critique

Overview of critique skill

What critique is for

The critique skill helps you evaluate an interface, feature, page, or component from a UX lens instead of giving vague “looks good” feedback. It is aimed at people who need a structured critique skill for UI/UX Design: product designers, founders, PMs, QA-minded reviewers, and AI agents that must explain what is working, what is not, and why.

What it actually does

This critique skill centers on visual hierarchy, information architecture, emotional resonance, cognitive load, heuristic checks, and actionable recommendations. The practical job-to-be-done is not just judgment, but decision support: identify friction, score quality, and produce feedback that can guide the next design iteration.

When it fits best

Use critique when the ask is to review, evaluate, audit, or compare a design artifact and you want a repeatable method rather than a generic prompt. It is a strong fit when the design context is available and the review needs nuance across user goals, constraints, and interaction quality.

Main decision factors

The repo is geared toward deeper critique work, not casual commentary. Its differentiators are the context-gathering step, independent assessments, persona-based testing, and scoring-oriented evaluation. That makes critique useful when you care about reducing reviewer bias and turning observations into concrete output.

How to Use critique skill

Install the critique skill

The documented install path in the skill is via the skills manager: npx skills add pbakaus/impeccable --skill critique. That is the critique install step to use when you want the skill available in your agent workflow rather than copied into a one-off prompt.

Feed it the right input

The critique usage pattern expects an area to review, such as a feature, page, or component. The argument-hint is explicit: give a focused target, not a whole product blob. Strong input looks like: “Critique the checkout shipping step for first-time mobile users” or “Critique this dashboard filter panel for cognitive load and clarity.”

Start with the right files

For a fast critique guide pass, read SKILL.md first, then the support references that shape evaluation quality: reference/cognitive-load.md, reference/heuristics-scoring.md, and reference/personas.md. These files tell you how the skill scores, what it looks for, and which user perspectives to simulate.

Use a complete prompt shape

A better prompt gives the design goal, audience, platform, and what changed. For example: “Critique this onboarding modal for new enterprise admins. Goal: reduce abandonment. Context: desktop web, step 2 of 4, copy and layout were changed yesterday.” That gives the skill enough signal to avoid generic advice and focus on actual failure modes.

critique skill FAQ

Is critique only for UI/UX Design?

No. It is strongest for UI/UX Design, but it can also review any interface-like experience where hierarchy, clarity, and interaction quality matter. If there is no user-facing flow or layout to assess, the critique skill is probably the wrong tool.

How is critique different from a normal prompt?

A normal prompt often produces subjective comments. The critique skill adds a method: context gathering, independent assessments, persona testing, and scoring. That structure makes the output easier to trust, compare, and act on.

Do I need design expertise to use it?

No. The critique skill is useful for beginners because it gives a review frame. What matters most is supplying enough context: what the interface does, who it is for, and what success looks like. Without that, even a strong critique skill will drift toward generic feedback.

When should I not use critique?

Do not use it for purely technical debugging, branding-only reviews, or content editing with no interaction design element. It is also a poor fit if you want a quick subjective opinion with no scoring or supporting rationale.

How to Improve critique skill

Give the skill a sharper brief

The biggest quality gain comes from better context. Include the target surface, user type, device, and business goal. “Critique the pricing table” is weak; “Critique the pricing table for SaaS buyers comparing plans on mobile, with the goal of increasing qualified trial starts” is much better.

Provide evidence, not just opinions

If you already know what feels off, say where and why. Mention the flow step, the label, the state, or the screenshot area that seems problematic. That helps critique focus on real usability issues instead of re-litigating obvious surface changes.

Watch for common failure modes

The most common miss is under-specifying the artifact, which leads to broad commentary. Another is asking for “feedback” without naming the user task; critique then has to infer the intent. For critique for UI/UX Design, the review gets better when you include the primary action the user should complete and any constraints like accessibility, mobile behavior, or time pressure.

Iterate after the first pass

Use the first critique to separate high-risk problems from polish issues, then re-run the skill on the riskiest area only. A second pass works best when you ask for a narrower critique such as hierarchy, copy clarity, or cognitive load, instead of asking the skill to re-review everything.