Dark Patterns Audit: AI Skill for UX Design
Dark Patterns Audit is an AI skill that sweeps any UI or flow for deceptive design. It names each pattern, classifies intent as Dark, Grey, or Mistake, and gives an honest-design fix.
npx skills add Uxcel-Lab/product-skills --skill ux-dark-patterns-audit
What this skill does
Sweeps any UI, flow, or copy for deceptive design and pairs every finding with an honest fix.
- Checks all eight pattern categories, from sneaking and obstruction to attention exploitation.
- Names each pattern it finds: hidden costs, roach motel, confirmshaming, forced continuity, and more.
- Classifies intent as Dark, Grey, or honest Mistake, so findings stay fair.
- Applies the Newspaper Test and the intent-gap question to catch subtle cases.
- Rates severity from Critical deception to Grey caution.
- Gives an honest-design fix for every finding, not just "remove it."
- Ends with a trust verdict and the prioritized fixes.
- Runs standalone or as a validation step on conversion, sign-up, and cancellation flows.
When to use it
Use it when you need an AI assistant to:
- audit a checkout, paywall, or pricing page for dark patterns;
- review a cancellation flow for roach-motel design;
- check sign-up defaults, pre-checked boxes, and renewal terms;
- catch confirmshaming, fake urgency, and manufactured scarcity;
- test whether a design would pass the Newspaper Test;
- validate AI-generated conversion flows before they ship.
What this audit checks
| Category | What it catches |
|---|---|
| Sneaking | Hidden costs, items snuck into the basket, forced continuity, bait and switch |
| Obstruction | Roach-motel cancellation, designs that block price comparison |
| Misdirection & tricks | Trick questions, disguised ads, preselected options, bad defaults |
| Nagging | Repeated prompts for something the user already declined |
| Forced action & privacy | Unrelated permission gates, friend spam, data reused beyond its stated purpose |
| Emotional manipulation | Confirmshaming, fake urgency and scarcity, scaremongering |
| Attention exploitation | Infinite scroll, autoplay, streak anxiety, notification overload |
| Microcopy tricks | Guilt-tripping copy, double negatives, vague OK and Cancel on consequential choices |
How the skill works
1
Sweep the catalog
The skill checks all eight pattern categories, from hidden costs and obstruction to attention exploitation.
2
Classify intent
It applies the intent-gap question and Newspaper Test, sorting each finding into Dark, Grey, or honest Mistake.
3
Deliver a trust verdict
It rates severity, gives an honest-design fix for every finding, and states the overall trust verdict.
Try it with
Related skills
1.0.0
MIT
UX Design
Audits
