Keyboard shortcuts

Press ← or → to navigate between chapters

Press S or / to search in the book

Press ? to show this help

Press Esc to hide this help

How to Use This Framework

Welcome to the practitioner's guide for Hackathon Alchemy.

This guide explains how to effectively extract maximum value from these prompts when pairing with modern Large Language Models (LLMs) and AI tools like ChatGPT, Claude, Gemini, Cursor, GitHub Copilot, and Windsurf.

⚡ The Core Workflow

  1. Pick Your Category: Navigate through the sidebar to your relevant hackathon track (e.g., 01. AI / ML Hackathons, 03. Automation, 06. Web3).
  2. Move Sequentially Through the Phases:
    • Phase 1 locks down your problem scope and requirements before any code is typed.
    • Phase 2 designs the technical contracts (database, API, architecture).
    • Phase 3 hardens the implementation with error boundaries, test scripts, and UI specs.
    • Phase 4 organizes deployment and your presentation pitch.
  3. Copy the Prompt: Use the built-in copy button on any prompt code block.
  4. Supply Your Context: Replace any bracketed placeholders like [PASTE PROBLEM STATEMENT] or [PASTE PRD] with your project's details.
  5. Save Outputs as Markdown: Save each AI-generated output in your project's repository under docs/ or your root folder (e.g., PRD.md, SYSTEM_ARCHITECTURE.md).
  6. Pass Outputs Forward: In subsequent phases, feed the generated documents into the next prompt! (e.g., feed PRD.md into the TRD.md prompt).

💡 Best Practices for Hackathons

1. Don't Skip Phase 1

The number one reason hackathon teams run out of time is building features nobody asked for or changing scope 8 hours before submission. Spending 30 minutes in Phase 1: Problem Definition guarantees consensus across your team and judges.

2. Treat Documents as Context Anchors

When working with AI coding tools (like Cursor, Claude Code, or Copilot Workspace), having crisp documents like SYSTEM_ARCHITECTURE.md, API_SPECIFICATION.md, and UI_SPEC.md directly in your repo gives your coding assistant clean grounding context.

3. Maintain an MVP Mindset

Every prompt in Hackathon Alchemy emphasizes:

  • What must be built for the demo.
  • What should NOT be built during a short sprint.
  • Realistic fallback mechanisms if an external API fails during live judging.

🛠️ Recommended AI Tooling

  • Reasoning & Planning: Claude 3.5 Sonnet, GPT-4o, Gemini 1.5 Pro
  • In-Editor Coding: Cursor, Windsurf, VS Code + GitHub Copilot
  • Mockups & Frontend: v0.dev, Bolt.new, Tailwind UI
  • Workflow Testing: Postman, Thunder Client, curl

🧭 Next Step

Head over to 01. AI / ML Hackathons or browse the sidebar to start your project.