YT · YOUTUBE// 32d agoPRODUCT UPDATE

Claude Code adds multi-agent PR review

Anthropic has added Code Review to Claude Code, a feature that runs multiple specialized agents on GitHub pull requests, checks candidate issues against actual code behavior, and posts inline findings with severity labels. It is in research preview for Teams and Enterprise, with reviews averaging about 20 minutes and pricing tied to PR size and complexity.

// ANALYSIS

This pushes Claude Code further from autocomplete helper toward full-stack software reviewer.

–Multiple agents analyze the diff and surrounding code in parallel, which gives Anthropic a stronger story on catching subtle logic bugs, regressions, and security issues
–A verification step filters candidates against real code behavior before posting comments, directly targeting the false-positive problem that makes many AI review bots noisy
–Findings are ranked by severity and do not block or approve PRs, so teams can add AI review without rewriting existing human review workflows
–CLAUDE.md and REVIEW.md let teams encode project-specific review rules, which makes the feature more useful than generic lint-style feedback
–Average review times and per-review token costs make this look aimed squarely at higher-value team workflows rather than casual solo use

// TAGS

claude-codeai-codingagentcode-reviewdevtool

DISCOVERED

32d ago

2026-03-10

PUBLISHED

32d ago

2026-03-10

RELEVANCE

9/ 10

AUTHOR

DIY Smart Code