Blog
Engineering notes from Zoroval.
- Code review can’t keep up with agents. Gates can. Coding agents write more code than any team can read. Reading every line does not scale, and blocking PRs kills your speed. We moved review from lines to evidence. Here are the gates we run, from a local branch to production, and a skill to set them up in your repo.
- Finding one failure mode in a 230,000 character trace A reviewer open-codes one specific failure in an agent trace. Now you must find it again in thousands of traces that are too long to read. We tried three approaches. Here is what we measured.