What is the Ralph Wiggum loop pattern?

The Ralph Wiggum loop is a continuous iteration pattern for Claude Code that keeps an AI agent working on a task until a specific, objective completion signal (a promise) is generated.

Can I run this skill overnight?

Yes, one of the primary use cases for the Ralph Wiggum loop is to enable autonomous development or testing sessions while the developer is away, provided safety caps are properly configured.

What is a 'Completion Promise' in this context?

A Completion Promise is a unique, exact-match phrase (like COMPLETE ) that the AI is instructed to output only when it has objectively verified that the task is finished.

How does this skill prevent excessive token usage?

The skill implements safety measures including a mandatory 'Max Iterations' cap (typically 10-50) and 'Stuck Detection' which terminates the loop if no progress is made after several attempts.

When should I avoid using the Ralph Wiggum workflow?

Avoid using it for tasks requiring subjective human judgment, such as UX design or architectural trade-offs, and for tasks with ambiguous success criteria that cannot be verified automatically.

Ralph Wiggum Continuous Iteration

概要

This skill implements the 'Ralph Wiggum' pattern, a specialized workflow for Claude Code that enables autonomous, multi-step iteration on well-defined coding tasks. By defining clear completion promises and safety limits, it allows AI agents to independently handle Test-Driven Development (TDD), bug fixes, and refactoring sessions. It is particularly effective for 'overnight' development or complex feature implementations where progress can be verified through automated tests, linters, or type checkers, ensuring the AI persists until the job is done.

主な機能

2 GitHub stars
Configurable safety caps for iterations to prevent infinite loops and token waste
Automated TDD iteration cycles that run until all tests pass
Integrated monitoring for stuck state detection and cost management
Completion promise verification for objective and deterministic task finalization
Specialized prompt templates for bug fixing, refactoring, and feature builds

ユースケース

Running autonomous bug-fixing sessions until specific unit tests are green
Automated refactoring of legacy codebases with continuous linting and regression testing
Implementing complex features overnight using continuous feedback loops and automated verification

概要

主な機能

2 GitHub stars
Configurable safety caps for iterations to prevent infinite loops and token waste
Automated TDD iteration cycles that run until all tests pass
Integrated monitoring for stuck state detection and cost management
Completion promise verification for objective and deterministic task finalization
Specialized prompt templates for bug fixing, refactoring, and feature builds

ユースケース

Running autonomous bug-fixing sessions until specific unit tests are green
Automated refactoring of legacy codebases with continuous linting and regression testing
Implementing complex features overnight using continuous feedback loops and automated verification

Continuous Development Loop (Ralph Wiggum)

Continuous Development Loop (Ralph Wiggum)

概要

主な機能

ユースケース

概要

主な機能

ユースケース