This is a submission for the DEV Challenges: April Fools' Day
What I Built
An AI code reviewer with 4 distinct personas โ each more brutal than the last. Paste your code and prepare for emotional damage. No constructive feedback. No silver linings. Just pure, unfiltered roasting.
"Your variable naming convention looks like a cat walked across your keyboard."
๐ Live Demo
Code
arnoldwender
/
ai-code-reviewer-that-only-complains
An AI code reviewer that only gives harsh, sarcastic complaints โ never praise. Built for the DEV April Fools Challenge 2026.
๐ก AI Code Reviewer That Only Complains
An AI code reviewer that never has anything nice to say. Ever.
Built by Arnold Wender
What is this?
Paste your code and prepare for emotional damage. This AI code reviewer has 4 distinct personas โ each more brutal than the last โ and they will never, under any circumstances, say something positive about your code. No constructive feedback. No silver linings. Just pure, unfiltered roasting.
"Your variable naming convention looks like a cat walked across your keyboard." โ The Reviewer
Features
- 4 Reviewer Personas โ Each with their own flavor of cruelty (from passive-aggressive to unhinged)
- Severity Meter โ Watch the complaint level rise from "Mildly Disappointed" to "Existential Crisis"
- Line-by-Line Roasts โ Every single line of your code gets individually destroyed
- Achievements System โ Unlock badges for surviving the worst reviews
- Share Report Card โ Generate a shareable report card ofโฆ
How I Built It
Built from the passenger seat on a 5-hour road trip to the Ostsee โ phone only, no laptop. My wife drove. My daughters sang their favorite songs on repeat. I shipped 10 April Fools projects. Sending 2 publicly because some jokes are better kept private.
React 18 + TypeScript + Tailwind + Framer Motion. The reviewer personas write themselves when you're running on highway coffee and cartoon soundtracks.
Prize Category
Funniest Project โ Because nothing is funnier than a tool that exists solely to tell you your code is terrible. The Severity Meter goes from "Mildly Disappointed" all the way to "Existential Crisis." The fake CI/CD pipeline always fails. The Grafana dashboard shows your declining performance over time. It's accurate.
What I Learned
AI reviewers and real reviewers have more in common than I'd like to admit.

Top comments (0)