Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Wenn es um Japan BAV und Corporate Pension — Deutsch-Japanische Perspektive in BAV Strategie Konzern — Treuenfels Yamamoto Rechtsanwälte geht: entwickelt Verhandlungsziel, Vergleichskorridor und Eskalationspfad; liefert eine Verhandlungs- oder Eskalationslinie mit Optionen.
.claude/skills/klotzkette-japan-bav-und-corporate-pension-iorp/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 29% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 22% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 105% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 103% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 81% | 0% |
expatriate-pensionsplanung-und-totalization)| Deutsch | Japanisch (Kanji) | Romaji | |---------|------------------|--------| | Betriebliche Altersversorgung | 企業年金 | Kigyou nenkin | | Defined Benefit (Leistungszusage) | 確定給付 | Kakutei kyuufu | | Defined Contribution (Beitragszusage) | 確定拠出 | Kakutei kyoshutsu | | Stiftungstyp-Plan | 基金型企業年金 | Kikin-gata kigyou nenkin | | Vertragstyp-Plan | 規約型企業年金 | Kiyaku-gata kigyou nenkin | | Einmalzahlung | 一時金 | Ichijikikin | | Rentenleistung | 年金給付 | Nenkin kyuufu | | Unverfallbarkeit | 受給資格 | Jukyu shikaku | | Arbeitnehmer-Vertreter | 代議員 | Daigiin | | Ministerium Gesundheit/Arbeit | 厚生労働省 | Kosei-rodo-sho | | Finanzaufsicht | 金融庁 | Kinyu-cho | | Altersrentenversicherung (AN) | 厚生年金保険 | Kosei nenkin hoken | | Nationale Rentenversicherung | 国民年金 | Kokumin nenkin | | Treuhandbank (Verwaltung) | 信託銀行 | Shintaku ginko |
BERATUNGSMEMORANDUM
確定給付企業年金プランの見直しについて
(Überprüfung des Defined Benefit Corporate Pension Plans)
Von / 発信: Japan-/Datenschutz-Team, bengoshi
bAV-Projektteam
Kyoto-Büro, Gion-Higashi, Shijō-dōri, Kyoto
An / 宛先: [Japanische Konzerngesellschaft, HR und Legal]
Datum: [Datum]
Ref.: [Aktennummer]
Sprache / 言語: Deutsch (mit japanischen Fachbegriffen) / 日本語抄録付き
ZUSAMMENFASSUNG (EXECUTIVE SUMMARY):
[Deutsch-Text]
要約:
[Japanisch-Text durch Japan-/Datenschutz-Team]ENTSCHEIDUNGSMATRIX DB vs. DC — JAPAN
bAV-Projektteam / Kyoto-Büro
Kriterium | Kakutei-kyuufu (DB) | Kakutei-kyoshutsu (DC)
-----------------------|--------------------|-----------------------
Leistungsgarantie | Arbeitgeber trägt Risiko | Arbeitnehmer trägt Anlagerisiko
Bilanzierung | Rückstellungspflicht | Kein Bilanzansatz
MHLW-Genehmigung | Erforderlich | Registrierung
Portabilität | Begrenzt (Kikin-Auflösung) | Hoch (ichi-gen)
Anlageentscheidung | Arbeitgeber/Treuhänder | Arbeitnehmer wählt
Steuervorteil AG | Beitrag abzugsfähig | Beitrag abzugsfähig
