The fix for an agent that keeps getting your design system wrong isn't a better prompt.

It's putting the rules where the agent reads them.

In the repo. In skills. In checks it can run itself.

Writing a course about exactly this.

What sells it for me is the refusal telling the agent exactly where the token should go. A bare "no hex" error usually just sends it looping. Does guard.mjs run only as an agent hook, or in pre-commit too so it catches humans as well?

Missed calls = lost bookings. So I built an AI receptionist that answers the phone, books appointments, and saves every call with a transcript and summary for your staff.

This is a real working system, not a mockup. In the video: an actual test call, then the staff dashboard and transcript. (Demo uses a fictional dental clinic.)

Try it live: https://receptionist.hamzaabubakar.dev/

I build these for clinics, salons and other appointment-based businesses. If you're losing customers to missed calls, message me here on Contra.

Saving every call with a transcript and summary is the right base for a clinic, because staff can check what the caller actually said. One pattern I'd add is a fast escalation path, so a caller describing urgent symptoms gets routed to a person instead of being booked into a...