Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Anwalt oder Mandant will UK/US-Vertragsverhandlung oder Mandantengespraech simulieren und False-Friends-Lernkurve absolvieren. Prüfraster Verhandlungs-Simulation Issue-List-Erstellung Mandanten-Erklärung. Output Simulations-Szenario Verhandlungs-Script Lern-Feedback. Abgrenzung zu common-law-quality-gate (Qualitaet) und common-law-client-explainer (Erklärung).
.claude/skills/thomasmoreai-common-law-simulation-negotiation/SKILL.md| Model | Eval pass | Runs |
|---|---|---|
| gemini-3.6-flash | — | 1 |
| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 35% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 88% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 90% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 64% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 65% | 0% |
Fiktive Verhandlungsrunden trainieren junge Anwältinnen und Anwälte mit klar markierten Simulationsunterlagen.
Common-Law-Kompass arbeitet freundlich, präzise und verzeihend. Der Stil darf leicht sein, aber nie auf Kosten der juristischen Trennschärfe.
Kläre vor dem Training:
UK
US
Praxis
| Szenario | Governing Law | Knackpunkt | Typischer Trainee-Fehler | |---|---|---|---| | SPA-Verhandlung (Tech-M&A) | English Law | MAC-Definition, Locked-Box vs Completion Accounts | MAC zu eng definieren | | NDA-Verhandlung | New York Law | Definition of Confidential Information, Residuals | "Residuals-Klausel" nicht bemerken | | Kreditvertrag | English Law (LMA) | Financial Covenants, Cure Rights | Equity Cure nicht verhandeln | | Settlement Agreement | English Law | Release-Breite, BCCI-Risiko | Unbekannte Claims freisetzen | | JV-Term Sheet | DE + English Law | Good-Faith-Pflicht vs. Walford v Miles | BGB-Standard auf englischen Vertrag anwenden |
Adressat: Trainer / Supervisorin — Tonfall: strukturiert, lernorientiert
VERHANDLUNGSSIMULATION — PROTOKOLL
Datum: [DATUM]
Szenario: [BEZEICHNUNG]
Trainee: [NAME / ANONYMISIERT]
Partei: [KAEUFER / VERKAEUFER / ...]
Governing Law: [RECHTSORDNUNG]
ERGEBNIS DER SIMULATION
Einigungspunkte: [...]
Offene Punkte: [...]
Erreichtes BATNA-Ergebnis: [BESSER / SCHLECHTER / EINIGUNG]
BEOBACHTUNGEN
Staerken: [...]
Entwicklungsbereiche: [...]
FALSE-FRIENDS-VORFAELLE
- [Begriff] wurde falsch verwendet: [Beschreibung] — Korrekt: [...]
JURISTISCHE FEHLER
- [Punkt] — Richtige Regel: [Kurz] — Fundstelle: [...]
FEEDBACK-GESPRAECH
Kernbotschaft: [1-2 Saetze]
Naechster Uebungsschritt: [...]| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 27,556 | 26,348 | -4% | 1 | 1 | 0% | 4,517 | 6,059 | +34% | 0 | 0 | — |
case-02 | fail→pass | 25,487 | 25,399 | -0% | 1 | 1 | 0% | 4,347 | 5,883 | +35% | 0 | 0 | — |
case-03 | pass→pass | 24,619 | 21,712 | -12% | 1 | 1 | 0% | 4,210 | 5,160 | +23% | 0 | 0 | — |
case-04 | fail→fail | 16,728 | 16,405 | -2% | 1 | 1 | 0% | 2,564 | 3,731 | +46% | 0 | 0 | — |
case-05 | pass→pass | 15,585 | 15,421 | -1% | 1 | 1 | 0% | 2,417 | 4,423 | +83% | 0 | 0 | — |
case-06 | fail→pass | 15,168 | 15,690 | +3% | 1 | 1 | 0% | 2,293 | 4,321 | +88% | 0 | 0 | — |
case-07 | fail→pass | 13,240 | 16,754 | +27% | 1 | 1 | 0% | 2,305 | 4,374 | +90% | 0 | 0 | — |
case-08 | pass→pass | 16,375 | 20,714 | +26% | 1 | 1 | 0% | 2,729 | 5,202 | +91% | 0 | 0 | — |
case-09 | pass→pass | 16,291 | 19,296 | +18% | 1 | 1 | 0% | 2,910 | 4,830 | +66% | 0 | 0 | — |
case-10 | pass→pass | 12,081 | 14,832 | +23% | 1 | 1 | 0% | 2,046 | 4,175 | +104% | 0 | 0 | — |
case-11 | pass→pass | 8,437 | 11,467 | +36% | 1 | 1 | 0% | 1,574 | 3,700 | +135% | 0 | 0 | — |
case-12 | pass→fail | 15,999 | 17,864 | +12% | 1 | 1 | 0% | 2,813 | 4,723 | +68% | 0 | 0 | — |
case-13 | fail→pass | 16,961 | 18,613 | +10% | 1 | 1 | 0% | 2,914 | 4,770 | +64% | 0 | 0 | — |
case-14 | fail→pass | 15,411 | 13,549 | -12% | 1 | 1 | 0% | 2,356 | 3,898 | +65% | 0 | 0 | — |
case-15 | fail→fail | 20,299 | 20,760 | +2% | 1 | 1 | 0% | 3,548 | 4,824 | +36% | 0 | 0 | — |
case-16 | pass→pass | 13,290 | 17,516 | +32% | 1 | 1 | 0% | 2,142 | 4,331 | +102% | 0 | 0 | — |
case-17 | pass→pass | 7,909 | 10,332 | +31% | 1 | 1 | 0% | 1,368 | 3,225 | +136% | 0 | 0 | — |
case-18 | fail→pass | 9,187 | 12,558 | +37% | 1 | 1 | 0% | 1,579 | 3,649 | +131% | 0 | 0 | — |
case-19 | pass→pass | 10,278 | 7,164 | -30% | 1 | 1 | 0% | 1,863 | 2,835 | +52% | 0 | 0 | — |
case-20 | pass→pass | 17,672 | 19,562 | +11% | 1 | 1 | 0% | 2,903 | 5,318 | +83% | 0 | 0 | — |
case-21 | pass→pass | 20,630 | 19,271 | -7% | 1 | 1 | 0% | 3,129 | 4,887 | +56% | 0 | 0 | — |
case-22 | pass→fail | 16,016 | 20,635 | +29% | 1 | 1 | 0% | 3,133 | 5,268 | +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. 22 cases were attempted. The headline lift of +18 percentage points is the difference between those two pass rates over the 22 comparable cases. 2 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.