Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Wenn es um Änderungsvorbehalt 308 in AGB-Recht-Prüfer geht: zerlegt Ergebnis, Frist, Zuständigkeit, Beweislast und Gegenposition; liefert eine Gegenprüfung mit Fehler-, Beweis- und Fristencheck.
.claude/skills/klotzkette-aenderungsvorbehalt-308/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-06 | ✗→✓ | ▲ Improved | 53% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 34% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 95% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 133% | 0% |
| case-19 | ✗→✓ | ▲ Improved | 103% | 0% |
> Der Anbieter ist berechtigt, die in Anlage X] aufgeführten Leistungsmerkmale anzupassen, soweit dies aufgrund (a) gesetzlicher Vorgaben, (b) zwingender Anforderungen an die IT-Sicherheit oder (c) der Weiterentwicklung der zugrundeliegenden Software erforderlich ist und für den Kunden zumutbar ist. Über die geplante Anpassung wird der Anbieter den Kunden mindestens 6 Wochen vor Wirksamwerden in Textform informieren. Verändert die Anpassung den vertragstypischen Leistungsumfang erheblich, kann der Kunde den Vertrag bis zum Wirksamwerden außerordentlich kündigen.
Siehe references/QUELLEN.md, references/PRUEFLOGIK.md und references/KLAUSELFAMILIEN.md.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-12 | pass→pass | 15,786 | 13,987 | -11% | 1 | 1 | 0% | 2,798 | 4,215 | +51% | 0 | 0 | — |
case-01 | pass→pass | 13,456 | 16,907 | +26% | 1 | 1 | 0% | 2,243 | 4,460 | +99% | 0 | 0 | — |
case-02 | pass→pass | 8,425 | 8,623 | +2% | 1 | 1 | 0% | 1,503 | 3,140 | +109% | 0 | 0 | — |
case-03 | pass→pass | 15,959 | 16,656 | +4% | 1 | 1 | 0% | 2,755 | 4,419 | +60% | 0 | 0 | — |
case-04 | pass→pass | 14,237 | 19,332 | +36% | 1 | 1 | 0% | 2,590 | 4,990 | +93% | 0 | 0 | — |
case-05 | pass→pass | 10,143 | 14,225 | +40% | 1 | 1 | 0% | 1,943 | 4,265 | +120% | 0 | 0 | — |
case-06 | fail→pass | 14,333 | 12,734 | -11% | 1 | 1 | 0% | 2,393 | 3,654 | +53% | 0 | 0 | — |
case-07 | pass→pass | 13,724 | 18,104 | +32% | 1 | 1 | 0% | 2,172 | 4,678 | +115% | 0 | 0 | — |
case-08 | pass→pass | 14,244 | 14,975 | +5% | 1 | 1 | 0% | 2,375 | 4,055 | +71% | 0 | 0 | — |
case-09 | pass→pass | 13,028 | 17,293 | +33% | 1 | 1 | 0% | 2,264 | 4,770 | +111% | 0 | 0 | — |
case-10 | pass→pass | 12,046 | 17,053 | +42% | 1 | 1 | 0% | 1,975 | 4,800 | +143% | 0 | 0 | — |
case-11 | fail→pass | 19,525 | 13,980 | -28% | 1 | 1 | 0% | 3,209 | 4,306 | +34% | 0 | 0 | — |
case-13 | pass→pass | 11,193 | 13,964 | +25% | 1 | 1 | 0% | 1,874 | 4,046 | +116% | 0 | 0 | — |
case-14 | fail→pass | 10,288 | 8,327 | -19% | 1 | 1 | 0% | 1,578 | 3,073 | +95% | 0 | 0 | — |
case-15 | fail→pass | 9,008 | 9,160 | +2% | 1 | 1 | 0% | 1,430 | 3,333 | +133% | 0 | 0 | — |
case-16 | pass→pass | 7,150 | 9,759 | +36% | 1 | 1 | 0% | 1,212 | 3,324 | +174% | 0 | 0 | — |
case-17 | pass→pass | 11,803 | 14,023 | +19% | 1 | 1 | 0% | 1,954 | 4,183 | +114% | 0 | 0 | — |
case-18 | pass→pass | 4,711 | 5,682 | +21% | 1 | 1 | 0% | 794 | 2,564 | +223% | 0 | 0 | — |
case-19 | fail→pass | 11,352 | 12,228 | +8% | 1 | 1 | 0% | 1,885 | 3,827 | +103% | 0 | 0 | — |
case-20 | pass→pass | 14,193 | 15,699 | +11% | 1 | 1 | 0% | 2,342 | 4,681 | +100% | 0 | 0 | — |
case-21 | fail→pass | 19,938 | 22,707 | +14% | 1 | 1 | 0% | 2,836 | 4,486 | +58% | 0 | 0 | — |
case-22 | pass→pass | 7,204 | 7,899 | +10% | 1 | 1 | 0% | 1,363 | 2,967 | +118% | 0 | 0 | — |
case-23 | pass→pass | 15,776 | 15,764 | -0% | 1 | 1 | 0% | 2,698 | 4,618 | +71% | 0 | 0 | — |
case-24 | pass→pass | 15,125 | 16,486 | +9% | 1 | 1 | 0% | 2,827 | 4,603 | +63% | 0 | 0 | — |
case-25 | pass→pass | 10,586 | 15,092 | +43% | 1 | 1 | 0% | 1,982 | 4,040 | +104% | 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. 25 cases were attempted. The headline lift of +24 percentage points is the difference between those two pass rates over the 25 comparable cases.
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.