Crisis Response Protocol

Implements the safety protocols you need when building AI coaching apps that touch mental health or addiction recovery. You get tiered response levels from gentle check-ins to emergency escalation, plus crisis detection logic that scans for primary indicators like suicidal ideation before your AI ever responds. The code shows exactly how to disable chat and surface 988 hotline info when risk hits critical, with built-in logging for human review. Pairs naturally with crisis detection skills since this handles the "what happens next" part. The key insight here is that it prevents your AI from being the sole responder by always providing paths to real counselors and emergency services, which is the only responsible approach.

Install to Claude Code

npx -y skills add erichowens/some_claude_skills --skill crisis-response-protocol --agent claude-code