agentleFS
Sign inStart free
BlackAngel242BlackAngel242/skill-market/skills/uxfix/SKILL.md

uxfix

Skill0 starsChanged 14 days agoMITView on GitHub

Fixe automatiquement les défauts UI détectés : boucle détecter → corriger → revérifier jusqu'à un score de 99/100. Contrastes WCAG, halos colorés, easings rebond, interlettrage, palettes IA, copy marketing.

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 1 is distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Don't just report UI defects: fix them alone, re-verify, and loop until the score reaches 99/100.

    line 10
  2. Very specific:

    Originals are safe: before the first write, everything is copied to .uxfix-backup/<timestamp>/ next to the target.

    line 27
  3. Distinctive:

    Run on the real built files, not on mockups: the detector resolves computed styles.

    line 53
  4. Very specific:

    Never run with --target 100 as a hard gate in CI unless a human reviews the diff: the last point is sometimes a judgment call.

    line 57

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

UxFix: autonomous UX fixing loop

Purpose

Don't just report UI defects: fix them alone, re-verify, and loop until the score reaches 99/100. The impeccable detector provides the analysis (findings + severity); UxFix applies the mechanical corrections and iterates. The agent supervises; the machine does the rounds.

Tooling

Engine: <skill-dir>/engine/uxfix.js (Node, no dependencies). Detector: the impeccable binary (installed globally by market install uxfix).

node <skill-dir>/engine/uxfix.js <fichier-ou-dossier> [--target 99] [--max-iterations 6] [--dry-run] [--json]

IMPECCABLE_BIN env overrides the detector path. --dry-run previews fixes without writing. --json emits a machine-readable report.

Workflow

  1. Run uxfix on the built UI files (single HTML/CSS file or directory). It scans .css files and <style> blocks in HTML.
  2. Read the report: score progression per iteration, fixes applied, manual remainder.
  3. Handle the manual remainder yourself: findings with no auto-fixer (unknown rules, brand palette choices the engine neutralized, copy the dictionary missed) need human judgment.
  4. Originals are safe: before the first write, everything is copied to .uxfix-backup/<timestamp>/ next to the target.

The 99% contract

  • Score = 100 − 1 per unique warning − 5 per unique error. Duplicates of the same finding count once. Advisories are informational and don't count (same as the detector's own failure semantics).
  • 99+ means zero primary findings, or a single trivial warning left.
  • The loop keeps fixing while the fixers make progress (max 6 iterations), then reports what's left instead of churning.

Auto-fix catalog

FindingAutomatic fix
low-contrastRecomputes the offending color with real WCAG math (binary search toward black/white) until the required ratio is met; touches only the color/background declarations involved
gradient-textReplaces the gradient with its darkest stop as a solid color; removes background-clip: text
dark-glowZero-offset chromatic box-shadow → neutral elevation shadow; chromatic text-shadow → none
wide-trackingletter-spacing above 0.05em → 0.02em (0.04em for uppercase labels)
cream-paletteFlagged warm page background → #ffffff
ai-color-paletteViolet/purple hues (265–300°) rotated toward blue, saturation/lightness kept
bounce-easingcubic-bezier with overshoot → cubic-bezier(0.4, 0, 0.2, 1)
pulsing-dotInfinite pulse animation on the flagged dot → animation: none
marketing-buzzwordDictionary rewrite of EN/FR buzzwords in HTML text nodes (scripts, styles, code blocks untouched)

Details and limits per rule: references/fix-catalog.md. Scoring details: references/scoring.md.

Operating rules

  • Run on the real built files, not on mockups: the detector resolves computed styles.
  • Palette changes (ai-color-palette, cream-palette) are applied with safe neutrals but must be flagged for human review: brand colors are a choice, not a bug. The report lists every one.
  • Copy rewrites use a fixed dictionary; nuanced marketing copy still deserves a human pass.
  • A 100/100 detector score is defect evidence, not proof of good design. Keep taste in the loop: run the loop, then look at the page.
  • Never run with --target 100 as a hard gate in CI unless a human reviews the diff: the last point is sometimes a judgment call.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free