I ended up documenting what I found while testing different CV layouts and parsing behavior across ATS.

It’s not a tool and it doesn’t generate CVs — it’s a technical, text-first system explaining what tends to break parsing, what survives reliably, and how to make structural decisions that don’t depend on folklore or guesswork.

Sharing it here in case it’s useful for others looking into ATS behavior:

https://gumroad.com/l/atspasskit

Happy to answer technical questions or clarify assumptions if anyone’s interested.