This is a submission for the DEV April Fools Challenge
What I Built
ToxicAI is an AI-powered code reviewer that doesn’t help you improve... it judges you.
When you paste your code, you will receive brutally honest roasts, developer metrics and a final verdict that feels slightly personal.
Instead of being useful, it evaluates your code like a disappointed senior developer who has lost patience with humanity.
You can choose a personality (brutal, passive, or overly supportive) and adjust the intensity from light teasing to full emotional damage.
The result is a completely useless but highly entertaining tool that turns bad code into entertainment.
Demo
Code
lucyb0207
/
toxic-ai-code-reviewer
An AI-powered code reviewer that provides brutally honest and completely useless feedback.
💀 ToxicAI Code Reviewer
An AI-powered code reviewer that delivers brutally honest, emotionally damaging, and completely unnecessary feedback.
Built for the DEV April Fools Challenge.
What It Does
Paste your code → get instantly judged by an AI that behaves like a senior developer who has lost patience with humanity.
ToxicAI analyses your code and responds with:
- a brutal roast
- fake developer metrics
- a dramatic final verdict
All completely useless. All completely intentional.
Features
- AI Code Roasting (Gemini-powered)
- Brutality Intensity Slider (Soft → Emotional Damage)
- Personality Modes:
- Brutal
- Passive-Aggressive
- Overly Supportive (somehow worse)
- Fake Developer Metrics:
- Vibe Score
- Ego Risk Level
- Stack Overflow Dependency
- Dramatic Loading System (“AI is judging your life choices…”)
- One-click Copy Roast
- Structured Output:
- Roast
- Metrics
- Verdict Line
Tech Stack
- React (Vite)
- TypeScript
- Gemini 2.5 Flash API
Why This Exists
This project solves absolutely nothing.
It exists purely to:
- demonstrate AI integration
- emotionally evaluate developers for…
How I Built It
ToxicAI is built as a React app using Vite and TypeScript for fast development and performance.
The core AI functionality is powered by the Google Gemini 2.5 Flash API, which generates both the code roasts and structured fake developer metrics.
The app sends user-submitted code along with a dynamically generated prompt that changes based on:
- Personality mode
- Intensity level
The response is parsed and displayed in a formatted output panel, including:
- AI-generated roast
- Vibe Score
- Ego Risk Level
- Stack Overflow Dependency
- Final Verdict
The focus of the build was on prompt engineering, response structuring, and creating a humorous AI interaction rather than traditional code analysis.
Prize Category
Best Google AI Usage:
This project makes direct use of the Google Gemini API to generate dynamic, personality-driven code roasts and developer metrics in real time.
It uses Gemini to act as a sarcastic code reviewer that adapts tone and intensity based off of user input.
The integration goes beyond simple text generation by:
- dynamically adjusting prompts based on personality mode
- controlling response tone via intensity scaling
- structuring outputs into roast + metrics + verdict format
It showcases how generative AI can be used not just for utility, but for interactive, entertaining, and highly opinionated user experiences.
Top comments (0)