ralph

Iteratively execute PRD-defined user stories until acceptance criteria pass and reviewer verification completes.

1|Updated Mar 30, 2026
One-click install
npx skills add https://github.com/Leap0920/Clean-Portfolio --skill ralph-leap0920
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ralph
Source: https://github.com/Leap0920/Clean-Portfolio/tree/main/%25USERPROFILE%25/.openclaude/plugins/cache/omc/oh-my-claudecode/4.14.0/skills/ralph
Command: npx skills add https://github.com/Leap0920/Clean-Portfolio --skill ralph-leap0920

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Ralph prevents silent failure by continuing work iteratively until every PRD user story is marked as passing and has been verified by a reviewer, rather than stopping at partial progress.

Core Features & Use Cases

  • PRD-driven persistence loop: Breaks tasks into user stories stored in prd.json and keeps iterating until each story’s acceptance criteria pass.
  • Automated story refinement & validation: Replaces generic acceptance criteria from scaffolds with task-specific, verifiable checks before implementation.
  • Tiered reviewer verification + regression guardrails: Runs reviewer approval and then performs a mandatory cleanup (deslop) pass followed by re-verification to ensure nothing breaks.

Quick Start

Ask the AI to run the ralph skill for your task and tell it to continue until all completion criteria are verified.

Frequently Asked Questions about ralph

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I ensure PRD user stories pass all acceptance criteria without silent failures?

To ensure PRD user stories pass acceptance criteria without silent failures, use an iterative execution loop that continuously runs tasks until every story is verified by a reviewer. This approach replaces generic criteria with task-specific, verifiable checks before implementation.

What is the best way to track task execution progress across multiple iterations and retries?

The best way to track task execution progress across multiple iterations is through session-persistent progress tracking stored in a prd.json file. This method breaks tasks into individual user stories and maintains state across retries until all completion criteria are met.

How does regression testing work after a mandatory deslop cleanup pass?

Regression testing after a deslop cleanup pass works by performing a mandatory re-verification of all previously approved user stories. After initial reviewer sign-off, the cleanup pass removes code bloat, followed by a regression guardrail to ensure the cleanup process breaks nothing.

Do I need a PRD scaffold to initialize task execution for multi-iteration engineering work?

Yes, you need a PRD scaffold to initialize task execution for multi-iteration engineering work. The process requires PRD scaffold initialization and refinement to break down tasks into user stories with verifiable acceptance criteria before iterative implementation begins.

Why does automated story refinement replace generic acceptance criteria before implementation?

Automated story refinement replaces generic acceptance criteria to prevent unverifiable partial progress. By generating task-specific, verifiable checks from the scaffold before implementation, the system ensures that iterative execution targets precise completion conditions rather than vague requirements.

Can I use session persistence to prevent stopping at partial progress during regression testing?

Yes, you can use session persistence to prevent stopping at partial progress during regression testing. By storing user stories in a persistent prd.json file, the execution loop continues across retries until all acceptance criteria pass and receive reviewer sign-off.