Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Wenn es um Internationale Fehlerkatalog in BAV Strategie Konzern — Treuenfels Yamamoto Rechtsanwälte geht: prüft Frist, Form, Zuständigkeit, Rechtsweg und Sofortmaßnahmen; liefert eine Fristen- und Risikoampel mit Sofortschritten.
.claude/skills/klotzkette-internationale-fehlerkatalog/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -14% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 274% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 43% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 22% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -6% | 0% |
Dieser Fehlerkatalog prüft Arbeitsergebnisse für Betriebliche Altersversorgung im Konzern vor Abgabe, Versand oder Mandantenfreigabe gegen die im Sachgebiet typischen Fehlerquellen — jeweils mit Symptom, Diagnose und Heilung.
Roter/gelber/grüner Befund je Fehlerachse; jeder rote Punkt mit konkreter Korrektur und verbleibendem Restrisiko. Quellenhygiene nach references/quellenhygiene.md.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-13 | pass→pass | 29,388 | 16,694 | -43% | 1 | 1 | 0% | 2,574 | 3,196 | +24% | 0 | 0 | — |
case-01 | fail→pass | 30,799 | 17,109 | -44% | 1 | 1 | 0% | 3,406 | 2,934 | -14% | 0 | 0 | — |
case-02 | fail→pass | 16,326 | 38,652 | +137% | 1 | 1 | 0% | 1,525 | 5,705 | +274% | 0 | 0 | — |
case-03 | fail→pass | 23,662 | 34,116 | +44% | 1 | 1 | 0% | 3,647 | 5,216 | +43% | 0 | 0 | — |
case-04 | pass→fail | 16,235 | 29,632 | +83% | 1 | 1 | 0% | 1,985 | 4,577 | +131% | 0 | 0 | — |
case-05 | pass→fail | 20,924 | 44,937 | +115% | 1 | 1 | 0% | 2,658 | 6,712 | +153% | 0 | 0 | — |
case-06 | pass→fail | 28,454 | 38,818 | +36% | 1 | 1 | 0% | 3,622 | 5,492 | +52% | 0 | 0 | — |
case-07 | fail→pass | 30,749 | 27,233 | -11% | 1 | 1 | 0% | 3,535 | 4,317 | +22% | 0 | 0 | — |
case-08 | fail→pass | 22,807 | 14,715 | -35% | 1 | 1 | 0% | 2,991 | 2,815 | -6% | 0 | 0 | — |
case-09 | fail→pass | 19,380 | 20,733 | +7% | 1 | 1 | 0% | 1,880 | 3,036 | +61% | 0 | 0 | — |
case-10 | fail→pass | 20,898 | 20,443 | -2% | 1 | 1 | 0% | 2,828 | 3,637 | +29% | 0 | 0 | — |
case-11 | fail→pass | 13,804 | 16,184 | +17% | 1 | 1 | 0% | 1,840 | 3,333 | +81% | 0 | 0 | — |
case-12 | fail→pass | 10,830 | 8,675 | -20% | 1 | 1 | 0% | 1,661 | 2,223 | +34% | 0 | 0 | — |
case-14 | fail→pass | 19,187 | 23,185 | +21% | 1 | 1 | 0% | 2,716 | 4,216 | +55% | 0 | 0 | — |
case-15 | fail→fail | 9,973 | 5,197 | -48% | 1 | 1 | 0% | 1,622 | 1,549 | -5% | 0 | 0 | — |
case-16 | fail→pass | 15,401 | 11,417 | -26% | 1 | 1 | 0% | 2,603 | 2,703 | +4% | 0 | 0 | — |
case-17 | fail→pass | 14,703 | 21,431 | +46% | 1 | 1 | 0% | 2,080 | 3,848 | +85% | 0 | 0 | — |
case-18 | pass→pass | 23,576 | 24,492 | +4% | 1 | 1 | 0% | 3,132 | 4,284 | +37% | 0 | 0 | — |
case-19 | pass→pass | 20,891 | 18,372 | -12% | 1 | 1 | 0% | 2,824 | 3,302 | +17% | 0 | 0 | — |
case-20 | fail→pass | 14,245 | 5,981 | -58% | 1 | 1 | 0% | 2,191 | 1,974 | -10% | 0 | 0 | — |
case-21 | pass→pass | 20,434 | 14,494 | -29% | 1 | 1 | 0% | 2,866 | 2,867 | +0% | 0 | 0 | — |
case-22 | fail→pass | 17,389 | 18,724 | +8% | 1 | 1 | 0% | 2,752 | 3,561 | +29% | 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 +50 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.
Other measured skills in the registry, with their headline benchmark lift.