CodeSentinel
AI-Powered Code Review System
CodeSentinel is an automated code review system that triggers on GitHub commits and pull requests. It runs a multi-agent AI analysis pipeline with specialized agents that are dynamically activated based on auto-detected project context.
Key Objectives
Data Sovereignty
Your code is processed securely with strict access controls and data protection.
Multi-Agent AI
6 specialized agents working together for comprehensive analysis.
Context Awareness
Automatic detection of frameworks, languages, and dependencies.
GitHub Integration
Seamless PR comments, commit status, and inline suggestions.
Cost Tracking
Monitor API usage and optimize your AI spending.
Real-time Reviews
Reviews complete in seconds, not hours or days.
AI Models
CodeSentinel is powered by Anthropic's Claude AI models:
Fast and efficient for standard reviews. Used on Hobby plan.
Maximum intelligence for complex analysis. Available on Pro and Enterprise plans.
How We Differentiate
What makes CodeSentinel unique
Competitive Comparison
| Feature | CodeRabbit | SonarQube | GitHub Copilot | CodeSentinel |
|---|---|---|---|---|
| Multi-Agent System | ||||
| Context Aware | ||||
| Pricing Model | Per seat | Per LOC | Per seat | Flat rate |
Privacy First
CodeSentinel is designed with data privacy at its core. Your source code is processed securely, ensuring complete data sovereignty and compliance with strict security requirements.
Multi-Agent Architecture
Instead of one general-purpose AI, we deploy 6 specialized agents that excel in their domains. This results in more accurate findings and fewer false positives than single-model approaches.
Automatic Context Discovery
CodeSentinel automatically detects your frameworks (Next.js, Django, Rails), ORMs (Prisma, TypeORM), and patterns without manual configuration. Reviews are tailored to your specific stack.
Predictable Pricing
No per-seat or per-line-of-code pricing. Our flat-rate plans give you unlimited reviews on the Pro tier, so you can scale your team without worrying about costs.
Architecture
How CodeSentinel works under the hood
System Overview
GitHub Integration
Webhooks receive PR and commit events. The GitHub App posts inline comments, review summaries, and commit statuses directly to your PRs.
Job Processing
A job queue manages review requests with retry logic and rate limiting. Jobs can run immediately or in background mode for large repositories.
AI Orchestrator
The orchestrator dispatches code to specialized agents in parallel, aggregates findings, deduplicates issues, and formats the final review.
Data Layer
PostgreSQL (via Prisma) stores repositories, reviews, findings, learned rules, and cost tracking data. All data is stored securely in the database.
Project Structure
codesentinel/ ├── src/ │ ├── app/ # Next.js App Router │ │ ├── api/ # API routes │ │ │ ├── webhooks/ # GitHub webhooks │ │ │ └── repositories/ # Repository endpoints │ │ ├── (protected)/ # Auth-required pages │ │ └── login/ # Authentication │ ├── lib/ │ │ ├── agents/ # AI review agents │ │ │ ├── orchestrator.ts │ │ │ ├── security-agent.ts │ │ │ └── performance-agent.ts │ │ ├── context/ # Context management │ │ ├── github/ # GitHub client │ │ ├── jobs/ # Job processing │ │ └── auth/ # NextAuth.js │ └── components/ # React components └── prisma/ # Database schema
AI Agents
6 specialized agents for comprehensive analysis
The Bouncer
Security Agent
Detects OWASP Top 10 vulnerabilities, injection attacks, authentication flaws, and exposed secrets.
The Optimizer
Performance Agent
Identifies N+1 queries, memory leaks, blocking operations, and inefficient algorithms.
The Pedant
Style Agent
Enforces consistent naming conventions, patterns, and best practices across your codebase.
The Architect
Architecture Agent
Analyzes dependencies, detects circular imports, and suggests structural improvements.
The Librarian
Documentation Agent
Checks for missing documentation, outdated comments, and incomplete API specs.
The Commander
Orchestrator
Coordinates all agents, aggregates findings, and manages the review pipeline.
Changelog
Recent updates and features
Phosphor Icons Migration
Migrated from Lucide to Phosphor Icons for a more consistent and comprehensive icon system.
Theme Toggle & Language Selector
Added dark/light mode toggle and multi-language support (EN/IT) to the landing page.
Advanced Security Infrastructure
Implemented comprehensive security features including rate limiting, input validation, and audit logging.
Sentinel Fortress Theme
Complete UI redesign with the new fortress-inspired theme featuring gold accents and glass panels.
Context Spider Web System
Unified context management with interactive dependency graph visualization and smart file sampling.
AI-Powered Context Generation
Automatic repository context generation using Claude AI for intelligent code understanding.
Auto-Review Toggle
Added ability to pause/resume AI reviews per repository with previous commit analysis.
Token Management & Caching
Improved API token management and implemented intelligent caching for faster reviews.