Where does the skill install files?

It scaffolds all necessary scripts, templates, and configurations into a 'scripts/ralph/' directory within your current git repository.

Can I run the autonomous loop inside Claude Code?

No, the autonomous loop scripts (ralph.sh) are designed to be executed directly from your terminal to allow for persistent background operation.

Which review backends are supported?

The skill supports RepoPrompt (ideal for macOS users with a visual builder) and Codex CLI (a cross-platform solution using high-tier GPT models).

What is the Ralph autonomous harness?

Ralph is a specialized framework for Claude Code that enables autonomous, long-running coding loops. It uses a plan-first approach and external review gates to ensure reliability during complex tasks.

Ralph Autonomous Coding Harness

Name: Ralph Autonomous Coding Harness
Author: gmickel

bygmickel

•

356

Productividad y Flujo de Trabajo

Scaffolds a repo-local autonomous agent harness for long-running, plan-first AI development loops.

Acerca de

The Ralph initialization skill sets up a specialized autonomous environment within your repository, enabling Claude to perform complex, multi-step coding tasks with built-in review gates. By scaffolding the Ralph harness under scripts/ralph/, it provides the necessary infrastructure for 'overnight' coding sessions where work is verified by secondary models like RepoPrompt or Codex. This skill automates the installation of flow control scripts and templates, ensuring a reliable, plan-first workflow that prevents context drift and maintains high code quality through automated verification loops.

Características Principales

Integrated plan-first development workflow templates
Automated local scaffolding of the Ralph autonomous harness
Automated permission handling and script configuration
Self-contained repository-specific automation scripts
Support for multi-model review gates via RepoPrompt and Codex
356 GitHub stars

Casos de Uso

Implementing complex features that require multiple planning and execution iterations
Establishing a multi-model verification pipeline for AI-generated code
Automating large-scale codebase refactoring during off-hours

Acerca de

Características Principales

Integrated plan-first development workflow templates
Automated local scaffolding of the Ralph autonomous harness
Automated permission handling and script configuration
Self-contained repository-specific automation scripts
Support for multi-model review gates via RepoPrompt and Codex
356 GitHub stars

Casos de Uso

Implementing complex features that require multiple planning and execution iterations
Establishing a multi-model verification pipeline for AI-generated code
Automating large-scale codebase refactoring during off-hours