Welcome to diffray
AI-Powered Code Review for GitHub Pull Requests
diffray automatically reviews your pull requests using artificial intelligence and posts detailed feedback directly in GitHub.
How It Works
1. PR Opened
Automatic trigger when you create or update a pull request
2. AI Analyzes
Advanced AI reviews your changes in depth
3. Review Published
Detailed feedback posted as a comment in your PR
Specialized Review Agents
diffray uses a team of specialized AI agents, each expert in their domain:
| Agent | What It Catches |
|---|---|
| Security Expert | Vulnerabilities, injection attacks, exposed secrets, OWASP Top 10 |
| Performance Specialist | N+1 queries, memory leaks, bottlenecks, scalability issues |
| Bug Hunter | Null errors, race conditions, edge cases, logic bugs |
| Quality Guardian | Code smells, dead code, unclear logic, missing validation |
| Architecture Advisor | SOLID violations, tight coupling, design pattern issues |
| Consistency Checker | Duplicated code, pattern deviations, naming inconsistencies |
| Documentation Reviewer | Outdated docs, missing API docs, misleading comments |
| Test Analyst | Missing coverage, flaky tests, poor isolation |
| General Reviewer | Project conventions, git hygiene, cross-cutting concerns |
| SEO Expert | Meta tags, OpenGraph, structured data, search visibility |
| Refactoring Advisor | Code smells, SOLID violations, technical debt, complexity |
diffray comes with 1,300+ built-in rules covering security, performance, bugs, compliance, and more across 49 categories including languages, frameworks, and standards. Rules are automatically applied based on what files changed in your PR — TypeScript changes get TypeScript rules, Python gets Python rules, and so on.
Compliance coverage: SOC 2, HIPAA, GDPR, LGPD, PCI-DSS rules included out of the box.
See all supported technologies →
Smart & Adaptive
- Context-aware — only relevant rules run based on changed files
- Noise-free — intelligent deduplication removes repetitive findings
- High-confidence — validation pass filters out uncertain issues
- Codebase-aware — analyzes how changes affect existing code
Customizable
Add project-specific rules to enforce your team's standards:
# .diffray/rules/team-standards.yaml
rules:
- id: use_internal_logger
agent: quality
title: "Use internal logger"
match:
file_glob: ["src/**/*.ts"]
...
Key Features
AI Code Review
Get intelligent code reviews from advanced AI models that understand your codebase
Fast & Efficient
Automatic review triggers on PR open, update, or on-demand via labels
Automatic Triggers
Reviews start automatically when:
- Opening a new PR
- Pushing new commits to PR
- Reopening a PR
- Adding the
diffray-reviewlabel
Status Labels
Track review progress through automatic labels:
diffray-review-started- review starteddiffray-review-completed- review completeddiffray-review-failed- review failed
Secure Integration
GitHub App with controlled permissions - you choose which repositories to grant access
What You Need
- A GitHub account
- A repository where you want AI code reviews
- Admin rights to install GitHub App
Quick Start
Explore More
Ready to improve your code review process? Let's get started!