Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Deep web page analysis with full content extraction. Import of web-browsing/web-research skill. Must fetch full page via apify — no shortcuts.
.claude/skills/yogsoth-ai-creative-ideation-web-research/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | 69% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -76% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -57% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 194% | 0% |
| case-17 | ✗→✓ | ▲ Improved | 119% | 0% |
Deep web page analysis with full content extraction.
Import — strictly follow web-browsing/web-research skill protocol.
Must fetch full page content via apify rag-web-browser. No conclusions from snippets or partial content.
Quantity target is set by the calling strategy's budget table. This SOP executes one unit = one full page fetch + analysis.
web-browsing repo → skills/web-research/SKILL.md
<!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use | | --- | --- | | web-research | Deep web research — fetches full page content for analysis. Snippets alone are PROHIBITED for conclusions. |
<!-- 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 | 4,630 | 5,909 | +28% | 1 | 1 | 0% | 688 | 616 | -10% | 0 | 0 | — |
case-02 | fail→fail | 4,957 | 15,330 | +209% | 1 | 1 | 0% | 726 | 2,618 | +261% | 0 | 0 | — |
case-03 | fail→fail | 5,584 | 11,525 | +106% | 1 | 1 | 0% | 819 | 1,979 | +142% | 0 | 0 | — |
case-04 | pass→pass | 2,582 | 1,703 | -34% | 1 | 1 | 0% | 392 | 398 | +2% | 0 | 0 | — |
case-05 | pass→pass | 8,983 | 8,205 | -9% | 1 | 1 | 0% | 1,428 | 1,432 | +0% | 0 | 0 | — |
case-06 | pass→fail | 4,500 | 7,892 | +75% | 1 | 1 | 0% | 692 | 619 | -11% | 0 | 0 | — |
case-07 | fail→pass | 8,505 | 11,501 | +35% | 1 | 1 | 0% | 1,219 | 2,065 | +69% | 0 | 0 | — |
case-08 | fail→pass | 13,124 | 2,029 | -85% | 1 | 1 | 0% | 1,992 | 486 | -76% | 0 | 0 | — |
case-09 | fail→fail | 9,387 | 5,558 | -41% | 1 | 1 | 0% | 1,322 | 1,121 | -15% | 0 | 0 | — |
case-10 | fail→pass | 10,574 | 3,055 | -71% | 1 | 1 | 0% | 1,536 | 654 | -57% | 0 | 0 | — |
case-11 | fail→fail | 4,348 | 5,469 | +26% | 1 | 1 | 0% | 644 | 511 | -21% | 0 | 0 | — |
case-12 | fail→pass | 4,527 | 14,885 | +229% | 1 | 1 | 0% | 720 | 2,120 | +194% | 0 | 0 | — |
case-13 | fail→fail | 5,319 | 4,572 | -14% | 1 | 1 | 0% | 804 | 382 | -52% | 0 | 0 | — |
case-14 | fail→fail | 5,220 | 7,741 | +48% | 1 | 1 | 0% | 777 | 503 | -35% | 0 | 0 | — |
case-15 | fail→fail | 3,449 | 5,790 | +68% | 1 | 1 | 0% | 430 | 464 | +8% | 0 | 0 | — |
case-16 | fail→fail | 4,697 | 7,715 | +64% | 1 | 1 | 0% | 673 | 1,398 | +108% | 0 | 0 | — |
case-17 | fail→pass | 2,438 | 4,167 | +71% | 1 | 1 | 0% | 375 | 823 | +119% | 0 | 0 | — |
case-18 | fail→fail | 3,990 | 13,365 | +235% | 1 | 1 | 0% | 669 | 1,957 | +193% | 0 | 0 | — |
case-19 | fail→pass | 10,523 | 9,639 | -8% | 1 | 1 | 0% | 1,774 | 1,798 | +1% | 0 | 0 | — |
case-20 | fail→fail | 5,329 | 14,181 | +166% | 1 | 1 | 0% | 764 | 2,496 | +227% | 0 | 0 | — |
case-21 | fail→fail | 4,350 | 11,581 | +166% | 1 | 1 | 0% | 589 | 2,111 | +258% | 0 | 0 | — |
case-22 | fail→fail | 3,679 | 9,382 | +155% | 1 | 1 | 0% | 501 | 784 | +56% | 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 15 counted toward the lift figure. The other 7 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 15 comparable cases. 2 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.