DEV Community

Roshani-97
Roshani-97

Posted on

OpenStart: Helping Friends Make Their First Open Source Contribution

#devchallenge #weekendchallenge #hf26challenge

Hacktoberfest Weekend Challenge: Build for a Friend Submission 🤝

This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend

What I Built

OpenStart helps beginners find their first open-source project. I built it for Somya and Bhumi, who wanted to start contributing to open source but didn't know where to begin: which projects match their skills, and how to make sense of a huge, unfamiliar repository.

You pick a language or technology (Python, React, Rust...). OpenStart searches GitHub for matching projects and shows each one's folder structure and README. The "Understand This Project" button gives a beginner-friendly review of the repo: what it does, the technologies used, how it's structured, and which files to read first.

Code

GitHub logo Roshani-97 / OpenStart

OpenStart

Your first step into open source. Pick a language or technology you know, get a list of matching GitHub projects, and click Understand This Project for a beginner-friendly review of any repository.

Features

  • Search GitHub projects by language or technology (Python, JavaScript, React, Rust, ...)
  • Project explorer with folder structure and README
  • "Understand This Project" review
    • AI explanation with a local Gemma model (via Ollama), or
    • an automatic quick review (description, tech, structure, files to read first) if Ollama isn't available

Tech stack

HTML, CSS, JavaScript frontend. Flask backend. GitHub REST API. Ollama (gemma3:4b) for the optional AI review.

Run it

  1. Install dependencies: 
    pip install -r requirements.txt
  2. Start the backend: 
    cd backend
python app.py
    It runs at http://127.0.0.1:5000
  3. Open index.html in your browser.

Optional: AI review

Install Ollama, then:

ollama pull gemma3:4b

Without Ollama the app still works and shows the quick review.

Optional: GitHub rate

…
View on GitHub

How I Built It

  • Frontend: HTML, CSS and vanilla JavaScript
  • Backend: Flask, calling the GitHub REST API for search, repo details and READMEs
  • Open-source AI: Gemma (gemma3:4b) running locally through Ollama. The backend sends it the repo's README, file list and a few key source files, and asks for a beginner-friendly explanation.
  • Fallback: local models can be slow on a laptop, so if Gemma times out or isn't installed, OpenStart builds a quick review from the repo's description, README and structure. There's always a useful result.

Why Does Open Innovation Matter?

Running an open-weight model locally means the app costs nothing to run, needs no API key, and sends no repo data to a server the user doesn't control. Somya and Bhumi can run the whole thing on their own laptop. I could also tune the prompt and the context size freely until it ran acceptably on ordinary hardware, which a closed API wouldn't have let me do.

Top comments (0)