TIP. Canary tokens for working with AI agents. Best helper ever!

moonlearning on YouTubeWatch on YouTube

SaveMark complete

Sign in to save videos and track what you’ve finished.

IntermediateUI4 min

About this video

The video shows how to make an AI agent print a visible marker each time it loads a context file, so you can see which sources it actually opened. It is aimed at people already setting up agent workflows, including design system work.

What you’ll learn

  • Add visible canary tokens so an AI agent reports each file it loads
  • Spot when an agent skips context files it was meant to read
  • Recognise that a file being opened is not the same as being read