Beitragsgrenze (2024) | Keine Obergrenze | JPY 55.000 p.m. (ohne anderen Plan)country-by-country-benefits-matrix-konzern — Japan in der Konzernmatrixexpatriate-pensionsplanung-und-totalization — DE-JP SV-Abkommeninternationale-harmonisierung-konzern-bav — Konzernweite Harmonisierung mit Japan-Bezuginternationale-buyout-datenflows-und-datenschutz — APPI Japan Datenschutz> Quellenregel: Entscheidungen nur nach Prüfung einer amtlichen oder frei zugänglichen Quelle mit Gericht, Entscheidungsform, Datum, Aktenzeichen und tragender Aussage ausgeben.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 57,278 | 60,813 | +6% | 1 | 1 | 0% | 8,279 | 10,655 | +29% | 0 | 0 | — |
case-02 | fail→pass | 49,853 | 44,182 | -11% | 1 | 1 | 0% | 6,312 | 7,677 | +22% | 0 | 0 | — |
case-03 | fail→fail | 43,286 | 44,733 | +3% | 1 | 1 | 0% | 5,727 | 7,981 | +39% | 0 | 0 | — |
case-04 | pass→pass | 19,952 | 20,704 | +4% | 1 | 1 | 0% | 3,264 | 5,662 | +73% | 0 | 0 | — |
case-05 | fail→pass | 30,398 | 30,862 | +2% | 1 | 1 | 0% | 3,265 | 6,684 | +105% | 0 | 0 | — |
case-06 | pass→pass | 24,304 | 32,191 | +32% | 1 | 1 | 0% | 3,197 | 6,884 | +115% | 0 | 0 | — |
case-07 | pass→pass | 26,586 | 46,701 | +76% | 1 | 1 | 0% | 3,939 | 8,844 | +125% | 0 | 0 | — |
case-08 | pass→pass | 27,465 | 36,443 | +33% | 1 | 1 | 0% | 4,359 | 7,592 | +74% | 0 | 0 | — |
case-09 | pass→pass | 26,696 | 41,708 | +56% | 1 | 1 | 0% | 4,037 | 8,641 | +114% | 0 | 0 | — |
case-10 | fail→fail | 21,863 | 27,149 | +24% | 1 | 1 | 0% | 3,016 | 6,761 | +124% | 0 | 0 | — |
case-11 | pass→pass | 27,252 | 42,124 | +55% | 1 | 1 | 0% | 3,911 | 9,615 | +146% | 0 | 0 | — |
case-12 | fail→fail | 19,101 | 21,184 | +11% | 1 | 1 | 0% | 3,028 | 5,724 | +89% | 0 | 0 | — |
case-13 | fail→pass | 18,891 | 26,130 | +38% | 1 | 1 | 0% | 3,251 | 6,593 | +103% | 0 | 0 | — |
case-14 | fail→fail | 21,920 | 26,483 | +21% | 1 | 1 | 0% | 3,249 | 6,637 | +104% | 0 | 0 | — |
case-15 | fail→pass | 24,983 | 23,835 | -5% | 1 | 1 | 0% | 3,914 | 7,095 | +81% | 0 | 0 | — |
case-16 | fail→pass | 17,973 | 24,282 | +35% | 1 | 1 | 0% | 2,985 | 6,615 | +122% | 0 | 0 | — |
case-17 | fail→pass | 19,523 | 25,756 | +32% | 1 | 1 | 0% | 2,480 | 6,330 | +155% | 0 | 0 | — |
case-18 | fail→pass | 26,860 | 17,594 | -34% | 1 | 1 | 0% | 4,331 | 5,366 | +24% | 0 | 0 | — |
case-19 | pass→pass | 18,263 | 33,550 | +84% | 1 | 1 | 0% | 3,022 | 5,906 | +95% | 0 | 0 | — |
case-20 | fail→fail | 24,885 | 24,612 | -1% | 1 | 1 | 0% | 3,861 | 6,675 | +73% | 0 | 0 | — |
case-21 | pass→pass | 25,999 | 53,462 | +106% | 1 | 1 | 0% | 4,125 | 8,306 | +101% | 0 | 0 | — |
case-22 | fail→fail | 27,078 | 39,557 | +46% | 1 | 1 | 0% | 4,453 | 8,833 | +98% | 0 | 0 | — |
case-23 | fail→fail | 33,108 | 23,086 | -30% | 1 | 1 | 0% | 3,634 | 5,912 | +63% | 0 | 0 | — |
case-24 | pass→pass | 29,164 | 27,294 | -6% | 1 | 1 | 0% | 3,737 | 6,283 | +68% | 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 +33 percentage points is the difference between those two pass rates over the 24 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.