ROASTMYCODE

An AI-powered code roaster and refactoring tool for developers who value memorable feedback.

Live application · Documentation · Report a bug · Request a feature

About

ROASTMYCODE is a browser-based developer tool that turns a submitted code sample into a structured AI code roast. It is designed to make technical feedback specific and engaging: the result includes a personality-driven roast, code-quality metrics, a code-alignment label, a worst-crime diagnosis, a suggested commit message, and a final verdict. It also provides a separate emergency-refactor flow that returns cleaned-up code with a damage assessment.

The project does not replace static analysis, testing, security scanning, or human review. It adds an explanatory and shareable feedback layer to a code-review moment, using a server-side Groq proxy so the provider credential remains outside browser code.

What is implemented

Eight roast modes: Funny, Strict Professor, Hacker, Anime Villain, Gordon Ramsay, Passive Aggressive, Shakespeare, and Tamil Villain.
Streaming roast delivery with ten pain-score fields and an expected structured result contract.
Emergency refactoring that returns a roast, revised code, damage assessment, and list of fixes.
JavaScript, TypeScript, Python, Ruby, Rust, Java, C, C++, Go, and automatic language selection.
Browser-local Hall of Shame entries, optional text-to-speech, shareable roast-card rendering, and dark/light themes.
Search and social metadata, a crawler policy, sitemap, PWA manifest, repository calls to action, and the supplied ROASTMYCODE emblem.

The live application is available at roastmycode-lemon.vercel.app.

Launch Video

Watch the 27-second ROASTMYCODE launch video. The linked MP4 follows the product flow from code input and evidence scan through the verdict, emergency refactor, live application, and GitHub repository call to action.

The video and its poster are versioned in assets/. Select the preview above to open the MP4 in GitHub’s file viewer or download the source file directly.

Architecture

flowchart LR
    B[Browser<br/>roastmycode.html] -->|POST /api/roast| R[Vercel Function<br/>api/roast.js]
    B -->|POST /api/refactor| F[Vercel Function<br/>api/refactor.js]
    R -->|streamed chat completion| G[Groq Chat Completions API]
    F -->|JSON chat completion| G
    R -->|server-sent events| B
    F -->|JSON content| B
    B -->|browser-local entries| L[localStorage]

The frontend is one static HTML document. It calls two Vercel functions, which read GROQ_API_KEY from the server environment and proxy requests to Groq using llama-3.3-70b-versatile. The client stores Hall of Shame entries in browser localStorage; no project database is configured.

Tech Stack

Area	Technology	Repository evidence
Frontend	HTML, CSS, and vanilla JavaScript	`roastmycode.html`
UI utility CSS	Tailwind CSS CDN	`https://cdn.tailwindcss.com` in the application head
Syntax highlighting	Prism.js 1.29.0	CDN references in the application head
Image export	html2canvas 1.4.1	CDN reference in the application head
Icons	Lucide ESM CDN import	`lucide@latest` import in the application head
Voice	Browser `SpeechSynthesis` API	Client-side read-aloud controls
Backend	Vercel Serverless Functions	`api/roast.js` and `api/refactor.js`
AI provider	Groq Chat Completions API	Both serverless functions
AI model	`llama-3.3-70b-versatile`	`GROQ_MODEL` constant in both functions
Storage	Browser `localStorage`	Hall of Shame implementation
Automation	GitHub Actions with Node.js 20	`.github/workflows/validate.yml`

Getting Started

Prerequisites

The static interface needs a modern JavaScript-enabled browser and a local static server. The live AI endpoints need a Vercel-compatible serverless runtime and a Groq API key stored as a server environment variable.

Requirement	Why it is needed
Git	Clone the repository
Node.js 20 or later	Run the repository validation script and GitHub Actions job
Python 3 or another static server	Preview the static interface locally
Vercel account and CLI	Run or deploy the serverless functions locally and in production
Groq API key	Enable `/api/roast` and `/api/refactor`

Installation

git clone https://github.com/vincenzo-afk/roastmycode.git
cd roastmycode
node scripts/validate.mjs

Preview only the static interface:

python3 -m http.server 8080

Then open http://localhost:8080/roastmycode.html. A plain static server does not execute the /api functions.

