Skip to content

broskell/Roast-My-Project

BranchesTags

Repository files navigation

Roast My Project

AI feedback for builders who want more than a generic thumbs-up.

Roast My Project lets users submit engineering projects, screenshots, resumes, and startup ideas, then receive structured AI feedback through multiple review modes. The upgraded app now runs on Next.js 16, React 19, Payload CMS 3, Gemini, Groq fallback routing, Cloudinary uploads, Firebase/Twilio authentication, and a new grounded idea-research workflow.

Roast My Project Screenshot

What It Does

  • Project roasts from screenshots, descriptions, GitHub links, and live URLs.
  • Multiple review modes: Funny Roast, Brutal Roast, Recruiter Review, Senior Developer Review, and Investor Review.
  • Resume review with PDF validation, AI scoring, and concrete improvement suggestions.
  • Idea Intelligence reports for startup concepts, including competitors, market risks, MVP roadmap, VC questions, and source-backed evidence.
  • User dashboard with projects, reviews, resume results, idea reports, and analytics.
  • Payload admin dashboard for managing users, prompts, reviews, uploaded records, and AI debug logs.
  • Google OAuth and phone OTP login with JWT-protected API routes.
  • Gemini-first AI routing with Groq fallback for quota and rate-limit failures.

Tech Stack

Layer Technology
App framework Next.js 16.2.7, App Router, React 19.2.4, TypeScript
Styling Tailwind CSS v4, PostCSS
CMS and database Payload CMS 3.85, MongoDB via @payloadcms/db-mongodb
AI providers Google Gemini via @google/generative-ai, Groq OpenAI-compatible chat completions
Uploads and media Cloudinary, Sharp image diagnostics
Authentication Firebase client/admin auth, Twilio OTP, app JWTs
Charts and UI Recharts, Lucide React, canvas-confetti
Tooling ESLint 9, TypeScript 5, tsx diagnostics

Main Workflows

Project Review

flowchart TD
  A["User submits project"] --> B["Upload screenshot to Cloudinary"]
  B --> C["Create project in Payload/MongoDB"]
  C --> D["POST /api/ai/roast"]
  D --> E["Verify JWT and project ownership"]
  E --> F["Download and validate screenshot with Sharp"]
  F --> G["Load selected prompt from Payload"]
  G --> H["Call Gemini through AI router"]
  H --> I{"Gemini quota or rate limit?"}
  I -->|"No"| J["Parse JSON response"]
  I -->|"Yes"| K["Fallback to Groq"]
  K --> J
  J --> L["Validate review schema"]
  L --> M["Save review and detected category"]
  M --> N["Return score, roast, review, strengths, weaknesses, suggestions"]
Loading

Resume Review

flowchart TD
  A["User uploads PDF resume"] --> B["Validate file size and PDF magic bytes"]
  B --> C["Load Resume Review prompt"]
  C --> D["Call AI router with PDF media"]
  D --> E{"Fallback provider needed?"}
  E -->|"Gemini works"| F["Gemini analyzes PDF"]
  E -->|"Groq fallback"| G["Extract PDF text and analyze with Groq"]
  F --> H["Validate score, roast, suggestions"]
  G --> H
  H --> I["Save resume review"]
Loading

Idea Intelligence

flowchart TD
  A["User enters startup idea"] --> B["POST /api/idea-research"]
  B --> C["Verify JWT"]
  C --> D["Compile DuckDuckGo HTML search evidence"]
  D --> E["Load Idea Research prompt"]
  E --> F["Call AI router"]
  F --> G["Validate structured report schema"]
  G --> H["Save idea report"]
  H --> I["Render tabs: overview, competitors, outcomes, strategy, roadmap, sources"]
Loading

Routes

User Pages

Route Purpose
/ Landing and entry experience
/login Google and phone login
/verify-otp OTP confirmation flow
/dashboard User workspace and stats
/submit Project submission
/resume Resume upload and review
/idea-research Startup idea validation reports
/profile User profile
/results/project/[id] Project review results
/results/resume/[id] Resume review results
/admin Payload CMS admin dashboard

API Routes

Route Purpose
/api/auth/firebase-login Firebase ID token verification and app JWT issue
/api/auth/send-otp Twilio or mock OTP generation
/api/auth/verify-otp OTP validation and app JWT issue
/api/upload Cloudinary upload endpoint
/api/projects Project list and creation
/api/projects/[id] Project detail, update, and delete
/api/ai/roast Project screenshot analysis
/api/ai/resume-review Resume PDF analysis
/api/reviews Review listing
/api/reviews/[id] Review detail
/api/idea-research Idea report list and creation
/api/idea-research/[id] Idea report detail with ownership check
/api/dashboard/stats User dashboard metrics
/api/admin/stats Payload admin analytics
/api/debug/roast-health Roast pipeline diagnostics

Data Model

Payload CMS manages the application collections:

  • admins: Payload dashboard users.
  • users: App users authenticated by Google or phone OTP.
  • otp_verifications: OTP records with a MongoDB TTL index.
  • projects: Submitted project metadata and Cloudinary screenshot references.
  • reviews: AI-generated project reviews.
  • resumes: Resume review results and uploaded PDF metadata.
  • prompts: Editable prompt templates for every review mode.
  • gemini_debug_logs: AI request diagnostics, provider used, model used, raw output, parse status, and errors.
  • idea_reports: Grounded startup research reports with market, risk, roadmap, and source data.

