Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Deep web full-text retrieval via Apify RAG browser. Import of web-browsing/web-research skill. Full page content for substantive analysis.
.claude/skills/yogsoth-ai-stress-test-web-research/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | -36% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -73% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -68% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -71% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -62% | 0% |
Deep web full-text retrieval for substantive analysis.
Import — strictly follow web-browsing/web-research skill protocol.
Must fetch full page via Apify RAG browser. No shortcuts — full text required for any substantive claim about web content.
Quantity target is set by the calling strategy's budget table. This SOP executes one unit = one apify/rag-web-browser call.
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-02 | fail→fail | 63,214 | 40,702 | -36% | 1 | 1 | 0% | 8,152 | 6,288 | -23% | 0 | 0 | — |
case-01 | fail→fail | 20,518 | 34,509 | +68% | 1 | 1 | 0% | 2,399 | 5,260 | +119% | 0 | 0 | — |
case-03 | fail→fail | 23,365 | 50,424 | +116% | 1 | 1 | 0% | 832 | 8,437 | +914% | 0 | 0 | — |
case-04 | pass→pass | 7,463 | 7,033 | -6% | 1 | 1 | 0% | 422 | 535 | +27% | 0 | 0 | — |
case-05 | pass→pass | 14,018 | 13,239 | -6% | 1 | 1 | 0% | 1,804 | 1,956 | +8% | 0 | 0 | — |
case-06 | pass→pass | 9,579 | 8,550 | -11% | 1 | 1 | 0% | 766 | 871 | +14% | 0 | 0 | — |
case-15 | pass→pass | 28,772 | 6,843 | -76% | 1 | 1 | 0% | 1,489 | 482 | -68% | 0 | 0 | — |
case-07 | fail→pass | 19,218 | 12,542 | -35% | 1 | 1 | 0% | 2,095 | 1,332 | -36% | 0 | 0 | — |
case-08 | fail→pass | 18,691 | 7,190 | -62% | 1 | 1 | 0% | 2,056 | 548 | -73% | 0 | 0 | — |
case-09 | fail→pass | 20,264 | 8,345 | -59% | 1 | 1 | 0% | 2,401 | 763 | -68% | 0 | 0 | — |
case-10 | fail→fail | 15,983 | 10,621 | -34% | 1 | 1 | 0% | 1,689 | 928 | -45% | 0 | 0 | — |
case-11 | fail→pass | 14,532 | 6,804 | -53% | 1 | 1 | 0% | 1,544 | 451 | -71% | 0 | 0 | — |
case-12 | fail→pass | 16,315 | 7,737 | -53% | 1 | 1 | 0% | 1,697 | 642 | -62% | 0 | 0 | — |
case-13 | fail→pass | 11,419 | 7,260 | -36% | 1 | 1 | 0% | 914 | 493 | -46% | 0 | 0 | — |
case-14 | fail→fail | 17,244 | 4,264 | -75% | 1 | 1 | 0% | 1,803 | 833 | -54% | 0 | 0 | — |
case-16 | fail→fail | 19,503 | 16,592 | -15% | 1 | 1 | 0% | 2,190 | 2,617 | +19% | 0 | 0 | — |
case-17 | pass→pass | 15,771 | 10,501 | -33% | 1 | 1 | 0% | 2,211 | 1,704 | -23% | 0 | 0 | — |
case-18 | fail→pass | 17,686 | 2,357 | -87% | 1 | 1 | 0% | 1,991 | 594 | -70% | 0 | 0 | — |
case-19 | fail→fail | 13,821 | 12,201 | -12% | 1 | 1 | 0% | 2,312 | 1,391 | -40% | 0 | 0 | — |
case-20 | fail→fail | 14,277 | 15,324 | +7% | 1 | 1 | 0% | 2,294 | 2,619 | +14% | 0 | 0 | — |
case-21 | fail→pass | 22,534 | 14,844 | -34% | 1 | 1 | 0% | 2,860 | 2,716 | -5% | 0 | 0 | — |
case-22 | fail→pass | 10,778 | 2,045 | -81% | 1 | 1 | 0% | 1,711 | 532 | -69% | 0 | 0 | — |
case-23 | fail→fail | 17,306 | 6,258 | -64% | 1 | 1 | 0% | 2,643 | 1,213 | -54% | 0 | 0 | — |
case-24 | pass→pass | 7,800 | 2,429 | -69% | 1 | 1 | 0% | 1,086 | 502 | -54% | 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. 24 cases were attempted, and 23 counted toward the lift figure. The other 1 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 +38 percentage points is the difference between those two pass rates over the 23 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.