By name: ai-safety description: Implement and review child-safe AI behavior, moderation gates, parent controls, and auditability for Ollama tool-calling features.
AI Safety Skill
Required controls
- Input moderation before LLM call.
- Output moderation before user rendering/apply.
- Parent approval for high-impact changes.
- Structured audit event for each AI action.
- Revert path for each AI mutation.
See references/policy-matrix.md for mode-specific restrictions.