Requirements that agents can actually use
Type
/prd with a feature or product description and Claude generates a machine optimized PRD. Unlike traditional PRDs written for humans, this format uses structured sections, explicit acceptance criteria, and unambiguous scope definitions that AI coding agents can parse and execute without interpretation gaps.1
Gather requirements
Asks about the product goal, target users, technical constraints, existing codebase context, and desired scope. Fills in reasonable defaults where possible.
2
Structure the PRD
Organizes requirements into machine readable sections: overview, user stories, acceptance criteria, technical constraints, out of scope, and dependencies.
3
Optimize for agents
Formats each section with explicit, unambiguous language. Removes vague terms, adds concrete examples, and ensures every requirement is testable.
4
Save the document
Writes the PRD as a markdown file in the project directory. Ready to be fed directly to a coding agent or used as a reference during implementation.