Get instant AI feedback on every change — before it reaches your PR. No waiting for reviewer comments. Works with Claude Code or Cursor.
Works with your Claude Code or Cursor subscription.
A flexible AI agent orchestrator under your full control. Configure agents and rules to match your project goals.

Uses your existing Claude Code or Cursor subscription.
Each agent focuses on a specific area. Create unlimited custom agents.
security-scanSQL injection, XSS, exposed secretsbug-hunterNull checks, race conditions, edge casesperformance-checkMemory leaks, slow queries, N+1consistency-checkNaming, patterns, code stylegeneralReadability, complexity, best practices+ create unlimited custom agents with YAML
This page
Two steps to your first AI code review
Install diffray globally using npm, yarn, or pnpm.
Navigate to any Git repository and run a review. Uses your existing Claude Code configuration.
Everything you need for comprehensive code reviews, right in your terminal
Get instant feedback on your code changes without leaving your terminal or switching context.
Detect security vulnerabilities, OWASP Top 10 issues, and potential exploits in your code.
Specialized AI agents for security, performance, architecture, and code quality.
Works with any Git repository. Review staged changes, branches, or specific commits.
Define your own review rules and guidelines using simple YAML configuration.
Easy to integrate into your CI/CD pipeline for automated code reviews.
Simple, intuitive CLI commands for every use case
diffray reviewReview current staged changes or working directory
diffray review --branch feature-branchReview all changes in a specific branch
diffray review --commit HEAD~3..HEADReview a range of commits
diffray configConfigure API keys and preferences
diffray initInitialize diffray configuration in your project
diffray agentsList and configure review agents
We believe code review should happen where you work. Not in a browser, not in a separate app, but right in your terminal alongside your code.
The diffray CLI brings the full power of our multi-agent AI review system directly to your command line, making it faster and more integrated into your workflow.