
AI
Why I Stopped Letting One Agent Grade Its Own Homework
A single AI agent that writes code and then decides for itself whether it is correct will lie to you eventually. Splitting the job into a builder and an independent checker fixed that for me.
Notes and long-form posts on AI, web dev, and technology.

A single AI agent that writes code and then decides for itself whether it is correct will lie to you eventually. Splitting the job into a builder and an independent checker fixed that for me.

Most reviews of AI coding tools focus on demos. Here is the practical checklist I use to decide whether one earns a permanent spot in my workflow.

The official docs explain the API. Here is what I wish I knew about schemas, loaders, and image handling before building a real site on content collections.