Two prompts that make Claude Code way more effective for building features

This repo has two markdown files: one generates a PRD from your feature description, the other breaks that PRD into granular tasks with sub-steps.

Why it works: instead of throwing a big request at Claude Code and hoping for the best, you're forcing a structured approach. Define scope first, then break it into small verifiable chunks. Claude works through tasks one at a time, you approve each step, and progress gets tracked in the task file.

It's not magic, but it's a more reliable way to build larger features without Claude going off the rails or losing context mid-way through.