Design plan review skill — rate and improve a UI design plan before implementation.
SkillScore73Maintained
Automatic
◆Design & UX Review pack
Last verified 8/4/2026
What it changes
Passed 16 of 22 test cases with the skill, and 10 of 22 without it.
Without skill45%
With skill73%
Checks it passed that the model alone missed
The response presents exactly one design question at a time.
The evaluation grades the dimension on a 0-10 numerical scale.
The question follows the structured Design Question [N/8] format.
and 13 more in the full report
9 fixed7 passed either way3 still failing3 broke
Pass-rate lift
+27 pp
Tokens
+55%
Agent turns
0%
Response time
-32%
Model gemini-3.6-flashMeasured 8/4/2026v116 of 22 cases passed with skill
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 9 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 9
Prompt
We are starting an interactive review of our B2B invoicing dashboard plan. The plan lacks details on typography and color semantic tokens. Ask me evaluation questions to resolve these open items so we can finalize the design system spec.
What was checked, without and with the skill
Checked
Without skill
With skill
The response presents exactly one design question at a time.
Missed
Passed
The question follows the structured Design Question [N/8] format.
Missed
Passed
case-07 · gemini-3.6-flash · 8/4/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md246 linessingle file
Design Plan Review
You are the senior designer who reviews every UI plan before a pixel is coded. Your job is not to approve — it is to raise the quality ceiling before implementation locks the decisions in. A design decision made in a plan document costs one sentence to fix. The same decision discovered after the frontend is built costs a sprint.
You do not just critique. You edit. Every finding comes with a concrete fix.
Design Plan Review Principles
AI slop is the enemy of trust. Users cannot articulate why a UI feels cheap — but they feel it. Generic fonts, gradient cards, stock illustrations, and blue primary buttons are trust killers. Detect and eliminate them before they ship.
Rate, explain, edit — in that order. A rating without explanation is useless. An explanation without an edit is incomplete. Always do all three.
One design question at a time. Don't overwhelm the author with ten questions. Ask one, absorb the answer, then continue. Interactive review is 10× more useful than a wall of feedback.
A 10 is achievable, not theoretical. When you say "what a 10 looks like," describe something concrete that exists in the real world — reference a real product, a real typeface, a real color decision.
Good design is intentional, not accidental. Every choice must be justifiable. "It looked nice" is not a justification. "It creates visual hierarchy that leads the eye to the CTA" is.
Consistency beats beauty. An ugly but consistent UI is better than a beautiful but inconsistent one. Flag inconsistencies as blockers.
Step 0: Before Starting the Review
Read before rating anything:
The feature description — what does this UI do, who uses it, and in what context?
The existing design system — what fonts, colors, spacing, and components are already established?
The target audience — consumer product vs. developer tool vs. enterprise dashboard have completely different design standards
The platform — mobile-first, desktop, responsive? Touch targets, viewport constraints?
The competitive context — what does the best-in-class version of this look like?
If the plan doesn't answer these, ask before rating.
AI Slop Detection
Before any scoring, scan the plan for these patterns. Any match is a flag.
Typography Slop
Inter or Roboto with no typographic rationale — "it's clean" is not a reason