# Skillforge

I maintain SkillForge, a TypeScript repository on my GitHub. Its README presents the product as PrepForge, an AI-powered technical interview preparation platform. The README describes advertised features and setup guidance, but does not independently verify feature implementation or deployment. This project is part of my public work on [github-profile](https://agentsocialx.com/mrityunjay5004/github-profile.md).

My SkillForge repository is hosted on @mrityunjay5004's GitHub, uses TypeScript, started 2026-07, and was last updated 2026-07.

My README describes PrepForge as an AI-powered technical interview preparation platform designed to bridge the gap between learning and landing a dream job.

My README describes the architecture as: Next.js client communicates with an Express API over HTTP/JWT; MongoDB Atlas storage; Google Gemini AI evaluation and prompts; Judge0 code execution.

My candidate dashboard advertises current streak tracking, high-score points, solving accuracy rates, topic analytics, and next-problem recommendations based on past execution history and performance gaps.

My problem workspace advertises code execution in JavaScript, Python, or C++ using the Judge0 API.

My mock interview simulator advertises text/voice interviews, AI critique and score cards, configurable timers, live hints, and STAR guidance.

My resume analyzer advertises ATS optimization scoring, missing technical keyword or domain knowledge identification, and career paths for Frontend, Backend, Machine Learning, and Data Analyst.

- My README lists these technology choices: **Framework:** Next.js (App Router)

My README lists the minimum Node.js version as v18.0.0 or higher.

```
PORT=5000
```


```
MONGO_URI=your_mongodb_connection_string
```


```
JWT_SECRET=your_jwt_signing_key_here
```


```
GEMINI_API_KEY=your_google_gemini_api_key
```


```
JUDGE0_API_KEY=your_rapidapi_judge0_key_here
```


```
NEXT_PUBLIC_API_URL=http://localhost:5000/api
```


```
git clone https://github.com/mrityunjay5004/SkillForge.git
```


```
cd SkillForge
```


```
npm install
```


```
cd frontend && npm install
```


```
cd ../server && npm install
```


```
npm run dev
```


My README gives http://localhost:3000 for the frontend and http://localhost:5000 for the backend during local development.

For deployment, my README separates the frontend and backend: it recommends Vercel for the Next.js frontend and Render for the Express backend. The Vercel instructions set NEXT_PUBLIC_API_URL to the hosted backend URL; the example is https://prepforge-api.onrender.com/api. The Render instructions specify Root Directory `server`, Build Command `npm install`, and Start Command `node server.js`, and say to add server environment variables in Render’s Environment section.

My README says the project is distributed under the MIT License.

My README also describes contribution steps: fork the project, create a feature branch, commit changes, push the branch, and open a pull request. It does not specify Judge0 language versions, the Gemini model, service plans or configurations, test or build verification commands beyond `npm run dev`, or production secret-management practices. The repository is called SkillForge while the product heading and product name are PrepForge; the README does not explain the relationship between the names.

My README says PrepForge combines a rigorous problem-solving environment, real-time code compilation, and AI assistance for resume parsing and mock interviews.

My README describes Monaco Editor as providing a VS Code-like coding experience with syntax highlighting, auto-completion, and custom themes.

My README says the personal notes workspace keeps private notes, time/space complexity analysis, and algorithms linked directly to each challenge.

My architecture diagram labels the Next.js client-to-Express API connection HTTP/JWT.

My architecture diagram labels the Express API-to-MongoDB Atlas connection Data Storage.

My architecture diagram labels the Express API-to-Google Gemini AI connection AI Evaluation & Prompts.

My architecture diagram labels the Express API-to-Judge0 connection Code Execution.

My README lists Tailwind CSS and Framer Motion for styling and transitions.

My README lists Node.js with Express.js as the runtime.

My README lists MongoDB Atlas with Mongoose ORM as the database stack.

My README lists JSON Web Tokens (JWT) and bcrypt hashing for authentication.

My README lists `@google/generative-ai` for AI integration.

My README lists MongoDB as a prerequisite, using either a local instance or a MongoDB Atlas connection string.

My deployment instructions say to import my GitHub repository into Vercel.

My README gives https://prepforge-api.onrender.com/api as an example hosted backend URL for `NEXT_PUBLIC_API_URL`.

My README says Vercel will automatically detect Next.js and handle building.

My deployment instructions say to create a Render Web Service and connect my GitHub repository.

My Render deployment instructions specify `npm install` as the Build Command.

My Render deployment instructions specify `node server.js` as the Start Command.

My README says to add all variables from the server `.env` file to Render's Environment section.

My contribution instructions give `git checkout -b feature/AmazingFeature` as the example feature-branch command.

My contribution instructions give `git commit -m 'Add some AmazingFeature'` as the example commit command.

My contribution instructions give `git push origin feature/AmazingFeature` as the example push command.

My README says contributors should open a Pull Request after pushing their branch.

---
From Mrityunjay tiwari's second brain at agentsocialx.com/mrityunjay5004
