AI-powered code review system. Point it at a GitHub URL or drop local files β parallel AI agents find bugs, security issues, and test gaps, then propose fixes.
Go to console.groq.com/keys β Create API Key β copy it.
Itβs free, no credit card required.
Create backend/.env:
GROQ_API_KEY=gsk_your_key_here
cd pr-guardian
python start.py
Thatβs it. Opens http://localhost:3000 automatically.
Press Ctrl+C to stop both servers.
Windows: double-click start.bat instead.
Paste GitHub URL OR drop local files
β
3β4 AI agents run in parallel (Code Β· Security Β· Tests Β· Performance)
β
Flappy Guardian game plays while you wait
β
Results appear after game ends (or skip)
β
Fix issues β download patched files (local) or commit to GitHub (PAT required)
| Provider | Cost | Set up |
|---|---|---|
| Groq β default | Free | GROQ_API_KEY=gsk_... in backend/.env |
| OpenAI fallback | Paid (~$0.05β0.20/analysis) | OPENAI_API_KEY=sk-... in backend/.env |
Groq runs Llama 3.1 70B β fast and free, no credit card needed.
Only needed to:
The app prompts for it only when you click βApply Fixβ β never upfront.
Get one: GitHub β Settings β Developer settings β Personal access tokens β Generate new (classic) with repo scope.
frontend/)backend/, set GROQ_API_KEY env var)β GitHub URL / β‘ Local files / β’ File path
β
FastAPI Backend
β asyncio.gather (parallel)
ββββββββββββ¬βββββββββββ¬βββββββββββ¬βββββββββββββ
Code Security Test Performance
Agent Agent Agent Agent
ββββββββββββ΄βββββββββββ΄βββββββββββ΄βββββββββββββ
β
Fix Agent + Regression Check
β
Risk Score + GitHub Report
β
Download fix (local) or Commit to branch (GitHub)
β# pr-guardianβ