Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →拆解陈述中的逻辑谬误，并进行事实一致性核查。
.claude/skills/sdsds222-logic-debunker/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 24% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 4% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 2% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 4% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -6% | 0% |
执行 logic-debunker 拆解陈述中的逻辑谬误，并进行事实一致性核查。 <SKILL_CLEAN: logic-debunker>
你是一个冷酷无情、极其严谨的逻辑与事实核查专家。请严格按以下步骤进行内部推演：
【金句总结】：(用一段极其简练、犀利的大白话直接拆穿谎言或澄清事实，一针见血)
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 12,864 | 15,381 | +20% | 1 | 1 | 0% | 2,208 | 2,738 | +24% | 0 | 0 | — |
case-02 | fail→pass | 13,050 | 12,065 | -8% | 1 | 1 | 0% | 2,188 | 2,277 | +4% | 0 | 0 | — |
case-03 | fail→pass | 26,602 | 13,659 | -49% | 1 | 1 | 0% | 2,480 | 2,536 | +2% | 0 | 0 | — |
case-04 | pass→fail | 9,950 | 9,760 | -2% | 1 | 1 | 0% | 2,242 | 2,340 | +4% | 0 | 0 | — |
case-05 | pass→fail | 8,035 | 14,811 | +84% | 1 | 1 | 0% | 1,308 | 2,718 | +108% | 0 | 0 | — |
case-06 | pass→fail | 10,283 | 11,986 | +17% | 1 | 1 | 0% | 1,778 | 2,268 | +28% | 0 | 0 | — |
case-07 | fail→pass | 21,253 | 12,527 | -41% | 1 | 1 | 0% | 2,200 | 2,295 | +4% | 0 | 0 | — |
case-08 | fail→pass | 14,761 | 11,623 | -21% | 1 | 1 | 0% | 2,389 | 2,236 | -6% | 0 | 0 | — |
case-09 | fail→pass | 12,799 | 11,191 | -13% | 1 | 1 | 0% | 2,232 | 2,182 | -2% | 0 | 0 | — |
case-10 | fail→pass | 17,367 | 16,133 | -7% | 1 | 1 | 0% | 2,846 | 2,860 | +0% | 0 | 0 | — |
case-11 | fail→pass | 15,400 | 13,161 | -15% | 1 | 1 | 0% | 2,603 | 2,644 | +2% | 0 | 0 | — |
case-12 | fail→pass | 16,099 | 14,529 | -10% | 1 | 1 | 0% | 2,556 | 2,316 | -9% | 0 | 0 | — |
case-13 | fail→pass | 13,172 | 13,027 | -1% | 1 | 1 | 0% | 2,269 | 2,559 | +13% | 0 | 0 | — |
case-14 | fail→pass | 16,520 | 14,894 | -10% | 1 | 1 | 0% | 2,599 | 2,723 | +5% | 0 | 0 | — |
case-15 | fail→pass | 13,772 | 12,066 | -12% | 1 | 1 | 0% | 2,199 | 2,163 | -2% | 0 | 0 | — |
case-16 | fail→pass | 13,426 | 13,282 | -1% | 1 | 1 | 0% | 2,328 | 2,600 | +12% | 0 | 0 | — |
case-17 | fail→pass | 15,702 | 12,402 | -21% | 1 | 1 | 0% | 2,575 | 2,509 | -3% | 0 | 0 | — |
case-18 | fail→pass | 13,495 | 11,968 | -11% | 1 | 1 | 0% | 2,161 | 2,165 | +0% | 0 | 0 | — |
case-19 | fail→pass | 14,546 | 33,630 | +131% | 1 | 1 | 0% | 2,499 | 2,614 | +5% | 0 | 0 | — |
case-20 | fail→pass | 14,894 | 10,216 | -31% | 1 | 1 | 0% | 2,395 | 2,102 | -12% | 0 | 0 | — |
case-21 | fail→fail | 14,927 | 14,609 | -2% | 1 | 1 | 0% | 2,383 | 2,699 | +13% | 0 | 0 | — |
case-22 | fail→pass | 14,123 | 10,858 | -23% | 1 | 1 | 0% | 2,424 | 2,181 | -10% | 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. The headline lift of +68 percentage points is the difference between those two pass rates over the 22 comparable cases. 3 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.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 8/3/2026 | +77% |
Other measured skills in the registry, with their headline benchmark lift.