Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Import SOP: quick web scan discovery (from web-browsing skill)
.claude/skills/yogsoth-ai-experiment-execution-web-search/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | -69% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -44% | 0% |
| case-19 | ✗→✓ | ▲ Improved | -51% | 0% |
| case-06 | ✓→✓ | ＝ Same ✓ | -72% | 0% |
| case-20 | ✓→✓ | ＝ Same ✓ | 25% | 0% |
Quick web scan discovery — imported from web-browsing skill.
Import — invokes web-browsing skill's web-search SOP.
Available to all campaigns for rapid information discovery (methodologies, tools, datasets, competing methods, etc.).
brave_web_search, brave_news_search<!-- 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-05 | fail→pass | 9,897 | 2,183 | -78% | 1 | 1 | 0% | 1,725 | 531 | -69% | 0 | 0 | — |
case-01 | fail→fail | 23,877 | 17,024 | -29% | 1 | 1 | 0% | 2,409 | 414 | -83% | 0 | 0 | — |
case-02 | fail→fail | 8,843 | 5,842 | -34% | 1 | 1 | 0% | 1,474 | 443 | -70% | 0 | 0 | — |
case-03 | fail→fail | 16,534 | 5,534 | -67% | 1 | 1 | 0% | 2,671 | 451 | -83% | 0 | 0 | — |
case-04 | fail→pass | 8,919 | 3,064 | -66% | 1 | 1 | 0% | 1,207 | 676 | -44% | 0 | 0 | — |
case-06 | pass→pass | 14,806 | 2,553 | -83% | 1 | 1 | 0% | 2,074 | 588 | -72% | 0 | 0 | — |
case-07 | fail→fail | 10,946 | 6,313 | -42% | 1 | 1 | 0% | 1,830 | 399 | -78% | 0 | 0 | — |
case-08 | fail→fail | 5,891 | 6,080 | +3% | 1 | 1 | 0% | 328 | 446 | +36% | 0 | 0 | — |
case-09 | fail→fail | 6,247 | 4,147 | -34% | 1 | 1 | 0% | 1,370 | 334 | -76% | 0 | 0 | — |
case-10 | fail→fail | 18,399 | 4,264 | -77% | 1 | 1 | 0% | 3,528 | 418 | -88% | 0 | 0 | — |
case-11 | fail→fail | 12,155 | 3,829 | -68% | 1 | 1 | 0% | 2,347 | 407 | -83% | 0 | 0 | — |
case-12 | fail→fail | 5,578 | 9,012 | +62% | 1 | 1 | 0% | 949 | 440 | -54% | 0 | 0 | — |
case-13 | fail→fail | 12,904 | 4,630 | -64% | 1 | 1 | 0% | 1,702 | 484 | -72% | 0 | 0 | — |
case-14 | fail→fail | 15,347 | 5,662 | -63% | 1 | 1 | 0% | 2,784 | 571 | -79% | 0 | 0 | — |
case-15 | fail→fail | 8,353 | 4,730 | -43% | 1 | 1 | 0% | 385 | 446 | +16% | 0 | 0 | — |
case-16 | fail→fail | 5,345 | 6,564 | +23% | 1 | 1 | 0% | 955 | 380 | -60% | 0 | 0 | — |
case-17 | fail→fail | 16,078 | 5,197 | -68% | 1 | 1 | 0% | 2,555 | 404 | -84% | 0 | 0 | — |
case-18 | fail→fail | 14,645 | 27,680 | +89% | 1 | 1 | 0% | 2,822 | 536 | -81% | 0 | 0 | — |
case-19 | fail→pass | 7,256 | 2,682 | -63% | 1 | 1 | 0% | 1,175 | 581 | -51% | 0 | 0 | — |
case-20 | pass→pass | 23,888 | 28,802 | +21% | 1 | 1 | 0% | 5,058 | 6,344 | +25% | 0 | 0 | — |
case-21 | pass→pass | 5,774 | 6,831 | +18% | 1 | 1 | 0% | 822 | 1,124 | +37% | 0 | 0 | — |
case-22 | pass→pass | 14,948 | 16,642 | +11% | 1 | 1 | 0% | 2,684 | 3,337 | +24% | 0 | 0 | — |
case-23 | pass→pass | 13,632 | 13,812 | +1% | 1 | 1 | 0% | 2,690 | 2,869 | +7% | 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. 23 cases were attempted, and 8 counted toward the lift figure. The other 15 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 +13 percentage points is the difference between those two pass rates over the 8 comparable cases.
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.