design-critique-template
Skill13 starsChanged 5 months agoMITView on GitHub
Structured critique of a design against the hypothesis it was drawn for, plus heuristic evaluation — for step 9 and any mockup review. Use for "critique the design", "review mockup", "ревью макета", "что не так с UX", "heuristic evaluation", "оцени дизайн".
In this repository23 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- ambiguity-resolverSkill
- challengeSkill
- deep-thinkSkill
- experiment-designSkill
- ingestSkill
- interview-analysisSkill
- multi-source-signal-synthesiserSkill
- next-advisorSkill
- pipeline-stepsSkill
- post-launch-reviewSkill
- problem-structuringSkill
- product-requirements-docSkill
- setup-initiativeSkill
- solution-scoringSkill
- strategic-narrative-generatorSkill
- system-design-docSkill
- technical-spec-documentSkill
- tracking-and-funnelsSkill
- ui-pattern-librarySkill
- user-persona-builderSkill
- user-story-generatorSkill
- user-testingSkill
The file
We split this skill into the rules it gives an agent
2 rules. About 0 are common, 0 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 43
Update output/solution-sketch.md (## Changelog section) with accepted changes.
- Very specific: line 45
If the critique overturns a solution assumption — that's registry news: flag it to the PM (hypotheses.py set <id> --note ...), don't bury …
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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
Design Critique (hypothesis-first)
Critique for the /sketch → design loop (step 9). The order matters: a
beautiful screen that doesn't implement the hypothesis mechanism fails the
review regardless of heuristics.
1. Hypothesis fit — the first and decisive pass
Load the solution hypotheses this design serves (registry solutions links
output/solution-hypotheses.md):
- Does each screen implement the mechanism from the hypothesis formula ("if X then Y because Z" — is Z actually on the screen)?
- Did anything essential from
output/solution-sketch.mdget lost or mutated on the way to the mockup? Name the deltas. - Does the design add scope no hypothesis asked for? (scope creep enters through mockups more often than through PRDs)
2. Heuristic pass (classic, kept short)
- Visibility & hierarchy: is the primary action of each screen unmistakable in 5 seconds?
- User control: reversibility, escape routes, no dead ends
- Consistency: platform conventions and the product's own patterns (see ui-pattern-library — novelty only where novelty IS the hypothesis)
- Error prevention & recovery: empty states, loading, failure paths — the states mockups always forget
- Recognition over recall: labels over icons-only, context over memory
3. Feedback format
Per issue: screen → what breaks (hypothesis-fit or heuristic, name which) → severity (blocker / major / minor / polish) → concrete suggestion. Blockers first. Praise what works — reviewers who only list defects train designers to hide work.
4. Output
Update output/solution-sketch.md (## Changelog section) with accepted
changes; unresolved disagreements go to decisions.md as open questions.
If the critique overturns a solution assumption — that's registry news:
flag it to the PM (hypotheses.py set <id> --note ...), don't bury it in
design comments.
Anti-patterns
- Critiquing taste ("I'd make it blue") instead of hypothesis fit and heuristics — unfalsifiable feedback teaches nothing.
- A defect list with no severity — the designer can't sequence fixes.
- Reviewing the mockup without opening the solution hypothesis — you'll approve a beautiful screen that tests nothing.
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.

