AI Code Review
What automated review actually produces: how CodeRabbit and Sourcery build line-by-line suggestions, which context MCP brings in, and why Copilot needs a fresh request after new commits.
Automated review is no longer plain diff reading. CodeRabbit posts change summaries and per-line patches, and through MCP can cite tickets and monitoring signals; Sourcery moves the same checks into the IDE and adds scheduled security scans; GitHub Copilot does not re-run after a fresh push until you ask again. Edge cases and missing tests are caught fastest, while payments, permissions and migrations still need a human decision.
