • element selector
  • change-text
  • component
  • change-style
  • file:line
  • add-feature
  • screenshot
  • improve-ux
  • console
  • network
  • viewport

Everything a developer needs, attached before they ask.

element
CSS selector, React component name and the source file it lives in.
screenshot
The selected element, or a region the reviewer dragged out.
console
Recent errors and warnings from the page.
network
Recent API calls, failed requests first.
page
URL, viewport and user agent.
Three nested boxes in the CSS box-model colours: a blue content box inside a salmon padding ring inside a green margin ring, drawn with dimension lines on a measuring grid.
The get_feedback_context payload from the file:line demo: component Button, file src/Button.tsx:12:3, one console error, a failed GET /api/theme returning 500, a replay summary and a matched Playwright flow.

One report as an agent reads it. Watch the 60s file:line demo

A printed checkout page on a drafting table, its grey button circled in orange pen with the note 'should be blue?', a sticky note, a steel ruler and an orange marker beside it.

Three steps, no translation layer.

A vague stakeholder comment becomes a code change without anyone playing interpreter in the middle.

  1. Add the script tag

    Drop the SDK into your site or create a project and share a preview link. The hook maps every DOM element back to its React component and source file.

    1. loader<script> one tag, no npm updates
    2. buildsource metadata injected into JSX
    3. runtime hook__FEEDBACK_DEVTOOLS_HOOK__
    4. react fiber<CheckoutButton>
    5. sourcesrc/checkout/Button.tsx:42

  2. Collect feedback

    Stakeholders click any element — or drag to capture a region — and describe the change. Screenshot, console errors and network logs attach automatically.

    The live widget in picking mode on this page: a banner reads 'Click any element on this page to leave feedback' and an orange selection box outlines the words 'code changes'.
    The widget's Leave feedback dialog, attached to the selected element, with 'What should change?' and 'Expected behavior' filled in and a Send feedback button.

    The real widget, running on this page.

  3. Triage with AI

    AI scores sentiment, intent and urgency so angry users surface first. One click generates a Claude Code prompt with full source mapping.

    sentiment
    frustrated
    intent
    fix-bug
    urgency
    high
    Claude Code prompt
Fix src/checkout/Button.tsx:42 <CheckoutButton>
Request: “This button should be blue.” · change-style
Attached: screenshot, console, network, viewport

Built for the feedback-to-code loop

  • Source mapping

    Every comment pins the exact component name and file location — no more "which button did they mean?"

  • Shareable feedback links

    Send a link, collect annotated feedback from anyone. No account needed for reviewers.

  • Kanban pipeline

    Track every item from new to deployed. Export CSV or JSON whenever you need it.

  • GitHub Issues export

    Reporters never need a GitHub account — but you still get issues. Connect your repo and items auto-file as labeled GitHub issues with the screenshot and source file attached.

  • Slack alerts

    Route new feedback to a Slack channel through an incoming webhook — title, page, file:line and a link back to the item, with automatic retries. Notify-only: no threads, no two-way sync.

  • AI sentiment & urgencyPRO

    Pro triage classifies tone, intent and urgency so the queue sorts itself.

Free to start. AI triage when you need it.

Free
$0
Unlimited feedback collection.
Pro
$9/mo per project
AI sorts the angry users from the nice-to-haves.