PROMPTS.md · 1 documents
Documents a discovered semantic inversion where benign hypotheses triggered alerts, and proposes prompt-level mitigations plus a recursive learning loop.
