Demo: AI agent safety (open to suggestions)

I built a small prototype around model optimization 026.

What it does

• Solves a tiny but annoying problem
• Focused on speed + clarity

Feedback I want

• UX flow
• Edge cases
• What to add/remove