ai-ui-heuristics
Skill4 starsChanged 3 months agoApache-2.0View on GitHub
Nielsen's 10 usability heuristics and Don Norman's 7 principles, reinterpreted for AI-native interfaces. Use when evaluating, auditing, or reviewing the UX of any AI-powered application. Each heuristic includes the AI-specific interpretation, a common failure mode, and a concrete fix. Triggers on: "heuristic evaluation", "Nielsen heuristics", "UX audit", "review this AI UI", "is this good design", "Don Norman principles".
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 1 is distinctive, and 0 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.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 40
Do not require users to remember across screens, decide more than once per screen, track >4–7 competing items, or learn a new interaction …
≈ “Common” is an estimate: at least 1 in 5 of the 308,232 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
AI UI Heuristics
Nielsen's 10 heuristics and Don Norman's principles, reinterpreted for AI-native interfaces. Keep this file as the audit entrypoint; depth lives in references/.
Sources: Nielsen Norman Group — Ten Usability Heuristics; Don Norman, The Design of Everyday Things.
Fast pass (ship blockers)
- Status — User knows AI is working (ambient > spinner-only).
- Control — Stop, undo/version, inline edit within 2 taps.
- Hero — Output owns ~80%+ of the canvas; one primary CTA.
- Recovery — Failures explain what / why / next step; keep partials.
- Fallback — Core job works when AI is unavailable.
Heuristic snapshot
| # | Heuristic | AI ask |
| H1 | Visibility of status | Processing, agent, confidence, ETA visible |
| H2 | Real-world match | Domain language; hide model knobs |
| H3 | User control | Stop, edit, undo, versions |
| H4 | Consistency | Predictable format or explained variation |
| H5 | Error prevention | Validate + cost + confirm high-risk |
| H6 | Recognition > recall | Actions + examples, not memorized prompts |
| H7 | Flexibility | Shortcuts, templates, batch for power users |
| H8 | Minimalist | Content hero; contextual chrome |
| H9 | Error recovery | Plain language + fix path |
| H10 | Help | In-context tips and examples |
Norman lens: discoverability, feedback, conceptual model, affordances, signifiers, mapping, constraints — see references/norman-principles-ai.md.
Cognitive load check
Do not require users to remember across screens, decide more than once per screen, track >4–7 competing items, or learn a new interaction model only for AI.
Reference Map
references/nielsen-10-ai.md— each heuristic with symptoms and fixesreferences/norman-principles-ai.md— Norman's principles applied to AI UIreferences/audit-checklist.md— printable Pass / Partial / Fail audit
Related Skills
- For calm technology test → use
calm-technology-ui - For common failure catalog → use
ai-ui-anti-patterns - For generative UI patterns → use
ai-native-ui-design - For multi-agent UI → use
agentic-ui-patterns
In this repository80 other files it gives its agents.
- a11y-accessibility-designSkill
- agentic-ui-patternsSkill
- ai-native-ui-designSkill
- ai-ui-anti-patternsSkill
- calm-technology-uiSkill
- conversion-growth-designSkill
- cross-platform-designSkill
- dark-mode-designSkill
- data-visualizationSkill
- design-systemsSkill
- design-thinking-processSkill
- error-empty-statesSkill
- ethical-design-patternsSkill
- form-uxSkill
- information-architectureSkill
- interaction-design-patternsSkill
- internationalization-uiSkill
- mobile-touch-designSkill
- modal-classes-lifecycleSkill
- modal-container-imagesSkill
- modal-dicts-queuesSkill
- modal-endpointsSkill
- modal-fundamentalsSkill
- modal-gpu-computeSkill
- modal-networkingSkill
- modal-observabilitySkill
- modal-parallel-executionSkill
- modal-sandboxesSkill
- modal-scaling-performanceSkill
- modal-schedulingSkill
- modal-secretsSkill
- modal-serversSkill
- modal-storageSkill
- modal-web-servingSkill
- modal-workspace-adminSkill
- motion-design-principlesSkill
- notifications-permissionsSkill
- onboarding-uxSkill
- rust-api-designSkill
- rust-async-programmingSkill
- rust-cargo-masterySkill
- rust-cli-applicationsSkill
- rust-clippy-lintsSkill
- rust-closures-iteratorsSkill
- rust-compiler-internalsSkill
- rust-concurrencySkill
- rust-design-patternsSkill
- rust-editionsSkill
- rust-embeddedSkill
- rust-error-handlingSkill
- rust-game-developmentSkill
- rust-lifetimes-borrowingSkill
- rust-macrosSkill
- rust-performanceSkill
- rust-rustdocSkill
- rust-rustup-toolchainSkill
- rust-smart-pointersSkill
- rust-testing-benchmarkingSkill
- rust-type-systemSkill
- rust-unsafe-ffiSkill
- The first 60 shown.
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.