ralph-loop-init

To initialize an AI workflow from an approved plan, this Skill generates a Ralph Loop scaffold including prd.json, loop.py, and progress.txt. It automatically validates the plan, extracts implementation steps into structured stories, and applies per-iteration execution constraints.

A Ralph Loop is an iterative AI-driven execution infrastructure that converts an approved plan into automated multi-step implementations. It uses generated artifacts like loop.py and guardrails to manage per-iteration tasks, track progress, and enforce quality gates throughout the execution cycle.

You can extract implementation steps from a plan into structured stories by invoking the initialization process on your plan path. The Skill automatically parses the approved plan, extracts the steps, and organizes them into a structured PRD for iterative execution.

You do not need to manually configure quality gates for plan-driven initialization. The Skill automatically detects and surfaces quality gates directly from your existing project configuration during the loop scaffolding process.

Setting up iterative AI-driven execution generates a .ralph directory containing prd.json, loop.py, CLAUDE.md, progress.txt, and guardrails. These artifacts collectively manage execution constraints, track per-story progress, and provide auditable recovery.

You can recover AI workflow progress if an iteration fails because the initialization generates guardrails and a progress.txt file. These artifacts provide safe, auditable execution tracking and per-story constraints designed specifically for recovery and resumption.