Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Wenn es um Agentur, Abnahme, Vergabe in Barrierefreiheit Web Checker geht: prüft Frist, Form, Zuständigkeit, Rechtsweg und Sofortmaßnahmen; liefert eine Fristen- und Risikoampel mit Sofortschritten.
.claude/skills/klotzkette-agentur-abnahme-vergabe/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 27% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 35% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 57% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 23% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 30% | 0% |
Dieses Fachmodul begleitet Websites, die neu gebaut, relauncht oder von einer Agentur nachgebessert werden.
| Bereich | Abnahmekriterium | Nachweis | Status | | --- | --- | --- | --- | | Navigation | Tastatur vollständig | Testprotokoll | offen/ok | | Formulare | Labels und Fehler | Screenshot + DOM | offen/ok | | Checkout | Vertragsschluss möglich | User Journey | offen/ok | | PDF | getaggt oder HTML | Dokumentencheck | offen/ok |
Accessibility-Overlay, Plugin oder Widget ersetzt keine barrierefreie Umsetzung. Es kann ergänzen, aber nicht die eigentlichen HTML-, Design-, Content- und Prozessfehler heilen.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-12 | pass→pass | 25,007 | 27,270 | +9% | 1 | 1 | 0% | 2,850 | 4,201 | +47% | 0 | 0 | — |
case-01 | fail→pass | 38,329 | 34,384 | -10% | 1 | 1 | 0% | 4,714 | 5,967 | +27% | 0 | 0 | — |
case-02 | fail→pass | 34,981 | 37,756 | +8% | 1 | 1 | 0% | 3,895 | 5,263 | +35% | 0 | 0 | — |
case-03 | fail→pass | 33,384 | 39,193 | +17% | 1 | 1 | 0% | 4,047 | 6,355 | +57% | 0 | 0 | — |
case-04 | pass→pass | 20,234 | 15,552 | -23% | 1 | 1 | 0% | 2,447 | 3,367 | +38% | 0 | 0 | — |
case-05 | pass→pass | 22,293 | 17,341 | -22% | 1 | 1 | 0% | 2,121 | 2,896 | +37% | 0 | 0 | — |
case-06 | fail→pass | 18,088 | 15,665 | -13% | 1 | 1 | 0% | 2,093 | 2,579 | +23% | 0 | 0 | — |
case-07 | fail→pass | 23,915 | 21,763 | -9% | 1 | 1 | 0% | 2,418 | 3,140 | +30% | 0 | 0 | — |
case-08 | pass→pass | 22,371 | 20,919 | -6% | 1 | 1 | 0% | 2,673 | 3,217 | +20% | 0 | 0 | — |
case-09 | fail→pass | 25,645 | 29,186 | +14% | 1 | 1 | 0% | 3,134 | 3,839 | +22% | 0 | 0 | — |
case-10 | fail→pass | 24,238 | 20,014 | -17% | 1 | 1 | 0% | 3,277 | 3,304 | +1% | 0 | 0 | — |
case-11 | pass→pass | 17,306 | 21,803 | +26% | 1 | 1 | 0% | 1,699 | 3,094 | +82% | 0 | 0 | — |
case-13 | fail→fail | 22,088 | 25,860 | +17% | 1 | 1 | 0% | 2,346 | 3,626 | +55% | 0 | 0 | — |
case-14 | pass→pass | 24,160 | 25,127 | +4% | 1 | 1 | 0% | 3,049 | 4,146 | +36% | 0 | 0 | — |
case-15 | fail→pass | 18,537 | 23,846 | +29% | 1 | 1 | 0% | 2,460 | 3,977 | +62% | 0 | 0 | — |
case-16 | fail→pass | 18,375 | 15,979 | -13% | 1 | 1 | 0% | 1,500 | 2,339 | +56% | 0 | 0 | — |
case-22 | pass→pass | 26,129 | 26,046 | -0% | 1 | 1 | 0% | 2,853 | 3,698 | +30% | 0 | 0 | — |
case-17 | fail→pass | 14,595 | 10,056 | -31% | 1 | 1 | 0% | 1,358 | 1,436 | +6% | 0 | 0 | — |
case-18 | pass→pass | 24,333 | 11,710 | -52% | 1 | 1 | 0% | 1,591 | 1,656 | +4% | 0 | 0 | — |
case-19 | fail→pass | 18,876 | 19,227 | +2% | 1 | 1 | 0% | 2,132 | 2,687 | +26% | 0 | 0 | — |
case-20 | pass→pass | 25,304 | 33,369 | +32% | 1 | 1 | 0% | 3,471 | 4,841 | +39% | 0 | 0 | — |
case-21 | pass→pass | 12,148 | 19,754 | +63% | 1 | 1 | 0% | 2,077 | 3,411 | +64% | 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.
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.