Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Paper landscape scan returning abstracts and metadata. Import of literature-engine/paper-overview skill. Abstracts only — no conclusions from abstracts.
.claude/skills/yogsoth-ai-stress-test-paper-overview/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 56% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 29% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 42% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 4% | 0% |
| case-20 | ✓→✗ | ▼ Worse | -88% | 0% |
Paper landscape scan — abstracts and metadata only.
Import — strictly follow literature-engine/paper-overview skill protocol.
Returns abstracts only. Do NOT draw conclusions about methodology, results, or contributions from abstracts. Use paper-search or paper-research for substantive claims.
Quantity target is set by the calling strategy's budget table. This SOP executes one unit = one discover_papers or relevanceSearch call.
literature-engine repo → skills/paper-overview/SKILL.md
<!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use | | --- | --- | | literature-overview | Quick landscape scan — discover papers on a topic without full-text reading | | stress-test-paper-research | Paper full text access via alphaxiv answer_pdf_queries or get_paper_content(fullText=true). Import of literature-engine/paper-research skill. Raw extracted text for precise claims. | | stress-test-paper-search | Paper AI summary report via alphaxiv get_paper_content. Import of literature-engine/paper-search skill. Structured AI-generated intermediate report. |
<!-- 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-08 | fail→fail | 23,635 | 40,891 | +73% | 1 | 1 | 0% | 4,218 | 5,266 | +25% | 0 | 0 | — |
case-01 | fail→fail | 27,388 | 16,827 | -39% | 1 | 1 | 0% | 4,024 | 782 | -81% | 0 | 0 | — |
case-02 | fail→pass | 22,296 | 29,749 | +33% | 1 | 1 | 0% | 3,053 | 4,759 | +56% | 0 | 0 | — |
case-03 | fail→fail | 38,407 | 18,116 | -53% | 1 | 1 | 0% | 4,875 | 848 | -83% | 0 | 0 | — |
case-04 | fail→pass | 31,471 | 30,634 | -3% | 1 | 1 | 0% | 4,473 | 5,784 | +29% | 0 | 0 | — |
case-05 | fail→fail | 25,369 | 19,876 | -22% | 1 | 1 | 0% | 3,967 | 1,168 | -71% | 0 | 0 | — |
case-06 | fail→pass | 30,223 | 42,065 | +39% | 1 | 1 | 0% | 3,783 | 5,376 | +42% | 0 | 0 | — |
case-07 | fail→fail | 24,540 | 18,301 | -25% | 1 | 1 | 0% | 3,481 | 910 | -74% | 0 | 0 | — |
case-09 | fail→fail | 21,394 | 19,269 | -10% | 1 | 1 | 0% | 3,581 | 1,007 | -72% | 0 | 0 | — |
case-10 | fail→fail | 67,053 | 14,354 | -79% | 1 | 1 | 0% | 4,284 | 841 | -80% | 0 | 0 | — |
case-11 | fail→pass | 23,961 | 26,737 | +12% | 1 | 1 | 0% | 4,177 | 4,324 | +4% | 0 | 0 | — |
case-12 | fail→fail | 36,693 | 16,371 | -55% | 1 | 1 | 0% | 6,237 | 1,606 | -74% | 0 | 0 | — |
case-13 | fail→fail | 41,963 | 12,410 | -70% | 1 | 1 | 0% | 6,278 | 909 | -86% | 0 | 0 | — |
case-14 | fail→fail | 36,308 | 38,001 | +5% | 1 | 1 | 0% | 5,521 | 6,850 | +24% | 0 | 0 | — |
case-15 | fail→fail | 19,408 | 19,955 | +3% | 1 | 1 | 0% | 3,298 | 3,775 | +14% | 0 | 0 | — |
case-16 | fail→fail | 24,610 | 38,729 | +57% | 1 | 1 | 0% | 3,515 | 4,995 | +42% | 0 | 0 | — |
case-17 | fail→fail | 34,668 | 19,997 | -42% | 1 | 1 | 0% | 6,094 | 1,047 | -83% | 0 | 0 | — |
case-18 | fail→fail | 37,828 | 15,852 | -58% | 1 | 1 | 0% | 5,613 | 974 | -83% | 0 | 0 | — |
case-19 | fail→fail | 43,655 | 59,237 | +36% | 1 | 1 | 0% | 8,238 | 8,535 | +4% | 0 | 0 | — |
case-20 | pass→fail | 30,933 | 17,633 | -43% | 1 | 1 | 0% | 6,705 | 780 | -88% | 0 | 0 | — |
case-21 | pass→fail | 30,494 | 11,910 | -61% | 1 | 1 | 0% | 5,098 | 789 | -85% | 0 | 0 | — |
case-22 | pass→pass | 30,958 | 34,640 | +12% | 1 | 1 | 0% | 2,797 | 4,299 | +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. 22 cases were attempted, and 10 counted toward the lift figure. The other 12 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 +9 percentage points is the difference between those two pass rates over the 10 comparable cases. 10 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.