Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Multi-stage PRISMA screening tactic — progressively filter papers from a large candidate pool to a focused set for deep reading. Four stages (identification, title/abstract screening, full-text screening, inclusion) with documented counts at each stage.
.claude/skills/yogsoth-ai-prisma-screening/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 15% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -31% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -50% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -43% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -26% | 0% |
Multi-stage screening following PRISMA methodology. Progressively filter papers from a large candidate pool to a focused set for deep reading.
paper-overview (import) — identification stagepaper-search (import) — full-text screening stagepaper-research (import) — inclusion stage (deep reading)prisma-flowchart (subagent) — generate PRISMA flow data<!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use | | --- | --- | | knowledge-acquisition-paper-overview | Abstract-level paper scanning for broad coverage. Import of literature-engine/literature-overview skill. Abstract-level only — no methodology conclusions from abstracts. | | knowledge-acquisition-paper-research | Full-depth paper reading with raw text extraction. Import of literature-engine/literature-research skill. Must read fullText (true) — equations, hyperparameters, specific claims extracted. | | knowledge-acquisition-paper-search | AI-summarized paper reading for intermediate depth. Import of literature-engine/literature-search skill. Must call get_paper_content for every analyzed paper. | | prisma-flowchart | Generate PRISMA-compliant flow data documenting the screening funnel — counts at each stage (identification, screening, eligibility, inclusion) with exclusion reasons. Used by systematic-survey via prisma-screening tactic. |
<!-- 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→pass | 44,709 | 48,794 | +9% | 1 | 1 | 0% | 7,667 | 8,795 | +15% | 0 | 0 | — |
case-02 | fail→fail | 51,676 | 53,074 | +3% | 1 | 1 | 0% | 8,269 | 8,798 | +6% | 0 | 0 | — |
case-03 | fail→pass | 17,821 | 9,817 | -45% | 1 | 1 | 0% | 2,011 | 1,389 | -31% | 0 | 0 | — |
case-04 | fail→pass | 18,544 | 10,567 | -43% | 1 | 1 | 0% | 2,371 | 1,180 | -50% | 0 | 0 | — |
case-05 | fail→pass | 16,818 | 8,489 | -50% | 1 | 1 | 0% | 1,852 | 1,060 | -43% | 0 | 0 | — |
case-06 | fail→pass | 16,242 | 13,531 | -17% | 1 | 1 | 0% | 2,822 | 2,088 | -26% | 0 | 0 | — |
case-07 | pass→pass | 18,096 | 10,496 | -42% | 1 | 1 | 0% | 1,794 | 1,235 | -31% | 0 | 0 | — |
case-08 | fail→pass | 14,034 | 10,326 | -26% | 1 | 1 | 0% | 1,556 | 1,043 | -33% | 0 | 0 | — |
case-09 | pass→pass | 21,154 | 16,613 | -21% | 1 | 1 | 0% | 2,578 | 2,638 | +2% | 0 | 0 | — |
case-10 | pass→pass | 12,117 | 11,772 | -3% | 1 | 1 | 0% | 1,789 | 1,573 | -12% | 0 | 0 | — |
case-11 | fail→pass | 22,660 | 13,423 | -41% | 1 | 1 | 0% | 2,761 | 2,424 | -12% | 0 | 0 | — |
case-16 | fail→pass | 18,085 | 8,606 | -52% | 1 | 1 | 0% | 1,946 | 1,164 | -40% | 0 | 0 | — |
case-12 | fail→fail | 22,971 | 21,869 | -5% | 1 | 1 | 0% | 2,897 | 3,444 | +19% | 0 | 0 | — |
case-13 | fail→fail | 21,209 | 9,668 | -54% | 1 | 1 | 0% | 2,432 | 1,369 | -44% | 0 | 0 | — |
case-14 | fail→pass | 19,093 | 18,813 | -1% | 1 | 1 | 0% | 2,784 | 2,728 | -2% | 0 | 0 | — |
case-15 | pass→pass | 13,981 | 9,524 | -32% | 1 | 1 | 0% | 1,635 | 1,362 | -17% | 0 | 0 | — |
case-17 | pass→pass | 15,546 | 12,912 | -17% | 1 | 1 | 0% | 1,652 | 1,272 | -23% | 0 | 0 | — |
case-18 | fail→fail | 24,098 | 39,827 | +65% | 1 | 1 | 0% | 3,942 | 8,772 | +123% | 0 | 0 | — |
case-19 | fail→fail | 24,779 | 14,893 | -40% | 1 | 1 | 0% | 2,824 | 2,256 | -20% | 0 | 0 | — |
case-20 | fail→fail | 57,891 | 42,907 | -26% | 1 | 1 | 0% | 8,228 | 6,527 | -21% | 0 | 0 | — |
case-21 | fail→fail | 12,890 | 25,620 | +99% | 1 | 1 | 0% | 2,256 | 4,518 | +100% | 0 | 0 | — |
case-22 | fail→pass | 17,615 | 12,300 | -30% | 1 | 1 | 0% | 666 | 930 | +40% | 0 | 0 | — |
case-23 | fail→pass | 8,651 | 2,337 | -73% | 1 | 1 | 0% | 1,198 | 952 | -21% | 0 | 0 | — |
case-24 | fail→pass | 25,606 | 13,654 | -47% | 1 | 1 | 0% | 4,283 | 1,969 | -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 +50 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.