Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Wenn es um Audit-Trail: Revisionssichere Freigabedokumentation im Aussenhandel in Außenwirtschaft, Sanktionen, Zoll und CBAM geht: prüft Frist, Form, Zuständigkeit, Rechtsweg und Sofortmaßnahmen; liefert eine Fristen- und Risikoampel mit Sofortschritten.
.claude/skills/klotzkette-aussenwirtschaft-audit-trail-freigaben/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 98% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 20% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -6% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 92% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 162% | 0% |
Aufbau und Pflege revisionssicherer Audit-Trails für Exportkontroll-Freigaben: Dokumentationsstandards nach AWG § 14 und AWV § 24 (Aufbewahrungsfristen) sowie Anforderungen der EU-Dual-Use-VO Art. 20. Sichert Freigabeprotokolle, Screening-Logs, Genehmigungsunterlagen und Kommunikation für Prüfungssituationen. Output: Audit-Trail-Struktur und Dokumentationsrichtlinie. Arbeite entlang dieser konkreten Prüfungslinie und trenne Rolle, Frist, Zuständigkeit, Beweislast und gewünschten Output.
Dieser Skill kann mit thematisch benachbarten Skills kombiniert werden, insbesondere:
aussenwirtschaft-sanktionsscreening-fuzzy-matchaussenwirtschaft-gueterlisten-klassifizierungaussenwirtschaft-freiwillige-offenlegung-bafa-zollaussenwirtschaft-icp-kontrollsystem| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-15 | pass→pass | 22,175 | 27,712 | +25% | 1 | 1 | 0% | 2,868 | 5,533 | +93% | 0 | 0 | — |
case-16 | pass→pass | 17,833 | 26,648 | +49% | 1 | 1 | 0% | 2,073 | 4,529 | +118% | 0 | 0 | — |
case-13 | pass→fail | 18,756 | 27,546 | +47% | 1 | 1 | 0% | 2,225 | 5,315 | +139% | 0 | 0 | — |
case-14 | pass→pass | 11,803 | 26,017 | +120% | 1 | 1 | 0% | 1,592 | 5,106 | +221% | 0 | 0 | — |
case-02 | fail→pass | 31,273 | 42,624 | +36% | 1 | 1 | 0% | 4,685 | 9,293 | +98% | 0 | 0 | — |
case-01 | fail→fail | 68,894 | 41,684 | -39% | 1 | 1 | 0% | 1,806 | 7,749 | +329% | 0 | 0 | — |
case-03 | fail→pass | 41,931 | 35,756 | -15% | 1 | 1 | 0% | 6,050 | 7,241 | +20% | 0 | 0 | — |
case-04 | fail→fail | 20,034 | 52,260 | +161% | 1 | 1 | 0% | 3,005 | 7,971 | +165% | 0 | 0 | — |
case-05 | fail→fail | 29,055 | 30,060 | +3% | 1 | 1 | 0% | 3,837 | 6,107 | +59% | 0 | 0 | — |
case-06 | fail→fail | 20,090 | 30,532 | +52% | 1 | 1 | 0% | 3,190 | 5,917 | +85% | 0 | 0 | — |
case-07 | fail→pass | 29,607 | 23,030 | -22% | 1 | 1 | 0% | 4,472 | 4,208 | -6% | 0 | 0 | — |
case-12 | fail→fail | 20,096 | 27,848 | +39% | 1 | 1 | 0% | 2,486 | 5,614 | +126% | 0 | 0 | — |
case-08 | pass→pass | 16,990 | 22,373 | +32% | 1 | 1 | 0% | 1,962 | 4,624 | +136% | 0 | 0 | — |
case-09 | fail→pass | 19,392 | 26,808 | +38% | 1 | 1 | 0% | 2,522 | 4,854 | +92% | 0 | 0 | — |
case-10 | fail→pass | 16,476 | 20,391 | +24% | 1 | 1 | 0% | 1,681 | 4,400 | +162% | 0 | 0 | — |
case-11 | fail→pass | 19,112 | 25,054 | +31% | 1 | 1 | 0% | 2,476 | 5,122 | +107% | 0 | 0 | — |
case-17 | pass→pass | 17,984 | 25,786 | +43% | 1 | 1 | 0% | 2,084 | 4,928 | +136% | 0 | 0 | — |
case-18 | fail→pass | 16,686 | 19,145 | +15% | 1 | 1 | 0% | 1,930 | 4,195 | +117% | 0 | 0 | — |
case-19 | fail→fail | 14,938 | 24,007 | +61% | 1 | 1 | 0% | 2,209 | 5,123 | +132% | 0 | 0 | — |
case-20 | fail→pass | 20,916 | 26,643 | +27% | 1 | 1 | 0% | 2,214 | 5,677 | +156% | 0 | 0 | — |
case-21 | pass→pass | 17,679 | 31,395 | +78% | 1 | 1 | 0% | 1,711 | 4,967 | +190% | 0 | 0 | — |
case-22 | pass→pass | 13,357 | 35,040 | +162% | 1 | 1 | 0% | 1,411 | 5,786 | +310% | 0 | 0 | — |
case-23 | pass→pass | 16,386 | 21,097 | +29% | 1 | 1 | 0% | 1,802 | 4,204 | +133% | 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. 23 cases were attempted, and 22 counted toward the lift figure. The other 1 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +30 percentage points is the difference between those two pass rates over the 22 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.