ralph

Run a persistent iteration loop with tests, build, lint, and architect validation until completion.

2|Updated May 11, 2026
One-click install
npx skills add https://github.com/xz1220/oh-my-kimi --skill ralph-xz1220
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ralph
Source: https://github.com/xz1220/oh-my-kimi/tree/main/skills/ralph
Command: npx skills add https://github.com/xz1220/oh-my-kimi --skill ralph-xz1220

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Ralph prevents complex tasks from silently stopping early or being marked complete without fresh evidence and a final verification sign-off.

Core Features & Use Cases

  • Persistent completion loop: Runs a durable cycle that continues until the task is actually finished and validated.
  • Fresh evidence validation: Ensures completion is backed by new checks like tests/build/lint and explicit zero-TODO conditions.
  • Architect gatekeeping: Routes the final approval step through architect verification, using higher rigor for larger or riskier changes.
  • Cleanup and regression safety: Applies a scoped “deslop” cleanup pass and then re-runs regression checks to maintain quality.

Quick Start

Tell your Kimi session to run the skill until completion with verification, for example: /skill:ralph finish this feature with tests and verification evidence

Frequently Asked Questions about ralph

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

FAQPage Schema
How do I prevent complex software tasks from stopping early or being marked complete without verification evidence?

To prevent complex software tasks from stopping early, you can use a persistent iteration loop that enforces fresh evidence validation, requiring passing tests, builds, and lint checks before completion. This ensures tasks are truly finished and backed by new verification data.

What is an architect review gate in software verification and when do I need it?

An architect review gate in software verification is a final approval step that routes completed work through higher rigor validation for larger or riskier changes. You need it to ensure complex integration work meets architectural standards before final task completion.

How do I automate task completion for multi-step software engineering work with regression testing?

You automate task completion by running a durable stateful workflow that applies parallel agent-driven implementation and verification lanes. This loop executes scoped cleanup passes and re-runs regression testing to maintain quality until the task is validated.

Does agent orchestration work for tasks requiring multiple retries and parallel implementation?

Yes, agent orchestration works for complex software and integration work by managing a stateful workflow that spans multiple attempts and retries. It coordinates parallel agent-driven implementation alongside continuous state tracking and explicit verification steps.

What are the limitations of using an iteration loop for software verification?

The iteration loop requires continuous state tracking and explicit verification steps for every change, meaning it demands computational overhead for running regression checks and scoped cleanup passes, making it less suited for trivial or single-step modifications.