Are you a Vibe Coder
or a Jive Coder?
AI can write code, but can you debug it? stresst introduces realistic bugs into real codebases to test if developers truly understand what they're shipping.
What is stresst?
🤖 AI-Powered Bug Generation
Uses Claude AI to introduce subtle, realistic bugs that developers actually make — off-by-one errors, null pointer issues, async/await mistakes, logic inversions, and more.
📊 Configurable Difficulty
Choose your stress level: Low (1-2 bugs), Medium (2-3 bugs), or High (3-5 devious bugs). Optionally focus on specific areas like async/await or null handling.
🔔 Realistic Bug Reports
Receive user-friendly symptom descriptions like "The posts are showing up blank" or "The app crashes when I select an item" — just like real production issues.
📋 Share & Collaborate
Copy bug reports to share with colleagues via email or Slack. Perfect for team training sessions, code reviews, or interview scenarios.
Why I Built This
AI code generation is everywhere — Cursor, Copilot, ChatGPT. Developers are shipping code faster than ever. But here's the uncomfortable question: do they actually understand what they're shipping?
When AI-generated code breaks (and it will), can your team debug it? Or are they just prompting until the errors go away?
stresst provides a challenging and engaging way for developers to learn how to fix bugs and diagnose code. It's essential for anyone using code-gen tools — because when things go wrong, it's imperative to be able to get things back on track.
How It Works
Connect
Sign in with GitHub to access your repositories
Select
Choose a repo, branch, and commit to stress
Stress
AI introduces realistic bugs — share & debug!
The Stress Process
When you hit "Stress", our AI analyzes your codebase, identifies the most interesting files to inject bugs into, and creates a new branch with realistic, hard-to-spot issues. You'll get a bug report describing symptoms — just like a real user complaint — without revealing where the bugs are.
Scoring System
Once you've fixed the bugs and committed your changes, return to stresst to check your score. The system automatically detects your start and completion commits and grades you based on how quickly you fixed all the bugs.
| Difficulty | A | B | C | D |
|---|---|---|---|---|
| 🌱 Easy | 0-5 min | 5-10 min | 10-15 min | 15+ min |
| 🔥 Medium | 0-7 min | 7-11 min | 11-15 min | 15+ min |
| 💀 Hard | 0-10 min | 10-15 min | 15-20 min | 20+ min |
Choose Your Codebase
Your Own Codebase
Use any repository you have access to. Great for testing your team's familiarity with your actual production code.
Our Practice Codebases
Fork one of our public "dummy" repositories designed specifically for practice. No risk to your real code, and they're structured to be great learning material.
Available Practice Repositories
Getting Started
Prerequisites
- ✓ Node.js 18+
- ✓ A GitHub account
- ✓An LLM provider — Anthropic (recommended), OpenAI, or a local LLMAnthropic is most robustly tested. Local requires a capable machine.
Quick Setup
$ git clone https://github.com/brenoneill/stresst.git$ cd stresst && npm install$ npm run devSee the README for full environment setup including GitHub OAuth and Anthropic API configuration.