Anthropic has released a research preview of Claude Code Review for Team and Enterprise users. The feature uses AI agents to automatically review every pull request, catching and ranking bugs by severity while filtering false positives. When a PR is opened, agents scale in number based on complexity, producing a high-signal overview comment and inline bug comments. Reviews average about 20 minutes and cost $15–25 based on token usage. Internally, Anthropic reports the share of PRs receiving substantive review comments jumped from 16% to 54% after adopting the tool. No developer configuration is needed; admins enable it via Claude Code settings and install the GitHub App.

2m read timeFrom sdtimes.com
Post cover image

Sort: