<< All versions

Skill v1.0.1

currentAutomated scan100/100
phnx-labs/.agents/critique

+5 new

──Details
PublishedOctober 21, 2026 at 11:28 PM
Content Hashsha256:cb47a14f1b19d5ed...
Git SHA515768dd1c64
Bump Typepatch
Compare with v1.0.0
──Files
Files (1 file, 3.0 KB)
SKILL.md3.0 KBactive
SKILL.md · 47 lines · 3.0 KB

version: "1.0.1" name: critique description: "Assess existing screens, graphics, design systems, and interactions against their own guidelines. Inspect evidence and return ranked findings with concrete fixes." argument-hint: "[URL, app, file, screenshots, recording, or pages to compare]" user-invocable: true


Design critique

Assess the supplied surface without silently expanding a review into implementation. Resolve the design skill through the current skill catalog and read design-core.md beside its SKILL.md. The checker scripts belong to this critique skill's scripts/ directory; resolve their paths from this skill's actual installed location.

Workflow

  1. Understand and observe. Read the applicable design guidelines and open their

visual examples. Inspect the target itself and representative peers for consistency. For interactive surfaces, follow the shared interaction preflight. Review supplied recordings; capture and inspect a short clip when sequence or timing matters. State what cannot be assessed from a screenshot or an unavailable surface.

  1. Check measurable properties. Run `bun <critique-skill-dir>/scripts/check-contrast.ts

--json '<actual foreground/background pairs>' on real tokens or computed colors. Each pair has fg, bg, and label fields; add large: true only for large text. Run bun <critique-skill-dir>/scripts/check-tells.ts <available-markup-file>` when markup is available. Use the browser's authenticated page when inspecting a live app; raw unauthenticated HTML may be a different surface. Treat style warnings as prompts to investigate, not automatic failures or permission to override the product's brand.

  1. Assess the user experience. Check hierarchy, density, readability, alignment,

asset consistency, component reuse, and copy against the task. For interactions, assess discoverability, keyboard/focus, feedback, state continuity, recovery, and relevant responsive/motion behavior. State continuity includes the hover or selected highlight in a row of sibling controls: it slides from one item to the next rather than blinking off one and on the other. Compare pages or variants to identify drift.

  1. Report actionable findings. Prioritize blockers to task completion and accessibility,

then inconsistency and polish. Each finding names the page/element, observed behavior, evidence (capture, timestamp/action trace, computed result, or source line), impact, and a concrete fix. Separate observed facts from proposed changes and unverified claims.

  1. Route the next step. Summarize the smallest useful correction. Use design:system

for recurring rules, design:graphics for assets, or design:prototype for screens and flows when changes are authorized. Do not impose new standing design laws during a read-only review. Update existing guidance when that work is in scope.

Keep a small review in the response. For a substantial report, use the artifacts skill and inspect its rendering. Share captures only within the user's authorized scope.

← v1.0.0All versions