BankBench-MY

Scenario Factory · Sinar Civic Tech
⚙ Settings & API key not set
$/Mtok
Keys are stored in your browser's localStorage only — never sent anywhere except the chosen provider endpoint, never written to disk or the repo.

Prompt a scenario category

Factory pipeline: dedup against your existing 42 scenarios → LLM seeds 5–7 in the fixed BankBench schema → you approve/edit/reject. Every seed is stamped with provenance. Try the gap classes from unaddressed-safety-gaps.md: E5 indirect · E6 mandate drift · E7 safety-theater · E8 multi-turn · E9 attack-register · E10 sycophancy.

Dedup check vs. your library

Existing scenarios
0
    Category coverage