Anthropic announced Code Review on March 9, 2026, a new feature for Claude Code that deploys a team of specialized AI agents to automatically review pull requests. The system has been running internally at Anthropic on nearly every pull request for several weeks, addressing what the company identifies as a critical bottleneck in modern software development.
Multi-Agent System Reviews Entire Codebases in Parallel
When a pull request opens, Claude Code Review spins up multiple specialized agents that work in parallel to analyze code. Unlike traditional code review tools that focus on changed files, the system reviews entire projects to catch logic bugs across the codebase. According to Anthropic employees, the system produces audit-ready reports and can integrate with CI/CD gates.
The feature costs approximately $15-25 per pull request depending on complexity, and is currently available in research preview for Team and Enterprise plan customers.
Internal Testing Shows Dramatic Increase in Review Coverage
Anthropic built the system to address an internal problem: code output per engineer increased 200% this year, making human review the limiting factor in development workflows. Before implementing Code Review, only 16% of pull requests received substantive review. After deployment, 54% now receive deep review.
Employees testing the system report that it catches bugs human reviewers often miss. The system does not automatically approve pull requests, but provides detailed analysis to help developers maintain code quality as AI tools generate code faster.
Addresses AI-Generated Code Review Bottleneck
The announcement comes as development teams face a new challenge: as AI coding assistants enable developers to write code faster, human review capacity has become the constraint. Code Review aims to scale review capabilities to match the increased velocity of AI-assisted development.
The feature is available now for Claude Code Team and Enterprise customers through Anthropic's platform.
Key Takeaways
- Anthropic launched Code Review for Claude Code on March 9, 2026, using multi-agent AI to automatically review pull requests
- The system costs $15-25 per PR and reviews entire projects, not just changed files
- Internal testing at Anthropic showed review coverage increased from 16% to 54% of pull requests
- Code output per Anthropic engineer increased 200% this year, making human review a bottleneck
- Available in research preview for Claude Code Team and Enterprise customers