PRD Maker
Open-source workflow tool for AI coding agents
Interview-driven PRD generator that produces an executable PRD.md for Claude Code, Codex, and agent skills—plus a structure linter.
Overview
- My role
- Author & maintainer
Operational context
Vague prompts make coding agents fail confidently. A good agent PRD needs explicit non-goals, verifiable acceptance criteria, and phased scope.
Users & stakeholders
- Builders who need agent-executable requirements without guessing
Problem
People struggle to turn a rough idea into a PRD that AI coding agents can execute reliably.
Constraints
- Keep output pure markdown with minimal dependencies
- Work for non-developers as well as engineers
My ownership
- Designed the adaptive interview and agent-oriented PRD template
- Shipped Claude Code / Codex plugin paths and validation script
Solution
An interview flow that adapts depth to the user, writes PRD.md, and lints structure for agent-ready requirements.
Key decisions
Optimize for agent execution, not slide decks
Traditional PRDs leave gaps humans fill; agents fill them incorrectly.
Enforce non-goals, phased scope, and machine-checkable acceptance criteria.
Trade-off: Stricter template; higher upfront interview effort.
Deployment & operations
Open source on GitHub (6 stars, 4 forks as of 2026-08-06).