Is there a limit to how many times the loop will run?

Yes, users must set a max-iterations parameter to prevent infinite loops and manage API costs effectively. It is recommended to start with a low number like 5.

How do I enable the Ralph Loop skill?

You must explicitly opt-in by setting enabled: true in your .opencode/oh-my-opencode.json configuration file, as it is disabled by default for safety.

How are execution logs handled?

Logs are saved to a local .opencode/runs directory. To keep the conversation clean, users are encouraged to use memory_add to reference the logs rather than pasting them directly into the chat.

What is the primary purpose of the Ralph Loop skill?

It automates the cycle of coding and testing, repeatedly refining the code until a specific completion condition, like passing tests or reaching coverage goals, is reached.

What is a completion promise in this context?

A completion promise is a specific natural language or technical condition (e.g., 'All tests pass and coverage > 80%') that tells the skill when it has successfully finished the task.

OpenCode Ralph Loop

Name: OpenCode Ralph Loop
Author: taiyousan15

bytaiyousan15

0•

Productivity & Workflow

Automates iterative development cycles by repeatedly refining code until specific completion criteria and test requirements are met.

OpenCode Ralph Loop is a specialized development skill designed to handle repetitive coding tasks and feature implementations through a controlled feedback loop. It allows developers to set specific goals—such as passing all unit tests or reaching a code coverage threshold—and automates the process of code modification and verification. By incorporating strict iteration limits and structured state management, it provides a safe, opt-in framework for autonomous task completion while preventing runaway resource consumption and ensuring every change is logged and verified.

Key Features

01Integrated state management and run logging for full traceability of code changes

02Goal-oriented execution based on custom completion promises and success criteria

03Strict opt-in safety mechanisms to prevent accidental or excessive API usage

04Memory-optimized log handling using memory_add to maintain clean conversation contexts

05Controlled iterative development with configurable maximum iteration limits

060 GitHub stars

Use Cases

01Refactoring complex legacy code until it passes a comprehensive suite of unit tests

02Iteratively improving codebase quality to meet specific test coverage thresholds

03Automating multi-step feature implementations that require sequential refinement and verification

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add taiyousan15/taisun_agent opencode-ralph-loop

For use in Claude.ai and ChatGPT

Download Skill