Prompt Driven Development
PDD is a development approach where prompts — instructions given to AI models — are treated as first-class artifacts, not throwaway inputs.
Just as traditional development has source code, tests, and documentation, PDD has a structured set of prompts, context files, and outputs that together define how a project is built.
The Problem
Without structure, AI-assisted development drifts. You re-explain the same context every session, generate code that contradicts earlier decisions, and lose the reasoning behind why things were built a certain way.
The Solution
PDD gives you four layers:
| Layer | Purpose |
|---|---|
| Context | Permanent project briefing — stack, conventions, decisions |
| Prompts | Modular, single-purpose instructions for each feature |
| Output | Reviewed and accepted AI-generated artifacts |
| Eval | Validation checklists and automated checks |
Quick Start
Claude Code (plugin): Run inside Claude Code:
/plugin marketplace add harshal2802/pdd-skill
/plugin install pdd-skill
Claude Code (manual):
git clone https://github.com/harshal2802/pdd-skill.git .claude/skills/pdd-skill
echo '{ "skills": [".claude/skills/pdd-skill/skills/pdd/SKILL.md"] }' > .claude/settings.json
Then use PDD — Claude auto-detects your intent, or use slash commands:
/project:pdd-scaffold— new project/project:pdd-init— existing project/project:pdd-help— see all commands
Learn More
- The Philosophy Behind PDD — Why structure matters, the four layers, project type flavors, and how to get started
- Efficiency Tips — Practical habits for reducing token usage and cost without changing the PDD workflow
- GitHub Repository — Source code, installation instructions, and examples
- Copilot Version — Same workflows adapted for GitHub Copilot Chat
Built by Harshal Chourasiya · GitHub