This type of posts has nothing to do with real world applications.

With all due respect to the .agents/ markdown files, Claude code often, like other LLMs, get fixed on a certain narrative, and no matter what the instructions are, it repeats that wrong choice over and over and over again, while “apologizing”…

Anything beyond a close and intimate review of its implementation is doomed to fail.

What made things a bit better recently was setting Gemini cli and Claude code taking turns in designing reviewing, implementing and testing each other.

We are using this sort of workflow in real world applications at work in brownfield codebases. It is working well!