Comment “REVIEW” and I’ll send you the link. Anthropic just revealed how they’ve been using Claude to review their own code, and the results are pretty wild. Instead of relying only on human reviewers, Claude Code Review launches multiple agents that scan pull requests in parallel, search for bugs, verify findings to reduce false positives, and rank issues by severity before posting a single high-signal summary with inline comments. After running it internally for months, the percentage of PRs receiving substantive review comments jumped from 16% to 54%, while less than 1% of findings were marked incorrect by engineers. On large PRs, the system surfaces real issues consistently, showing how AI agents are starting to change the entire code review process. #claudeai #aicoding #devtools #anthropic #genai