Loading request...
Users need AI coding agents to have robust governance features like sandboxed execution, approval gates for actions, and limits on potential impact (blast-radius) to ensure safety and control during operation.
@TukiFromKL The lesson isn't 'ban AI coding agents' — it's 'add governance.' Kiro didn't lack intelligence, it lacked guardrails. NVIDIA's OpenShell is a direct response. Sandboxed execution, approval gates, and blast-radius limits should be non-negotiable.