Announcing diffray CLI
Free & Open-Source
Run AI-powered multi-agent code reviews directly from your terminal. No account required. Powered by Claude Code or Cursor agents.
Today we're releasing diffray CLI — a free, open-source command-line tool that brings the power of multi-agent AI code review to every developer.
Install it with a single command. Run it in any Git repository. Get actionable feedback in seconds — all without creating an account or sharing your code with any cloud service.
Get Started in 30 Seconds
Uses your existing Claude Code or Cursor CLI configuration. No additional setup needed.
Why Open Source?
We believe every developer deserves access to intelligent code review — not just teams with enterprise budgets. The CLI brings diffray's multi-agent architecture to anyone with a terminal.
100% Free Forever
No trial periods. No credit card. No usage limits. Use it as much as you want, forever.
Your Code Stays Local
Code is processed locally by your AI agent. Nothing is sent to our servers. Complete privacy by design.
Fully Open Source
MIT licensed. Inspect the code, contribute features, or fork it for your own needs.
Community Driven
Share your custom agents and rules. Learn from what others have built. Grow together.
Powered by Claude Code or Cursor
diffray CLI doesn't include its own AI model — it leverages the AI agents you already have installed. This means you get the full power of Claude or Cursor's intelligence for code understanding.
Supported AI Executors
Claude Code CLI
Default executor. Uses your existing Claude Code authentication.
Cursor Agent CLI
Alternative executor. Use your existing Cursor subscription.
The CLI orchestrates the review pipeline — cloning the diff, selecting relevant rules, running specialized agents, deduplicating findings, and presenting actionable results. The AI does the heavy thinking.
Multi-Agent Review Pipeline
Unlike single-prompt AI tools, diffray CLI runs multiple specialized agents — each focused on a specific domain. This is the same architecture that powers our cloud platform.
Security
OWASP, injection, auth flaws
Bug Hunter
Logic errors, edge cases
Performance
N+1, memory leaks, complexity
Quality
Code smells, readability
Architecture
SOLID, patterns, coupling
Consistency
Style, conventions, naming
Each agent only sees the code and context relevant to its domain. This focused approach reduces noise and increases the quality of findings.
Usage Examples
The CLI is designed for flexibility. Use it interactively during development or integrate it into your CI/CD pipeline.
Review uncommitted changes
Compare against a branch
Run specific agents
JSON output for CI/CD
Custom Agents & Rules
The CLI is fully extensible. Create your own agents and rules to match your team's specific needs and coding standards.
How It Works
- Create agents as Markdown files in
~/.diffray/agents/ - Define rules as Markdown files in
~/.diffray/rules/ - List available agents with
diffray agents - List available rules with
diffray rules
Your custom agents can enforce project-specific patterns, framework conventions, or internal coding standards. Rules can capture the lessons learned from past bugs and security incidents.
Who Should Use diffray CLI?
Solo developers and indie hackers
Get a second pair of eyes on every change — for free
Open-source maintainers
Catch issues in contributions before merging
Teams wanting to evaluate before committing
Try diffray's multi-agent approach in your workflow
Security-conscious developers
Keep code local while getting AI-powered security analysis
Start Using diffray CLI Today
One command to install. Zero configuration. Unlimited reviews.