1Prompt eval cues predicted refusal shifts across 32k LLM rollouts (opens in new tab)(medium.com)1ratnaditya1mo ago0Save
2Show HN: AgentWard – After an AI agent deleted files, I built a runtime enforcer (opens in new tab)(github.com)GitHub1ratnaditya4mo ago1Save
