Skip to content

groundhog-21/digitalocean_gradient_ai_hackathon

Repository files navigation

Hackathon Helper

Hackathon Helper is an AI-powered orchestration tool designed to eliminate the "cold-start" problem of hackathons. By automating the transition from a dense Devpost overview to a structured technical blueprint, it ensures developers spend their first hour coding instead of configuring.

🌟 Core Philosophy: Strategic Delivery

Success in a hackathon is about consistency and reliability. This tool is built to help you meet every requirement and ship a functional, polished project by identifying a clear, achievable path through the requirements from the very start.


🚀 Features

  • Strategic Analysis: Extracts mandatory requirements and proprietary platform constraints (like DigitalOcean or GitLab integrations).
  • Dynamic Scaffolding: Automatically generates .gitignore and requirements.txt based on the AI's prediction of your project's specific library needs.
  • Creative Brainstorming: Suggests achievable project concepts tailored to the extracted hackathon criteria.
  • Professional Dashboard: A Flask-based web interface (app.py) that renders your blueprint into a polished, browser-based report for high-quality presentation.

📂 Project Structure

The project is organized to support both the multi-agent logic and the professional presentation layer:

DIGITALOCEAN_GRADIENT_AI_HACKATHON
├── app.py                # Flask Web Interface (Professional Reporting)
├── main.py               # Core Multi-Agent Logic (LangGraph + Gradient)
├── devpost_input.txt     # The "Source of Truth" (Paste Hackathon text here)
├── hackathon_report.md   # Markdown Output
├── hackathon_report.json # Data Output
├── output/               # Generated Project Sandboxes
└── docs/                 # Documentation & Logs

🛠️ How It Works

The system uses a LangGraph multi-agent workflow powered by Gradient AI:

  1. Analyst Node: Maps out the mandatory "Must-Builds" and technical constraints.

  2. Scaffolder Node: Reasons about Python dependencies and creates the local environment files.

  3. Creative Node: Identifies valid, high-potential ideas that fit the specific hackathon scope.

  4. Flask Layer (app.py): Dynamically executes the workflow and renders the results in a clean, professional dashboard.


⚡ Quick Start

1. Setup Environment

Ensure your .env file contains your DIGITALOCEAN_API_TOKEN and GRADIENT_MODEL_ACCESS_KEY.

pip install -r requirements.txt

2. Add Your Hackathon

Paste the text from the hackathon overview page on DEVPOST into devpost_input.txt.

3. Run the Helper & Initialize Project

Launch the web dashboard. This single command triggers the multi-agent workflow, generates your project scaffolding, and prepares your report:

python app.py

4. Results & Scaffolding

  • Browser Report: Navigate to http://localhost:8081 to view the professionally rendered analysis and project concepts.
  • Physical Scaffolding: Check the /output/ directory. The agent will have automatically created a new folder named after the hackathon containing your customized .gitignore and requirements.txt.
  • Data Exports: Find the raw hackathon_report.json and hackathon_report.md in the root directory for your records.

🎓 What We Learned

Building this tool demonstrated the power of Multi-Agent Systems (MAS). By breaking down the problem into specialized roles (Analyst, Scaffolder, Creative), we achieved much higher accuracy and utility than a single LLM prompt could provide.

🔭 What's Next

GitLab Integration: Auto-committing scaffolded files to a new repository via webhooks.

Cloud Deployment: Migrating the local dashboard to a persistent cloud environment.

Multi-Language Support: Expanding scaffolding beyond Python to include Node.js, Rust, and Go.


Built for the DigitalOcean & Gradient AI Hackathon.

About

The Hackathon Helper uses DigitalOcean Gradient AI to set up projects in seconds. From the competition overview, it extracts requirements, generates config files and suggests ideas to pursue.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages