taste-audit
Skill4 starsChanged 16 days agoMITView on GitHub
Reviews an existing UI, screenshot, URL or codebase for AI slop, UX debt, accessibility failures and design-system drift, then returns a scored, prioritised fix plan and (on request) applies it. Use for "review my UI", "why does this look AI-generated", "audit", "make it look professional", "roast my landing page", "is this accessible", or before a launch.
1 other project uses a version of this.
Adopted by
Most recently taken up by
- obakeng-develops/tastetheir own version, first seen Jun 29, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- obakeng-develops/taste★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository12 other files it gives its agents.
- dist/AGENTS.mdAGENTS.md
- dist/copilot-instructions.mdCopilot instructions
- taste-bannersSkill
- taste-brandSkill
- taste-briefSkill
- taste-componentsSkill
- taste-directionSkill
- taste-guardSkill
- taste-identitySkill
- tastemakerSkill
- taste-slidesSkill
- taste-tokensSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 3 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 42
Keep the summary to 12 lines.
- Distinctive: line 47
Do not restyle what works.
- Distinctive: line 49
Never invent analytics or user research to justify a finding.
≈ “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.
Files
Taste Audit: diagnose, score, fix
Find what makes the interface feel generated or hard to use, rank it by user impact, and fix it with the smallest diffs.
Inputs (any)
Code (repo or files) · screenshot(s) · live URL (use a browser tool if available) · description. With code, run the scanner first; with images, apply the catalog by eye.
Procedure
- Context (3 lines): who uses it, the critical flow, platform. Infer; do not interrogate.
- Automated pass (code only):
node <taste-guard>/scripts/slop-scan.mjs <src> --project --json. - Heuristic pass, by lens, using catalog IDs where they exist:
- Slop: V/C-series visuals and copy, including Review-only items (V15, V17, C07).
- Hierarchy: one focal point per view? primary action obvious within 3 seconds? squint test.
- Consistency: radius, spacing, type scale, color roles, icon set; count distinct values.
- States: loading, empty, error, permission on each async surface (U10).
- Accessibility: contrast (measure, don't guess), keyboard path, names, focus visibility, reflow at 320px, motion.
- Copy: specific, verb-led, consistent voice, no hype.
- Responsiveness and performance: layout shift, image weight, font loading, mobile tap targets.
- Score: Taste score from the scanner (or estimate from findings) + BASIC (0-2 each for Beauty, Accessibility, Simplicity, Intuitiveness, Consistency).
- Prioritise: P0 blocks use or excludes users (a11y errors, broken states) · P1 erodes trust (slop, inconsistency) · P2 polish.
- Fix plan: for each finding, the exact change (file:line, class or token to use). Group by component so fixes are batched.
- Apply (when asked or when the user said "fix"): P0 then P1, smallest diffs, re-scan, report before/after score.
Output
Tastemaker audit: <product/page>
Score <before>/100 (<grade>) · BASIC <n>/10 · <e> errors · <w> warnings
Top 3 problems (user impact first):
1. <finding> [ID] -> <fix>
...
P0 | P1 | P2 tables: ID · where · problem · fix
After fixes (if applied): score <after>/100, remaining items with reasonsKeep the summary to 12 lines; tables only as long as the findings.
Rules
- Measure contrast with
taste-tokens/scripts/contrast.mjs; never eyeball a borderline pair. - Do not restyle what works. A good layout with slop copy gets copy fixes, not a redesign.
- Screenshot audits: state what cannot be verified from pixels (keyboard, screen-reader names, states not shown).
- Never invent analytics or user research to justify a finding; reason from heuristics and say so.
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
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.