Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when the user asks to review a skill, analyze skill quality, update a skill version, or run a repeatable keep/disable/archive decision loop from real failures instead of abstract best practices.
.claude/skills/cnfjlhj-skill-governance-loop/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-23 | ✗→✓ | ▲ Improved | -51% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 131% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -20% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -39% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -23% | 0% |
Use this skill for evidence-based governance of one skill or a small target set. It starts from a concrete problem, audits the skill, and ends with an explicit keep, disable, merge, split, or archive decision.
Use when:
Do not use when:
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-23 | fail→pass | 15,597 | 7,695 | -51% | 1 | 1 | 0% | 2,645 | 1,299 | -51% | 0 | 0 | — |
case-01 | fail→fail | 10,518 | 9,395 | -11% | 1 | 1 | 0% | 1,655 | 1,715 | +4% | 0 | 0 | — |
case-02 | pass→pass | 12,062 | 8,556 | -29% | 1 | 1 | 0% | 2,099 | 1,816 | -13% | 0 | 0 | — |
case-03 | fail→fail | 2,907 | 11,462 | +294% | 1 | 1 | 0% | 217 | 1,380 | +536% | 0 | 0 | — |
case-04 | fail→pass | 5,508 | 9,953 | +81% | 1 | 1 | 0% | 791 | 1,829 | +131% | 0 | 0 | — |
case-05 | pass→pass | 9,919 | 8,878 | -10% | 1 | 1 | 0% | 1,406 | 1,419 | +1% | 0 | 0 | — |
case-06 | fail→pass | 14,423 | 11,171 | -23% | 1 | 1 | 0% | 2,284 | 1,826 | -20% | 0 | 0 | — |
case-07 | pass→pass | 14,117 | 10,791 | -24% | 1 | 1 | 0% | 2,228 | 1,879 | -16% | 0 | 0 | — |
case-08 | pass→pass | 13,255 | 8,370 | -37% | 1 | 1 | 0% | 2,122 | 1,745 | -18% | 0 | 0 | — |
case-09 | pass→pass | 15,320 | 6,530 | -57% | 1 | 1 | 0% | 2,391 | 1,280 | -46% | 0 | 0 | — |
case-10 | pass→pass | 8,599 | 6,651 | -23% | 1 | 1 | 0% | 1,150 | 1,247 | +8% | 0 | 0 | — |
case-11 | pass→pass | 11,661 | 8,294 | -29% | 1 | 1 | 0% | 1,924 | 1,514 | -21% | 0 | 0 | — |
case-12 | fail→pass | 15,044 | 8,000 | -47% | 1 | 1 | 0% | 2,529 | 1,533 | -39% | 0 | 0 | — |
case-13 | fail→pass | 16,383 | 10,889 | -34% | 1 | 1 | 0% | 2,460 | 1,897 | -23% | 0 | 0 | — |
case-14 | pass→fail | 15,581 | 10,238 | -34% | 1 | 1 | 0% | 2,364 | 1,590 | -33% | 0 | 0 | — |
case-15 | pass→pass | 10,970 | 6,550 | -40% | 1 | 1 | 0% | 1,684 | 1,240 | -26% | 0 | 0 | — |
case-16 | fail→pass | 10,225 | 5,770 | -44% | 1 | 1 | 0% | 1,619 | 1,146 | -29% | 0 | 0 | — |
case-17 | fail→pass | 10,099 | 6,754 | -33% | 1 | 1 | 0% | 1,502 | 1,253 | -17% | 0 | 0 | — |
case-18 | pass→pass | 12,237 | 9,477 | -23% | 1 | 1 | 0% | 2,122 | 1,793 | -16% | 0 | 0 | — |
case-19 | pass→pass | 7,378 | 8,483 | +15% | 1 | 1 | 0% | 1,162 | 1,658 | +43% | 0 | 0 | — |
case-20 | pass→pass | 9,375 | 6,246 | -33% | 1 | 1 | 0% | 1,403 | 1,165 | -17% | 0 | 0 | — |
case-21 | fail→pass | 10,672 | 7,452 | -30% | 1 | 1 | 0% | 1,653 | 1,251 | -24% | 0 | 0 | — |
case-22 | pass→pass | 9,707 | 7,211 | -26% | 1 | 1 | 0% | 1,421 | 1,311 | -8% | 0 | 0 | — |
case-24 | pass→pass | 14,241 | 10,051 | -29% | 1 | 1 | 0% | 2,071 | 1,791 | -14% | 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. The headline lift of +29 percentage points is the difference between those two pass rates over the 24 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.