Code reviews that
hurt your feelings.
Connect your repository. Open a Pull Request. Get roasted by an AI trained to emulate a cynical Senior Engineer who hates spaghetti code.
const handleSubmit = async (data) => {
if (data.password.length < 6) {
alert('Bad password');
return false;
}
await db.users.create(data);
window.location.reload();
}
Score: 2/10
🚨 CRITICAL
Line 145: Plaintext password handling detected. Do you even read the news?
⚠️ VIOLATIONS
Line 150: window.location.reload() is not a state management strategy.
[REQUESTED CHANGES] Mention @roastbot once you've fixed this mess.
Terrorizing engineering teams at
Why use standard linting
when you can have psychological warfare?
Most tools catch syntax errors. RoastBot catches skill issues. It enforces best practices through rigorous shame-based reinforcement learning.
Merge Protection
RoastBot uses the GitHub Reviews API to explicitly reject PRs with critical violations, effectively blocking merges.
Conversational Roasts
The bot listens. Mention @roastbot in your PR comments to argue, apologize, or ask for refactoring advice.
Instant Autopsy
Manual roast mode lets you pick any file from your repo for an immediate, detailed breakdown of its failures.
Occasionally Nice?
If you actually write clean, well-documented, tested code, RoastBot might offer a rare compliment. (Note: This feature is currently in beta and rarely triggered).
"This PR... doesn't entirely suck. Good job."
Zero Config
Install the GitHub app. Grant permissions. Prepare your ego.