Loading page...Загружаем страницу...正在加载页面...
AtlasRepo

The screen looked right. The file was wrong.

Description and use

An AI agent built a Figma screen that looked correct but had hidden design system violations and detached components. Audits detected all planted defects, yet post‑fix checks revealed remaining focus and flow issues needing designer input.

Problem

An AI agent built a Figma screen that visually appeared correct but contained hidden design system violations and component detachment defects that were not detected by the agent’s initial write operations.

Outcome

The audit skills successfully detected all planted defects but revealed additional design system violations and flow gaps that the fix pass did not resolve, demonstrating the need for designer intervention.

Workflow

  • Agent 1 builds Team members screen and plants 10 defects.
  • Agent 2 runs figma-slop-check and figma-handoff-gate to audit the file.
  • Agent 3 compares audit results with the answer key and confirms all defects are found.
  • Agent 4 runs figma-preflight, applies a fix pass, and re‑runs the audit skills.
  • Audit shows reduced slop findings but new handoff blockers remain.

Evidence

The title uses #111827. The file has a color variable, color/text/primary, with the same value. Another text layer is bound to it. The title is not. On the canvas, you can't tell the difference.

Source

The third row, Sofia Martins, matches the rows around it. But it is no longer an instance of the ListItem component. It is a plain frame that kept the component's name.

Source

All 10 planted defects were found, with 0 partial matches and 0 missed.

Source

The slop check went from 16 findings to 6, with 1 critical finding: no focus state on Button or ListItem.

Source

The handoff gate went from 11 issues to 13, with 8 blockers, mostly flows and states the fix pass had not drawn.

Source

  • Record date: 2026-10-07T21:41:17.952Z