ralph

Execute a Read-Act-Learn-Feedback loop until a binary success condition is met.

2|Updated Mar 19, 2026
One-click install
npx skills add https://github.com/alex-voloshin-dev/ai-skills --skill ralph-alex-voloshin-dev
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ralph
Source: https://github.com/alex-voloshin-dev/ai-skills/tree/main/plugin/skills/ralph
Command: npx skills add https://github.com/alex-voloshin-dev/ai-skills --skill ralph-alex-voloshin-dev

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) and assets (resource) components.

What problem does it solve?

This skill solves the problem of inefficient, manual trial-and-error cycles when attempting to reach a specific, verifiable goal in code or data generation.

Core Features & Use Cases

  • Mechanically-Verifiable Iteration: Automatically runs tasks until a binary success condition (oracle) is met, such as passing tests or validating schemas.
  • Safety Guardrails: Prevents runaway costs and infinite loops through configurable token budgets, time caps, and iteration limits.
  • Use Case: Use this to refactor a codebase until all unit tests pass or to generate test cases until they meet a specific rubric score threshold.

Quick Start

Invoke the ralph skill to write a Rust CLI that parses CSV files by using the test-cli.sh script as an oracle and stopping when the oracle reports success.

Frequently Asked Questions about ralph

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

FAQPage Schema
How do I automate code refactoring until all unit tests pass?

Automated iteration loops handle verifiable tasks by running a Read-Act-Learn-Feedback cycle until a binary success condition is satisfied. The mechanism relies on oracle scripts to validate results and kill-on signals to maintain deterministic control over the agentic loop.

What is an agentic feedback loop for test generation?

An agentic feedback loop for test generation automatically iterates code generation tasks until a binary success condition is met. It uses oracle scripts to validate outputs against a rubric score threshold, ensuring generated tests meet specific quality criteria before stopping.

How do I prevent infinite loops and runaway costs during automated iteration?

To prevent infinite loops and runaway costs during automated iteration, configure safety guardrails like token budgets, time caps, and iteration limits. These constraints ensure deterministic control over the agentic loop and stop execution when limits are exceeded.

Do I need oracle scripts to run automated refactoring loops?

Yes, oracle scripts are required to run automated refactoring loops. The iteration cycle depends on these scripts to evaluate task completion and report a binary success condition, which triggers the kill-on signal to stop the agentic loop.

Can I use automated iteration loops for schema validation tasks?

Yes, you can use automated iteration loops for schema validation tasks. The loop manages complex development workflows including schema validation by iterating until the oracle script confirms the generated data meets the strict schema constraints.