Roast My Code - Find AI code smells before humans do

Roast My Code is an open-source AI tool that scans your repository for common code issues—especially patterns found in AI-generated code—and delivers brutally honest, funny, and actionable feedback. It combines static analysis with LLM-generated “roasts” to highlight TODOs, bad practices, and hidden problems, then outputs a shareable HTML report with a quality score. Built for developers who want fast, honest feedback without boring code reviews.

I built Roast My Code after noticing how much low-quality AI-generated code was slipping through reviews—things like TODOs, placeholder variables, and empty exception blocks. Traditional tools catch bugs, but they don’t call you out in a way that sticks. I wanted something that feels like a brutally honest senior engineer reviewing your code—but also makes you laugh. The project started as a simple static analyzer, but evolved into combining rule-based detection with LLM-generated roasts. Getting the balance right between humor and actually useful feedback was the hardest part. Would love feedback—especially on what “code smells” or patterns you think should be roasted more

Roast My Code - Find AI code smells before humans do

Replies