Skip to content

Home

Jump to bottom
Alp Yalay edited this page Aug 8, 2025 · 11 revisions

Welcome to the vibe-coding-prompt-template wiki!

This repository provides a structured workflow for turning an app idea into a functional Minimum Viable Product (MVP) using AI agents. It's designed for "vibe-coders"—people with great ideas but limited traditional coding skills.

The core of this repository is a 4-part process that uses a series of markdown-based prompt templates. You feed these templates to a powerful AI model (like Gemini 2.5 Pro or Claude 4) to generate key documents that guide the development process.

The 4-Step Workflow

The process is broken down into four distinct stages, each with its own template file.

1️⃣ Part 1: Deep Research (part1-deepresearch.md)

  • Goal: To validate your app idea and understand the market.
  • Process: You answer a series of questions based on your technical comfort level. The AI then uses your answers to generate a comprehensive research document.
  • Output: A research-[YourAppName].txt file containing market analysis, competitor breakdown, technical recommendations, and cost estimates.

2️⃣ Part 2: Product Requirements Document (PRD) (part2-prd-mvp.md)

  • Goal: To clearly define what you are building, for whom, and why.
  • Process: You provide your research document and answer questions about core features, target users, and success metrics.
  • Output: A PRD-[YourAppName]-MVP.md file that acts as the blueprint for your product.

3️⃣ Part 3: Technical Design (part3-tech-design-mvp.md)

  • Goal: To decide how to build the app using the best modern tools.
  • Process: The AI analyzes your PRD and asks about your platform choice, budget, and timeline to recommend the best tech stack.
  • Output: A TechDesign-[YourAppName]-MVP.md file outlining the architecture, tools, and implementation plan.

4️⃣ Part 4: AI Agent Instructions (part4-notes-for-agent.md)

  • Goal: To create a final, consolidated set of instructions for your AI coding agent.
  • Process: The AI combines your PRD and Technical Design into a single, actionable instruction file.
  • Output: A NOTES.md file that your chosen AI coding assistant (like Cursor or a terminal agent) will use to write the actual code.

How to Use This Repository

  1. Start with Part 1: Open part1-deepresearch.md, copy its content, and paste it into a powerful AI chat model like Gemini 2.5 Pro.
  2. Answer the Questions: The AI will ask you a series of questions. Your answers will tailor the output to your specific project.
  3. Generate the Document: The AI will produce the first document (your research findings).
  4. Repeat for All Parts: Continue this process for part2, part3, and part4, feeding the previously generated documents into the AI when prompted.
  5. Build: Give the final NOTES.md file to your AI coding agent and let it build your application.

Recommended Tools

The workflow is flexible, but the README.md suggests a few key tools for 2025:

  • AI Chat Platforms:
    • AI Studio: For Gemini 2.5 Pro (recommended for research).
    • Claude.ai: For Claude 4 models (recommended for technical accuracy).
    • ChatGPT: For GPT-5 models.
  • AI Coding Agents/IDEs:
    • Cursor: A powerful, AI-native code editor.
    • Terminal Agents: Claude Code, Gemini CLI for more advanced users.
    • No-Code Platforms: Bolt.new, Lovable for the fastest results.

This structured approach ensures that by the time the AI starts writing code, it has a deep, contextual understanding of the project, leading to a much higher chance of success for your MVP.

Clone this wiki locally