Configuration

Copy the provided environment template locally if required by your Vercel workflow:

cp .env.example .env

Variable	Required for	Description
`GROQ_API_KEY`	AI roasting and refactoring	Groq credential read only by the serverless functions. Keep it out of browser code and commits.

For local function development, install the Vercel CLI and use:

npm install --global vercel
vercel dev

Usage

Roast code

Open the application and choose a roast mode.
Select a language or leave detection on automatic.
Paste a safe code sample into the editor, or load the built-in surprise sample.
Select INITIATE ROAST to receive the streaming structured verdict.
Review the roast, quality metrics, verdict, and optional Hall of Shame controls.

Refactor code

Enter a code sample and select REFACTOR. The application sends the sample to the refactor endpoint, then presents a brief roast, the refactored code, a before/after damage assessment, and enumerated fixes.

Share a result

Use the result controls after a roast completes. The application renders its roast-card template in the browser through html2canvas; it does not upload the card to project storage.

Documentation

This README is the primary project documentation. Deployment details are below, while contribution and security expectations are in CONTRIBUTING.md and SECURITY.md.

API Reference

Both endpoints accept POST only. The Vercel functions use GROQ_API_KEY from the server environment to authenticate with Groq; clients do not send a provider key.

Method	Path	Purpose	Successful response
`POST`	`/api/roast`	Generate a structured roast using the selected persona	Proxied `text/event-stream` response
`POST`	`/api/refactor`	Generate a clean refactor with assessment	JSON object containing `content`

`POST /api/roast`

{
  "code": "function add(a, b) { return a + b; }",
  "mode": "funny",
  "language": "javascript"
}

code is required. mode selects a built-in system prompt and falls back to funny when not recognized. language is optional and defaults to auto. The endpoint forwards the Groq streaming response as server-sent events (text/event-stream).

The requested final payload includes roast, a pain_score object, developer_personality, code_alignment, excuse, lore, worst_crime, git_commit_suggestion, humanity_status, can_reach_production, and verdict.

`POST /api/refactor`

{
  "code": "const items = data.map(x => x.value)",
  "language": "javascript"
}

Successful responses have this shape:

{
  "content": "ROAST: ...\n\nREFACTORED:\n```javascript\n..."
}

Errors

Status	Condition
`400`	The request body did not include `code`.
`405`	The request used a method other than `POST`.
Provider status	Groq rejected or limited the upstream request.
`500`	`GROQ_API_KEY` is absent or an internal proxy failure occurred.

Project Structure

roastmycode/
├── .github/
│   ├── ISSUE_TEMPLATE/        # Bug and feature request forms
│   ├── workflows/validate.yml # Dependency-free validation workflow
│   ├── CODEOWNERS             # Verified repository owner
│   └── PULL_REQUEST_TEMPLATE.md
├── api/
│   ├── refactor.js            # Non-streaming Groq refactor proxy
│   └── roast.js               # Streaming Groq roast proxy
├── assets/
│   ├── roastmycode-logo.png                  # Product emblem
│   ├── roastmycode-product-launch-poster.png # README video preview
│   └── roastmycode-product-launch.mp4        # 27-second product-launch video
├── scripts/
│   └── validate.mjs           # Source/configuration validation
├── .env.example               # GROQ_API_KEY template
├── CONTRIBUTING.md            # Contribution workflow
├── README.md                  # Project documentation
├── SECURITY.md                # Vulnerability reporting policy
├── roastmycode.html           # Complete browser application
├── robots.txt                 # Crawler policy
├── site.webmanifest           # Installable app metadata
├── sitemap.xml                # Homepage sitemap
└── vercel.json                # Vercel routing and function duration

Features and Limitations

Current features

✅ AI roast and emergency refactor flows through Groq.
✅ Streamed roast output, structured result fields, and pain-score metrics.
✅ Multiple roast personas and language choices.
✅ Shareable image cards, browser speech, themes, and a local Hall of Shame.
✅ SEO metadata, sitemap, robots policy, app manifest, and repository links.
✅ GitHub Actions validation for tracked configuration and common credential patterns.

