Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Quick web scanning for landscape understanding. Import of web-browsing/web-search skill. Snippets only — no conclusions from snippets alone.
.claude/skills/yogsoth-ai-creative-ideation-web-search/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-09 | ✗→✓ | ▲ Improved | 17% | 0% |
| case-02 | ✓→✗ | ▼ Worse | -82% | 0% |
| case-03 | ✓→✗ | ▼ Worse | -79% | 0% |
| case-04 | ✓→✗ | ▼ Worse | -92% | 0% |
| case-08 | ✓→✗ | ▼ Worse | -79% | 0% |
Quick web scanning for landscape understanding.
Import — strictly follow web-browsing/web-search skill protocol.
Snippets only — do not draw conclusions from search result snippets alone. Snippets provide leads for deeper investigation (web-research or paper-search).
Quantity target is set by the calling strategy's budget table. This SOP executes one unit = one brave_web_search call (count=10 results).
web-browsing repo → skills/web-search/SKILL.md
<!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use | | --- | --- | | web-search | Quick web scanning — discover pages, get snippets, find URLs. For orientation only, not substantive analysis. |
<!-- END available-tables (generated) -->
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 5,974 | 5,990 | +0% | 1 | 1 | 0% | 251 | 554 | +121% | 0 | 0 | — |
case-06 | fail→fail | 12,064 | 5,261 | -56% | 1 | 1 | 0% | 1,852 | 439 | -76% | 0 | 0 | — |
case-02 | pass→fail | 14,607 | 5,088 | -65% | 1 | 1 | 0% | 2,381 | 434 | -82% | 0 | 0 | — |
case-03 | pass→fail | 34,822 | 17,257 | -50% | 1 | 1 | 0% | 6,173 | 1,273 | -79% | 0 | 0 | — |
case-04 | pass→fail | 32,618 | 5,198 | -84% | 1 | 1 | 0% | 6,172 | 466 | -92% | 0 | 0 | — |
case-05 | fail→fail | 11,916 | 8,827 | -26% | 1 | 1 | 0% | 2,053 | 667 | -68% | 0 | 0 | — |
case-07 | fail→fail | 18,602 | 6,758 | -64% | 1 | 1 | 0% | 542 | 542 | 0% | 0 | 0 | — |
case-08 | pass→fail | 12,408 | 5,068 | -59% | 1 | 1 | 0% | 2,126 | 448 | -79% | 0 | 0 | — |
case-09 | fail→pass | 8,421 | 11,450 | +36% | 1 | 1 | 0% | 1,346 | 1,579 | +17% | 0 | 0 | — |
case-10 | fail→fail | 4,757 | 5,057 | +6% | 1 | 1 | 0% | 196 | 508 | +159% | 0 | 0 | — |
case-11 | fail→fail | 15,135 | 4,528 | -70% | 1 | 1 | 0% | 2,311 | 393 | -83% | 0 | 0 | — |
case-12 | fail→fail | 17,733 | 4,362 | -75% | 1 | 1 | 0% | 198 | 422 | +113% | 0 | 0 | — |
case-13 | fail→fail | 15,248 | 4,753 | -69% | 1 | 1 | 0% | 2,874 | 374 | -87% | 0 | 0 | — |
case-14 | fail→fail | 15,610 | 3,797 | -76% | 1 | 1 | 0% | 2,798 | 319 | -89% | 0 | 0 | — |
case-15 | fail→fail | 4,844 | 5,016 | +4% | 1 | 1 | 0% | 233 | 402 | +73% | 0 | 0 | — |
case-16 | fail→fail | 11,327 | 5,133 | -55% | 1 | 1 | 0% | 1,627 | 375 | -77% | 0 | 0 | — |
case-17 | fail→fail | 8,741 | 4,680 | -46% | 1 | 1 | 0% | 1,353 | 428 | -68% | 0 | 0 | — |
case-18 | pass→fail | 10,538 | 4,206 | -60% | 1 | 1 | 0% | 1,813 | 308 | -83% | 0 | 0 | — |
case-19 | fail→fail | 16,348 | 4,914 | -70% | 1 | 1 | 0% | 1,990 | 374 | -81% | 0 | 0 | — |
case-20 | fail→fail | 13,592 | 4,065 | -70% | 1 | 1 | 0% | 2,474 | 344 | -86% | 0 | 0 | — |
case-21 | fail→fail | 11,380 | 4,564 | -60% | 1 | 1 | 0% | 1,917 | 407 | -79% | 0 | 0 | — |
case-22 | fail→fail | 23,711 | 4,980 | -79% | 1 | 1 | 0% | 260 | 456 | +75% | 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 1 counted toward the lift figure. The other 21 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 -18 percentage points is the difference between those two pass rates over the 1 comparable cases. 5 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.