ralph

Automate multi-session tasks with persistence, retries, and architect verification.

Updated Feb 15, 2026
One-click install
npx skills add https://github.com/leejaedus/pepcode --skill ralph-leejaedus
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ralph
Source: https://github.com/leejaedus/pepcode/tree/main/skills/ralph
Command: npx skills add https://github.com/leejaedus/pepcode --skill ralph-leejaedus

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Tasks often stall after partial progress or silent failures. Ralph enforces persistence and architect verification to ensure a task runs to true completion rather than stopping at the first working version.

Core Features & Use Cases

  • Persistence across iterations: the loop resumes where it left off after failures or retries.
  • Parallel delegation: agents of different tiers handle subtasks simultaneously to speed up delivery.
  • Mandatory verification: an architect review confirms the goal is truly achieved before completion.
  • State and evidence tracking: keeps progress, results, and verification artifacts for audit.

Quick Start

Initiate a long-running task and let Ralph persistently iterate until you receive a verifiable completion.

Frequently Asked Questions about ralph

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

FAQPage Schema
How do I automate task completion across multiple sessions without losing progress?

Automated task completion across multiple sessions requires a persistent loop that preserves progress and resumes where it left off after failures. Ralph enforces this persistence to ensure tasks run to true completion rather than stopping at partial progress.

What is architect verification in agent orchestration?

Architect verification in agent orchestration is a mandatory review step that confirms a goal is truly achieved before marking it complete. Ralph applies this strict completion criteria to prevent silent failures and ensure verifiable results.

How do I handle parallel delegation for subtasks in multi-agent orchestration?

Parallel delegation in multi-agent orchestration assigns subtasks to agents of different tiers simultaneously to speed up delivery. Ralph manages this parallel execution while tracking state and evidence for audit.

Why do tasks stall after partial progress and how can I prevent it?

Tasks stall after partial progress due to silent failures or lack of strict completion criteria. Prevent this by applying persistent iteration and mandatory architect verification to ensure the task runs to true completion.

Can I use persistent iteration for long-running tasks that require fresh evidence collection?

Yes, persistent iteration supports long-running tasks that require fresh evidence collection. Ralph keeps progress, results, and verification artifacts across iterations to satisfy strict completion criteria and audit requirements.

What are the limitations of relying on the first working version of a task?

Relying on the first working version risks silent failures and incomplete goals. Ralph overcomes this limitation by enforcing persistence and architect verification across iterations to guarantee true task completion.