ralph-init_skill

Overview

This skill initializes a Product Requirements Document (PRD) to drive structured ralph-loop execution. It converts a short project or feature description into a testable, linked PRD that Ralph can use as the definition of done. The PRD is saved in the project plans directory and prepared for automated iteration.

How this skill works

Provide a project or feature description and the skill conducts an interactive requirements gathering if needed. It generates a structured PRD (.omc/plans/prd-{slug}.md) containing problem statement, goals and non-goals, acceptance criteria, technical constraints, and implementation phases. The PRD is linked so the ralph executor can consume it and iterate until all acceptance criteria are met.

When to use it

• Starting a new feature or project that will be implemented via Ralph multi-agent execution
• You need a clear, testable definition of done before automation begins
• Converting stakeholder notes or a short brief into a structured requirements artifact
• Preparing handoff materials for automated agents and human architects
• When you want phased delivery and explicit acceptance criteria for iterative runs

Best practices

• Provide a concise initial description to speed up PRD generation
• Be explicit about scope: list non-goals to prevent scope creep
• Define acceptance criteria as testable statements to enable automated verification
• Include technical constraints early to avoid late rework
• Use implementation phases to break the work into verifiable milestones

Example use cases

• Turn a one-paragraph feature brief into a PRD for Ralph to implement end-to-end
• Prepare a new integration by documenting constraints, goals, and acceptance tests
• Create phased rollout plans for a multi-team feature with clear verification steps
• Convert stakeholder interview notes into a machine-readable definition of done
• Seed a multi-agent workflow with a formal PRD so agents can iterate autonomously