ralph

Coordinate multi-phase agent loops with iterative validation and planner approval.

1|Updated Feb 19, 2026
One-click install
npx skills add https://github.com/moveonha/pixel-ai-studio --skill ralph-moveonha
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ralph
Source: https://github.com/moveonha/pixel-ai-studio/tree/main/.claude/skills/ralph
Command: npx skills add https://github.com/moveonha/pixel-ai-studio --skill ralph-moveonha

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Ralph provides an automated, self-referential loop to drive complex tasks from inception to verified completion by repeatedly executing phases, validating results, and requiring planner approval before finalization.

Core Features & Use Cases

  • Multi-phase orchestration for implementing tasks across multiple agents (execution, validation, planner, and documentation).
  • Evidence-based verification with fresh checks on every iteration and no reuse of prior results.
  • Planner-driven completion requiring explicit approval to emit the final <promise>, ensuring rigorous quality control.
  • Comprehensive state tracking through per-iteration documentation (TASKS, VERIFICATION, PROCESS, ITERATION).

Quick Start

Provide the original task as PROMPT and run the Ralph loop until completion and a final <promise> is emitted.

Frequently Asked Questions about ralph

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

FAQPage Schema
How do I automate multi-agent workflow verification for complex software tasks?

Multi-agent workflow verification is automated through a self-referential loop that executes implementation, validates results with fresh checks, and requires planner approval before finalizing. This enforces evidence-based completion across software development, documentation, and deployment workflows.

What is a self-referential loop for iterative task completion?

A self-referential loop for iterative task completion repeatedly executes phases, validates results, and requires planner approval before emitting a final promise. It tracks per-iteration provenance across TASKS, VERIFICATION, PROCESS, and ITERATION documentation to guarantee completion.

How do I set up a multi-phase orchestration loop with planner approval?

Multi-phase orchestration is set up by providing the original task as a prompt and running the loop until completion. The loop automatically enforces Phase 1 implementation, Phase 2 pre-deploy validation, Phase 3 planner verification, and Phase 4 final promise output.

Does multi-agent iterative validation reuse results from prior iterations?

Multi-agent iterative validation does not reuse results from prior iterations. It performs evidence-based verification with fresh checks on every iteration, requiring explicit planner approval before declaring completion to ensure rigorous quality control.

When do I need planner-driven completion for deployment workflows?

Planner-driven completion is needed for multi-phase, multi-agent projects requiring evidence-based verification before declaring completion. It applies to software development, documentation, and deployment workflows where rigorous quality control across implementation and validation phases is mandatory.