GitOps Guardian automatically reviews pull requests in GitOps repositories to identify security risks and compliance violations before code reaches production. It uses a multi-agent architecture combining AI-powered security analysis with rule-based Kubernetes best practices validation.

Key Features

Multi-Agent Architecture: Four specialized agents (Scanner, Security, Compliance, Ensemble)
Security Detection: Identifies hardcoded secrets, privileged containers, missing security contexts, insecure RBAC
Compliance Validation: Checks for :latest tags, missing resource limits, required labels
Risk Scoring: 0-1 risk score with SAFE/REVIEW/RISKY classification
Interactive Dashboard: Real-time Gradio web interface
Cost Tracking: Monitor OpenAI API costs (~$0.0004 per PR)
Export & Comments: Download results as JSON/CSV, post reviews to GitHub PRs
Historical Trends: Track risk scores over time

Requirements

Python 3.10+
GitHub Personal Access Token (repo read permissions)
OpenAI API Key
GitOps repositories with open pull requests

Quick Start

Install dependencies:

pip install -r requirements.txt

Configure environment:

cp .env.example .env
# Edit .env with your tokens and repository list

Run the application:

python app.py

Open browser at http://localhost:7860

Configuration

Create .env file with:

GITHUB_TOKEN=ghp_your_token
OPENAI_API_KEY=sk_your_key
GITOPS_REPOS=owner/repo1,owner/repo2

Project Structure

gitops-guardian/
├── models.py          # Pydantic data models
├── agents.py          # Multi-agent system (4 agents)
├── app.py             # Gradio UI & orchestration
├── requirements.txt   # Dependencies
└── .env.example       # Configuration template

How It Works

Scanner Agent fetches open PRs from GitHub repositories
Security Agent analyzes diffs using OpenAI GPT-4o-mini for security issues
Compliance Agent validates Kubernetes manifests against best practices
Ensemble Agent combines scores (60% security, 40% compliance) into overall risk
Results displayed in dashboard with recommendations

Week 8 Concepts Applied

Multi-agent system with base Agent class and color-coded logging
Ensemble pattern with weighted scoring
External API integration (GitHub + OpenAI)
Pydantic models for type safety
Gradio web interface
Persistent memory (JSON-based)
Real-world DevOps use case

Author

Salah - Week 8: Multi-Agent Systems - LLM Engineering Bootcamp 2025