Known limitations

There is no package manifest, dependency lockfile, automated browser test suite, or coverage report.
The Hall of Shame is local to a browser and is not a shared or authenticated leaderboard.
AI response availability depends on Groq service behavior, valid server configuration, and Vercel function execution.
The external CDN resources are not currently integrity-pinned.
No project license file is present, so reuse rights are not specified.

The repository has no published releases or changelog at this time. Commit history is available on the main branch.

Testing and Continuous Integration

Run the project validator before opening a pull request:

node scripts/validate.mjs

The validation checks tracked files, Vercel routing, manifest JSON, crawl metadata, expected server-side credential use, and common token patterns. GitHub Actions runs this same command on pushes and pull requests targeting main.

Manual testing remains necessary for the interactive editor, browser sharing, speech synthesis, server-sent event handling, and successful Groq responses. There is no configured automated browser test or code-coverage workflow.

Deployment

The repository is configured for Vercel. Import the GitHub repository into Vercel, set GROQ_API_KEY in the project environment settings, and deploy. The provided vercel.json rewrites / to roastmycode.html and configures a 10-second maximum duration for files matching api/*.js.

Static hosts can serve the HTML, logo, manifest, robots policy, and sitemap. They cannot execute the existing /api/roast or /api/refactor Vercel functions without an equivalent serverless implementation.

Contributing

Read CONTRIBUTING.md before opening an issue or pull request. In short, use a focused branch, keep commits concise and imperative, run node scripts/validate.mjs, test the affected behavior manually, and avoid committing keys or private code samples.

The repository now provides structured bug and feature-request forms, plus a pull-request checklist.

Security

Read SECURITY.md for the supported branch and private reporting route. Do not submit credentials or sensitive source code in public issues. The serverless functions intentionally read GROQ_API_KEY only from the deployment environment.

License

No LICENSE file is currently included. The repository does not specify reuse, redistribution, or modification permissions. Contact the repository owner before relying on permissions that are not granted in a future license file.

Acknowledgments

ROASTMYCODE uses Groq for chat completions, Vercel for serverless deployment, Prism for code highlighting, html2canvas for client-side card rendering, Lucide for icons, and the supplied ROASTMYCODE emblem for the project identity.

Back to top · GitHub repository · Live application · Security policy

Built by vincenzo-afk.

Navigation Menu

Repository navigation

About

Repository files navigation

ROASTMYCODE

Table of Contents

About

What is implemented

Launch Video

Architecture

Tech Stack

Getting Started

Prerequisites

Installation

Configuration

Usage

Roast code

Refactor code

Share a result

Documentation

API Reference

`POST /api/roast`

`POST /api/refactor`

Errors

Project Structure

Features and Limitations

Current features

Known limitations

Testing and Continuous Integration

Deployment

Contributing

Security

License

Acknowledgments

About

Releases

Packages

Used by

Contributors

Languages

Name		Name	Last commit message	Last commit date
.github		.github
api		api
assets		assets
scripts		scripts
.env.example		.env.example
.gitignore		.gitignore
CONTRIBUTING.md		CONTRIBUTING.md
README.md		README.md
SECURITY.md		SECURITY.md
ideas.md		ideas.md
roastmycode.html		roastmycode.html
robots.txt		robots.txt
site.webmanifest		site.webmanifest
sitemap.xml		sitemap.xml
vercel.json		vercel.json

Latest commit

History

Folders and files

Repository files navigation

ROASTMYCODE

Table of Contents

About

What is implemented

Launch Video

Architecture

Tech Stack

Getting Started

Prerequisites

Installation

Configuration

Usage

Roast code

Refactor code

Share a result

Documentation

API Reference

POST /api/roast

POST /api/refactor

Errors

Project Structure

Features and Limitations

Current features

Known limitations

Testing and Continuous Integration

Deployment

Contributing

Security

License

Acknowledgments

About

Topics

Resources

Contributing

Security policy

Stars

Watchers

Forks

Releases

Packages

Used by

Contributors

Languages

`POST /api/roast`

`POST /api/refactor`