Files
Jeremy Dawes 608fdd4300 feat(dev-tools): rewrite ux-audit, add ux-extract and ux-compare
Three-skill UX trilogy: extract captures patterns from reference apps,
compare synthesises design decisions across extracts, audit measures
a build against the bar.

ux-audit rewrite
- Single exhaustive mode (drops quick/standard/thorough — unused ladders)
- Thread-first flow testing, with element inventory as completeness check
- Coverage arithmetic: tested / inventoried, published in the report
- Persona required (argument, .jez/personas/, or one-question prompt)
- Chrome MCP required for authenticated apps, fail loud if not connected
- Viewport pinned 1440x900 at start, >2000px noted as harness-breaking
- 8-scenario battery: First Contact, Interrupted Workflow, Wrong Turn
  Recovery, Returning User, Keyboard Only, Heavy Data, Destructive
  Confidence, Second User
- Sub-agent screenshot review per thread
- Fix-and-verify loop closes findings in the same session
- Optional cross-reference to ux-extract libraries

ux-extract (new)
- Inverse of audit — asks "what is the bar?"
- Exhaustively captures screenshots (every state), verbatim copy,
  interaction prose, keyboard patterns, responsive treatments, motion
  GIFs, and Notable Absences
- Dedup phase collapses repeated patterns
- Output: docs/ux-extracts/<app>/pattern-library.md + screenshots/
- Ethics section on ToS, robots.txt, rate-limiting, authorised access

ux-compare (new)
- No browser — pure markdown synthesis across 2+ pattern libraries
- Walks category by category: convergence, divergence, unique, absent
- Opinionated recommendations (safe default / consider / skip / fill gap)
- Cites source libraries instead of re-embedding screenshots
- Honest about stale libraries and coverage gaps

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-19 12:28:16 +10:00

697 B

description, argument-hint, allowed-tools
description argument-hint allowed-tools
Exhaustively audit a web app — threads, elements, 8 scenarios, fix-and-verify loop [optional: persona or scope, e.g. 'as a busy broker' or 'the dashboard only'] *

Load the ux-audit skill and run the audit.

Single exhaustive mode. If $ARGUMENTS describes a persona ("as a busy broker", "first-time user"), adopt it. If $ARGUMENTS scopes the audit ("the dashboard", "just the settings flow"), scope to that area — still exhaustive within the scope.

URL is auto-detected from wrangler.jsonc or running dev server.

Examples:

  • /ux-audit
  • /ux-audit as a time-poor client logging in for the first time
  • /ux-audit the billing flow