agentsclimarketplace

Figma pixel verify

Skill FuDesign2008/open-skills/skills/figma-pixel-verify

Measure whether a running UI matches a Figma design-spec table (computed styles / box metrics + optional screenshots). Use when checking pixel alignment / 对照 Figma 验收 after implement. Bounded loop; honest residuals if no JS-eval channel. Triggers — 「检查像素对齐」「对照 Figma 验收」「设计保真检查」「是否对齐稿面」「figma 保真检查」 / verify Figma fidelity, pixel check. Do NOT use as the primary implement workflow (use figma-pixel-implement) or for code→Figma.From its SKILL.md

Install
npx -y skills add FuDesign2008/open-skills --skill figma-pixel-verify

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 6 stars6 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

4.1 KB, as published. Nobody here has run it

Figma Pixel Verify

Measure a running UI against a Figma design-spec table and report pass/fail with evidence. Optionally suggest bounded fixes; do not pretend measurement succeeded without a real measurement channel.

When to use

  • After figma-pixel-implement (or any Figma implement pass that left a spec table)
  • User asks to check pixel alignment / design fidelity / 对照稿面验收
  • Standalone: user provides Figma URL + running preview URL/route; extract or rebuild a minimal spec table first

Inputs

  1. Design-spec table — From implement artifacts, or rebuild via Figma metadata/variables (same columns as implement reference).
  2. Runnable UI — Local/dev URL, Storybook, or in-editor preview the Agent can observe.
  3. Optional: Figma screenshot for side-by-side vision (supporting evidence, not a substitute for numeric checks when JS-eval is available).

Workflow (ordered)

  1. Preflight — Confirm preview is reachable; list which rows will be measured. If no spec table exists, extract a minimal one from Figma before claiming verify.
  2. Vision (optional) — Capture UI vs Figma screenshot; note gross mismatches (wrong component, missing block, theme drift).
  3. Numeric measurement — For each critical row, read computed style and/or bounding box from the live DOM (or equivalent runtime). Map selectors carefully; prefer stable test ids when present.
  4. Compare — Apply tolerances in reference.md. Assign a verdict per row.
  5. Bounded fix loop — On actionable DRIFT, fix code → re-measure. Cap at about 3 meaningful iterations, then stop with remaining residuals listed.
  6. Report — Use the template in reference.md. Overall status is not PASS if any critical HARDCODED/DRIFT/MISSING-style remains unless the user explicitly accepts residuals.

Verdict taxonomy

VerdictMeaning
PASSWithin tolerance of expected
DRIFTMeasurable but outside tolerance
HARDCODEDValue matches by luck or literal, not via project token/variable when the spec required a token — flag when token mapping was part of the contract
VARIANTMatches a different documented variant (e.g. hover/dark) than the requested one
MISSING-styleProperty not measurable (no node, no style, or channel unavailable)

Tool intent (platform-agnostic)

IntentAgent may use
Open/preview UIBrowser MCP, simple browser, device preview
Read computed styles / boxesJS evaluate in page, DevTools-like APIs, test harnesses
ScreenshotsBrowser or Figma screenshot tools
Apply fixesNative edit tools, then re-measure

If no JS-eval / computed-style channel exists: run vision-only comparison, label numeric rows MISSING-style or residual, and do not claim pixel PASS.

Relationship

SkillBoundary
figma-pixel-implementProduces code + spec table; does not own final verdict
Agent-native Figma→code guidance / MCP fetchNot a substitute for measurement
Host PDCA workflowsOptional invoke — not a default strong dependency

Pitfalls

  • Declaring PASS from screenshots alone when measurement was possible.
  • Infinite tweak loops without re-measure.
  • Treating ±1 CSS px geometry noise on high-DPR as failure without noting DPR.
  • Silent skip of mask/currentColor issues — if implement violated asset rules, call that out as root cause of color DRIFT.

Keep looking

Skills are one crate of 627,801. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.