I pre-registered a small study, pointed a 12-reviewer pipeline at three merged Copilot PRs in major . NET organisations, and the failure mode I found was not the one everyone talks about. Disclosure first: I maintain review-pro , the open-source review system used as the instrument here.
Source: [Dev.to](https://dev.to/tufan_tunc/the-agent-didnt-hallucinate-it-ignored-what-the-repo-already-knew-2m44)