Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Execute a pairwise comparison between two candidates, producing a judgment with winner, confidence, and reasoning.
.claude/skills/yogsoth-ai-comparison-executor/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | -37% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 87% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -59% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -59% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -20% | 0% |
Executes a single pairwise comparison between two candidates. Produces a structured judgment indicating the winner, confidence level, and detailed reasoning supporting the decision.
Runs as a subagent. Receives a pair of candidates and evaluation context, returns a judgment.
Each comparison requires focused deliberation on the specific pair without being influenced by prior ranking expectations. Isolation prevents anchoring bias from the current leaderboard.
Output MUST declare exactly one winner from the pair (no ties unless context explicitly permits ties). Confidence MUST be a value in 0.5, 1.0]. Reasoning MUST reference specific attributes of both candidates.
<!-- 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 | 10,422 | 5,095 | -51% | 1 | 1 | 0% | 1,734 | 1,091 | -37% | 0 | 0 | — |
case-01 | fail→fail | 2,682 | 5,758 | +115% | 1 | 1 | 0% | 413 | 1,186 | +187% | 0 | 0 | — |
case-03 | pass→pass | 10,641 | 6,180 | -42% | 1 | 1 | 0% | 1,736 | 1,256 | -28% | 0 | 0 | — |
case-04 | fail→pass | 4,306 | 7,217 | +68% | 1 | 1 | 0% | 749 | 1,403 | +87% | 0 | 0 | — |
case-05 | pass→pass | 2,279 | 3,918 | +72% | 1 | 1 | 0% | 348 | 809 | +132% | 0 | 0 | — |
case-06 | pass→pass | 17,321 | 9,695 | -44% | 1 | 1 | 0% | 2,934 | 1,802 | -39% | 0 | 0 | — |
case-07 | fail→pass | 11,290 | 3,210 | -72% | 1 | 1 | 0% | 1,797 | 744 | -59% | 0 | 0 | — |
case-08 | pass→pass | 9,411 | 2,221 | -76% | 1 | 1 | 0% | 1,595 | 605 | -62% | 0 | 0 | — |
case-09 | pass→pass | 13,238 | 12,445 | -6% | 1 | 1 | 0% | 2,017 | 1,973 | -2% | 0 | 0 | — |
case-10 | fail→pass | 8,618 | 2,341 | -73% | 1 | 1 | 0% | 1,430 | 585 | -59% | 0 | 0 | — |
case-11 | fail→pass | 5,913 | 3,677 | -38% | 1 | 1 | 0% | 1,102 | 880 | -20% | 0 | 0 | — |
case-17 | pass→pass | 11,802 | 4,709 | -60% | 1 | 1 | 0% | 1,985 | 1,143 | -42% | 0 | 0 | — |
case-12 | fail→pass | 9,855 | 5,844 | -41% | 1 | 1 | 0% | 1,548 | 1,163 | -25% | 0 | 0 | — |
case-13 | pass→fail | 12,555 | 2,719 | -78% | 1 | 1 | 0% | 2,066 | 676 | -67% | 0 | 0 | — |
case-14 | fail→pass | 7,782 | 3,605 | -54% | 1 | 1 | 0% | 1,209 | 840 | -31% | 0 | 0 | — |
case-15 | fail→pass | 11,703 | 5,475 | -53% | 1 | 1 | 0% | 1,955 | 1,115 | -43% | 0 | 0 | — |
case-16 | fail→pass | 16,087 | 1,501 | -91% | 1 | 1 | 0% | 1,232 | 435 | -65% | 0 | 0 | — |
case-18 | pass→pass | 10,678 | 7,435 | -30% | 1 | 1 | 0% | 1,579 | 1,389 | -12% | 0 | 0 | — |
case-19 | fail→pass | 9,054 | 4,528 | -50% | 1 | 1 | 0% | 1,549 | 971 | -37% | 0 | 0 | — |
case-20 | pass→pass | 15,459 | 25,127 | +63% | 1 | 1 | 0% | 3,900 | 6,392 | +64% | 0 | 0 | — |
case-21 | pass→pass | 8,534 | 8,798 | +3% | 1 | 1 | 0% | 1,346 | 1,718 | +28% | 0 | 0 | — |
case-22 | pass→pass | 19,813 | 31,574 | +59% | 1 | 1 | 0% | 3,225 | 5,479 | +70% | 0 | 0 | — |
case-23 | fail→pass | 17,838 | 9,220 | -48% | 1 | 1 | 0% | 2,789 | 1,552 | -44% | 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. 23 cases were attempted, and 22 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 +43 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is 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.