Anthropic Launches Multi-Agent Code Review System for Claude, Tripling Substantive PR Feedback at $15–25 Per Review

Anthropic launches a multi-agent AI code review system for Claude that tripled substantive PR feedback from 16% to 54% internally, catching critical bugs before production — now available in research preview for Team and Enterprise users at $15–25 per review.

• Anthropic is launching Code Review for Claude Code, a multi-agent PR review system now available in research preview for Team and Enterprise plans, dispatching teams of agents to catch bugs in parallel, verify findings, and rank issues by severity.
• Internal data shows the system is dramatically improving review coverage at Anthropic, with substantive review comments jumping from 16% to 54% of PRs, including catching a critical authentication-breaking one-line change before it reached production.
• Reviews are billed on token usage and typically cost $15–25 per PR depending on size and complexity, with admins able to set monthly spending caps, control which repositories are reviewed, and track usage through an analytics dashboard.