deceptive-patterns
Deceptive patterns are design choices that use the product's power against its own users — exploiting the same psychological mechanisms that honest design employs, but in service of the company's goals at the user's expense. The work here is identification, classification, severity assessment, regulatory exposure, and the honest alternative. Every finding cites its pattern category and, where relevant, the mechanism being exploited.
Cross-reference: skills/design-for-ai/references/ai-tells.md is the structural twin — a ban-list of aesthetic tells (visual patterns that signal AI-generated defaults). This skill is the ethical twin — a ban-list of manipulation tells. Same format, different domain.
When this applies
- Auditing a flow for deceptive patterns: sign-up, checkout, subscription, cancellation, cookie consent, notification opt-in, pricing pages, free-trial-to-paid conversions.
- Reviewing a specific feature: a countdown timer, a social-proof block, a dark default, a friction-asymmetric cancel flow.
- Ethical design review: a product team wants to know whether their engagement mechanics cross the line from persuasion into manipulation.
- Regulatory exposure check: which patterns are explicitly named in EU DSA, FTC guidance, or the forthcoming EU Digital Fairness Act.
- Proposing the honest alternative: every dark pattern has a mechanism;
references/honest-alternatives.mddocuments what ethical design looks like for each category.
Not aesthetic tells (glassmorphism, Inter defaults, cyan-on-dark — those are ai-tells in the core skill), not legitimate persuasion mechanics (use behavioral), not accessibility audits (use usability).