Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Cluster papers by theme, method, or timeline. Produces natural groupings from a paper collection. Used by scoping-survey and narrative-review.
.claude/skills/yogsoth-ai-categorize-papers/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 25% | 0% |
| case-22 | ✓→✓ | ＝ Same ✓ | 60% | 0% |
| case-20 | ✓→✓ | ＝ Same ✓ | 17% | 0% |
| case-21 | ✓→✓ | ＝ Same ✓ | -33% | 0% |
| case-01 | ✗→✗ | ＝ Same ✗ | -77% | 0% |
Cluster papers by theme/method/timeline into natural groupings.
Subagent — spawned via subagent-spawning/spawn-agent skill.
Categorization requires holding the entire paper collection in context simultaneously to identify patterns and groupings. Subagent provides dedicated space for this comparative analysis.
<!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use | | --- | --- | | spawn-agent | Spawn a customized CC subagent with full MCP tool access. Used by SOPs that declare execution: subagent. |
<!-- 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→pass | 18,353 | 22,272 | +21% | 1 | 1 | 0% | 2,822 | 3,536 | +25% | 0 | 0 | — |
case-01 | fail→fail | 9,946 | 6,224 | -37% | 1 | 1 | 0% | 1,580 | 358 | -77% | 0 | 0 | — |
case-03 | fail→fail | 21,083 | 41,944 | +99% | 1 | 1 | 0% | 3,449 | 6,194 | +80% | 0 | 0 | — |
case-04 | fail→fail | 10,070 | 8,767 | -13% | 1 | 1 | 0% | 1,485 | 498 | -66% | 0 | 0 | — |
case-05 | fail→fail | 6,320 | 24,275 | +284% | 1 | 1 | 0% | 964 | 4,300 | +346% | 0 | 0 | — |
case-06 | fail→fail | 8,057 | 8,754 | +9% | 1 | 1 | 0% | 1,237 | 847 | -32% | 0 | 0 | — |
case-22 | pass→pass | 6,424 | 10,030 | +56% | 1 | 1 | 0% | 1,039 | 1,661 | +60% | 0 | 0 | — |
case-07 | fail→fail | 8,513 | 8,191 | -4% | 1 | 1 | 0% | 1,550 | 529 | -66% | 0 | 0 | — |
case-08 | fail→fail | 24,551 | 5,272 | -79% | 1 | 1 | 0% | 3,957 | 1,000 | -75% | 0 | 0 | — |
case-09 | fail→fail | 9,695 | 3,270 | -66% | 1 | 1 | 0% | 1,515 | 622 | -59% | 0 | 0 | — |
case-10 | fail→fail | 4,776 | 9,923 | +108% | 1 | 1 | 0% | 725 | 1,218 | +68% | 0 | 0 | — |
case-11 | fail→fail | 13,314 | 6,052 | -55% | 1 | 1 | 0% | 2,069 | 925 | -55% | 0 | 0 | — |
case-12 | fail→fail | 11,438 | 2,408 | -79% | 1 | 1 | 0% | 1,636 | 463 | -72% | 0 | 0 | — |
case-13 | fail→fail | 16,435 | 7,270 | -56% | 1 | 1 | 0% | 2,622 | 829 | -68% | 0 | 0 | — |
case-14 | fail→fail | 6,195 | 3,674 | -41% | 1 | 1 | 0% | 981 | 665 | -32% | 0 | 0 | — |
case-15 | fail→fail | 6,267 | 13,732 | +119% | 1 | 1 | 0% | 945 | 1,394 | +48% | 0 | 0 | — |
case-16 | fail→fail | 2,826 | 5,157 | +82% | 1 | 1 | 0% | 380 | 816 | +115% | 0 | 0 | — |
case-17 | fail→fail | 4,284 | 11,985 | +180% | 1 | 1 | 0% | 664 | 775 | +17% | 0 | 0 | — |
case-18 | fail→fail | 8,729 | 5,998 | -31% | 1 | 1 | 0% | 1,375 | 1,046 | -24% | 0 | 0 | — |
case-19 | fail→fail | 2,767 | 11,748 | +325% | 1 | 1 | 0% | 387 | 681 | +76% | 0 | 0 | — |
case-20 | pass→pass | 7,898 | 9,305 | +18% | 1 | 1 | 0% | 1,411 | 1,655 | +17% | 0 | 0 | — |
case-21 | pass→pass | 7,121 | 4,146 | -42% | 1 | 1 | 0% | 1,319 | 890 | -33% | 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 14 counted toward the lift figure. The other 8 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 +5 percentage points is the difference between those two pass rates over the 14 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.