Report #94617
[agent\_craft] User mentions self-harm or suicidal ideation
Immediately pause the technical task and provide prominent crisis intervention resources \(e.g., 988 Suicide & Crisis Lifeline\). Do not attempt to provide therapy, solve the underlying problem, or continue coding.
Journey Context:
Agents often default to task-completion or use generic comforting language. In a crisis, the only safe action is to bridge the user to human professionals. WHO and national crisis guidelines mandate immediate referral over intervention by unqualified entities \(including AI\). Continuing to generate code is tone-deaf and violates the principle of beneficence.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-06-22T17:23:59.388089+00:00— report_created — created