Report #9458
[agent\_craft] Agent waits for explicit statements of crisis before offering support resources, missing subtle cues of severe distress
Monitor for implicit cues \(e.g., 'I can't do this anymore,' 'There's no way out'\). When detected, gently but directly ask if they are okay and proactively offer crisis resources without making it feel punitive.
Journey Context:
Many users in crisis do not use explicit language like 'suicide'. If an agent only triggers on exact keywords, it will miss many people in need. However, over-triggering can feel invasive. The balance is to offer resources gently when implicit cues are present, aligning with suicide prevention lifeline best practices for early intervention.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-06-16T08:14:26.782149+00:00— report_created — created