Default prompts are seeded on Payload initialization when missing.

Environment Variables

Create .env in the project root.

# Database
MONGODB_URI=mongodb://127.0.0.1:27017/roast-my-project
# or
DATABASE_URI=mongodb://127.0.0.1:27017/roast-my-project

# Payload
PAYLOAD_SECRET=replace_with_a_long_secret
PAYLOAD_PUBLIC_SERVER_URL=http://localhost:3000

# App auth
JWT_SECRET=replace_with_a_long_jwt_secret

# AI providers
GEMINI_API_KEY=your_gemini_api_key
GROQ_API_KEY=your_groq_api_key
ENABLE_GEMINI_DEBUG=true

# Cloudinary
CLOUDINARY_CLOUD_NAME=your_cloud_name
CLOUDINARY_API_KEY=your_api_key
CLOUDINARY_API_SECRET=your_api_secret

# Twilio OTP
TWILIO_ACCOUNT_SID=your_account_sid
TWILIO_AUTH_TOKEN=your_auth_token
TWILIO_PHONE_NUMBER=your_twilio_number

# Firebase client config
NEXT_PUBLIC_FIREBASE_API_KEY=your_firebase_api_key
NEXT_PUBLIC_FIREBASE_AUTH_DOMAIN=your_project.firebaseapp.com
NEXT_PUBLIC_FIREBASE_PROJECT_ID=your_project_id
NEXT_PUBLIC_FIREBASE_STORAGE_BUCKET=your_project.appspot.com
NEXT_PUBLIC_FIREBASE_MESSAGING_SENDER_ID=your_sender_id
NEXT_PUBLIC_FIREBASE_APP_ID=your_app_id
NEXT_PUBLIC_FIREBASE_MEASUREMENT_ID=your_measurement_id

MONGODB_URI or DATABASE_URI is required. The runtime env validator also requires PAYLOAD_SECRET, GEMINI_API_KEY, GROQ_API_KEY, and the Cloudinary keys. If Twilio is not configured, the OTP route can fall back to local mock behavior for development.

Getting Started

Install dependencies:

npm install

Run the development server:

npm run dev

Open http://localhost:3000.

Open the Payload admin dashboard at http://localhost:3000/admin. On first setup, create the admin user; prompt templates and the OTP TTL index are initialized during Payload startup.

Scripts

Command Description
npm run dev Start the Next.js development server
npm run build Build for production
npm start Start the production server
npm run lint Run ESLint
npm run typecheck Run TypeScript without emitting files
npm run test:gemini-models Check available Gemini models using .env

Project Structure

.
├── payload.config.ts
├── next.config.ts
├── package.json
├── src
│   ├── app
│   │   ├── (app)
│   │   │   ├── api
│   │   │   ├── dashboard
│   │   │   ├── idea-research
│   │   │   ├── login
│   │   │   ├── profile
│   │   │   ├── resume
│   │   │   ├── results
│   │   │   ├── submit
│   │   │   └── verify-otp
│   │   └── (payload)
│   │       ├── admin
│   │       └── api
│   ├── collections
│   ├── components
│   ├── config
│   ├── context
│   ├── lib
│   ├── types
│   └── utils
└── public

Notes For Development

  • This repository uses the upgraded Next.js package installed in node_modules. Check node_modules/next/dist/docs/ before changing framework-specific code.
  • Keep prompt behavior editable through the prompts collection instead of hardcoding new review instructions in route handlers.
  • AI responses are treated as untrusted JSON and validated in src/types/ai.ts before saving.
  • The AI router prefers Gemini and uses Groq only for quota or rate-limit fallback paths.
  • Idea research uses app-side search evidence from DuckDuckGo HTML results, then asks the AI model to produce a structured report from that evidence.

About

Get brutal, funny, or professional AI feedback on your engineering projects and resumes. Boost your designs, improve your system architecture, and prepare for interviews using Gemini AI.

roast-my-project.vercel.app

Resources

Readme
Activity

Stars

0 stars

Watchers

0 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

 
 
 

Contributors

Languages