Feature

Point,
don't describe

Reporters click the element that is wrong. The report carries its label, component and selector, and outlines in red the elements tied to an error or a failed request. Included on every plan.

  • The reporter clicks the element that is wrong, or several of them
  • The report carries the element's label, component and selector
  • Elements tied to an error or a failed request are outlined in red
  • No screenshot needed, so it works on every plan
  • Your agent reads the element over MCP and goes straight to the code
Start free
Element Picker: Point at the Problem Instead of Describing It

The hardest part of a bug report is saying where

"The button on the second tab doesn't work" costs a round trip every time. With the element picker the reporter clicks the thing that is wrong. The report records which element it was: its label, the component it belongs to and a selector that finds it again.

Red outlines show what is already broken

When an element is tied to an error or a failed request, the picker outlines it in red. A reporter does not have to know which button fired the 422: the page already does. The report keeps the pairing, so the element, the error and the request travel together.

One element, or several

Pick one element or more than one. A misaligned header and the card under it are one report, not two.

Why an agent needs this

A coding agent that is told "the checkout button is broken" searches the whole repo. One that reads the element, the component name and the request goes to the file. get_issue_context returns all three over MCP.

On every plan

The element picker needs no rendered screenshot, so it is on every plan, including Free. Screenshots and markup are on paid plans: see screenshot and markup and pricing.

For teams shipping with AI agents

Let your users find the bugs. Let your agent fix them.

Free to start, no card. One script tag, then your agent reads every report over MCP.

Add to Cursor