Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Reduction pass — cuts content, structure, visuals, and dead code that doesn't answer a user question or drive an action, respecting CRAFT_LEVEL. Use when the UI feels cluttered, has too many CTAs, walls of text, or decorative noise, or when the user says "simplify this" / "it feels too busy". Invoke when the user asks for distill on their UI, or mentions 'distill' alongside design / UI / frontend work.
.claude/skills/educlopez-distill/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | 40% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 3% | 0% |
| case-21 | ✗→✓ | ▲ Improved | -57% | 0% |
| case-03 | ✓→✗ | ▼ Worse | -89% | 0% |
| case-05 | ✓→✗ | ▼ Worse | -76% | 0% |
<!-- HARNESS MIRROR — do not edit here. Canonical source: skills/ or commands/. After editing source, copy into cli/assets/<harness>/ and repo-root harness mirrors. -->
Context: this sub-skill is one lens of the broader ui-craft skill. If the ui-craft skill is also installed, read its SKILL.md first for Discovery + Anti-Slop + Craft Test, then apply the specific lens below.
Distill the UI at $ARGUMENTS to its essence. Load the ui-craft skill.
The test for every section, component, prop, and word:
> Does this answer a question the user is asking or drive an action they need to take? If not — cut it.
Knob gating (CRAFT_LEVEL):
≤ 4 → be more aggressive with cuts; strip anything that doesn't directly drive action.5-7 → standard cuts per the test below.8+ → preserve signature details even if they don't directly drive action — the one memorable element that makes this feel designed stays.Apply ruthlessly:
references/modern-css.md) or 1px borders — never both. If every section is full-bleed and high-contrast, let some breathe.console.log, TODO notes without tickets — gone.Overlapping commands: for color overload (too many accents, decorative color), use /colorize. For motion noise, use /animate with MOTION_INTENSITY ≤ 3.
Output:
| Cut | Why | | --- | --- |
Guardrail: do not cut accessibility affordances (labels, focus styles, error text, reduced-motion fallbacks) — those always earn their space.
Next step: /polish — what survived the cut deserves the details (rung 1).
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 5,595 | 4,080 | -27% | 1 | 1 | 0% | 374 | 898 | +140% | 0 | 0 | — |
case-02 | fail→fail | 4,483 | 6,048 | +35% | 1 | 1 | 0% | 332 | 959 | +189% | 0 | 0 | — |
case-03 | pass→fail | 37,111 | 4,871 | -87% | 1 | 1 | 0% | 8,255 | 874 | -89% | 0 | 0 | — |
case-04 | fail→fail | 4,889 | 7,654 | +57% | 1 | 1 | 0% | 323 | 1,261 | +290% | 0 | 0 | — |
case-05 | pass→fail | 19,046 | 5,460 | -71% | 1 | 1 | 0% | 3,964 | 969 | -76% | 0 | 0 | — |
case-06 | pass→fail | 34,636 | 5,424 | -84% | 1 | 1 | 0% | 6,778 | 989 | -85% | 0 | 0 | — |
case-07 | fail→pass | 10,969 | 10,736 | -2% | 1 | 1 | 0% | 1,994 | 2,799 | +40% | 0 | 0 | — |
case-08 | fail→pass | 12,451 | 9,341 | -25% | 1 | 1 | 0% | 2,417 | 2,488 | +3% | 0 | 0 | — |
case-09 | fail→fail | 14,805 | 4,333 | -71% | 1 | 1 | 0% | 3,117 | 885 | -72% | 0 | 0 | — |
case-10 | fail→fail | 12,399 | 6,020 | -51% | 1 | 1 | 0% | 2,292 | 1,170 | -49% | 0 | 0 | — |
case-11 | fail→fail | 13,082 | 4,290 | -67% | 1 | 1 | 0% | 2,559 | 893 | -65% | 0 | 0 | — |
case-12 | pass→fail | 12,689 | 5,643 | -56% | 1 | 1 | 0% | 2,316 | 857 | -63% | 0 | 0 | — |
case-13 | pass→fail | 21,337 | 4,702 | -78% | 1 | 1 | 0% | 4,118 | 832 | -80% | 0 | 0 | — |
case-14 | pass→fail | 17,213 | 5,669 | -67% | 1 | 1 | 0% | 2,826 | 1,093 | -61% | 0 | 0 | — |
case-15 | fail→fail | 5,054 | 5,665 | +12% | 1 | 1 | 0% | 157 | 1,162 | +640% | 0 | 0 | — |
case-16 | fail→fail | 17,510 | 5,832 | -67% | 1 | 1 | 0% | 3,852 | 870 | -77% | 0 | 0 | — |
case-17 | fail→fail | 11,760 | 6,135 | -48% | 1 | 1 | 0% | 2,414 | 982 | -59% | 0 | 0 | — |
case-18 | pass→fail | 12,220 | 5,776 | -53% | 1 | 1 | 0% | 2,067 | 1,138 | -45% | 0 | 0 | — |
case-19 | fail→fail | 14,620 | 4,995 | -66% | 1 | 1 | 0% | 2,796 | 975 | -65% | 0 | 0 | — |
case-20 | fail→fail | 9,375 | 5,495 | -41% | 1 | 1 | 0% | 1,708 | 872 | -49% | 0 | 0 | — |
case-21 | fail→pass | 13,757 | 2,318 | -83% | 1 | 1 | 0% | 2,285 | 978 | -57% | 0 | 0 | — |
case-22 | pass→fail | 13,167 | 6,825 | -48% | 1 | 1 | 0% | 2,610 | 881 | -66% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 3 counted toward the lift figure. The other 19 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of -23 percentage points is the difference between those two pass rates over the 3 comparable cases. 11 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.