Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Award NOS's (Newcastle-Ottawa Scale) stars item-by-item across Selection (up to 4), Comparability (up to 2), and Outcome/Exposure (up to 3) — a binary award-or-not action per item, distinct from a 5-value signalling judgment. Use this after study-design-tool-gate has dispatched to NOS, as the first step before sum-threshold-scoring.
.claude/skills/yogsoth-ai-star-awarding/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-22 | ✓→✓ | ＝ Same ✓ | 30% | 0% |
| case-23 | ✓→✓ | ＝ Same ✓ | 31% | 0% |
| case-24 | ✓→✓ | ＝ Same ✓ | 77% | 0% |
| case-11 | ✗→✗ | ＝ Same ✗ | 92% | 0% |
| case-01 | ✗→✗ | ＝ Same ✗ | -17% | 0% |
NOS's item-by-item star awarding — binary per item, not a signalling-question judgment. Added per coverage-audit M8: the original graph had NOS's stars appearing already-summed at an aggregation node, with no node actually doing the item-level awarding those sums depend on.
Subagent — spawned via spawn-agent skill.
<!-- BEGIN available-tables (generated) -->
| SOP | When to use | | --- | --- | | spawn-agent | Spawn a customized CC subagent with full MCP tool access. |
<!-- 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-11 | fail→fail | 23,121 | 39,771 | +72% | 1 | 1 | 0% | 3,162 | 6,085 | +92% | 0 | 0 | — |
case-01 | fail→fail | 12,021 | 29,665 | +147% | 1 | 1 | 0% | 1,240 | 1,025 | -17% | 0 | 0 | — |
case-02 | fail→fail | 12,404 | 26,739 | +116% | 1 | 1 | 0% | 1,211 | 3,975 | +228% | 0 | 0 | — |
case-03 | fail→fail | 12,777 | 22,806 | +78% | 1 | 1 | 0% | 1,350 | 685 | -49% | 0 | 0 | — |
case-04 | fail→fail | 23,210 | 33,243 | +43% | 1 | 1 | 0% | 3,138 | 5,366 | +71% | 0 | 0 | — |
case-05 | fail→fail | 29,245 | 23,159 | -21% | 1 | 1 | 0% | 4,088 | 3,477 | -15% | 0 | 0 | — |
case-06 | fail→fail | 21,707 | 19,643 | -10% | 1 | 1 | 0% | 2,584 | 2,534 | -2% | 0 | 0 | — |
case-07 | fail→fail | 27,780 | 23,919 | -14% | 1 | 1 | 0% | 4,143 | 1,609 | -61% | 0 | 0 | — |
case-08 | fail→fail | 26,109 | 35,390 | +36% | 1 | 1 | 0% | 3,532 | 5,610 | +59% | 0 | 0 | — |
case-09 | fail→fail | 20,827 | 36,013 | +73% | 1 | 1 | 0% | 2,720 | 2,562 | -6% | 0 | 0 | — |
case-10 | fail→fail | 17,556 | 18,569 | +6% | 1 | 1 | 0% | 2,302 | 2,598 | +13% | 0 | 0 | — |
case-12 | fail→fail | 23,393 | 24,707 | +6% | 1 | 1 | 0% | 3,358 | 863 | -74% | 0 | 0 | — |
case-13 | fail→fail | 26,692 | 47,258 | +77% | 1 | 1 | 0% | 3,803 | 5,150 | +35% | 0 | 0 | — |
case-14 | fail→fail | 29,038 | 23,059 | -21% | 1 | 1 | 0% | 3,965 | 3,195 | -19% | 0 | 0 | — |
case-15 | fail→fail | 23,881 | 27,736 | +16% | 1 | 1 | 0% | 3,305 | 4,296 | +30% | 0 | 0 | — |
case-16 | fail→fail | 17,593 | 16,760 | -5% | 1 | 1 | 0% | 2,003 | 1,975 | -1% | 0 | 0 | — |
case-17 | fail→fail | 23,551 | 24,300 | +3% | 1 | 1 | 0% | 3,297 | 3,574 | +8% | 0 | 0 | — |
case-18 | fail→fail | 17,604 | 18,732 | +6% | 1 | 1 | 0% | 1,977 | 2,481 | +25% | 0 | 0 | — |
case-19 | fail→fail | 22,246 | 23,820 | +7% | 1 | 1 | 0% | 3,050 | 3,360 | +10% | 0 | 0 | — |
case-20 | fail→fail | 25,060 | 22,632 | -10% | 1 | 1 | 0% | 3,477 | 908 | -74% | 0 | 0 | — |
case-21 | fail→fail | 22,457 | 33,782 | +50% | 1 | 1 | 0% | 2,929 | 5,256 | +79% | 0 | 0 | — |
case-22 | pass→pass | 12,745 | 14,145 | +11% | 1 | 1 | 0% | 1,382 | 1,795 | +30% | 0 | 0 | — |
case-23 | pass→pass | 23,286 | 28,066 | +21% | 1 | 1 | 0% | 3,203 | 4,184 | +31% | 0 | 0 | — |
case-24 | pass→pass | 21,536 | 33,465 | +55% | 1 | 1 | 0% | 2,557 | 4,515 | +77% | 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 18 counted toward the lift figure. The other 6 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 0 percentage points is the difference between those two pass rates over the 18 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
Other measured skills in the registry, with their headline benchmark lift.