ralph

Automate a persistence loop requiring architect verification before task completion.

Updated Apr 3, 2026
One-click install
npx skills add https://github.com/pextacy/stellar --skill ralph-pextacy
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ralph
Source: https://github.com/pextacy/stellar/tree/main/skills/ralph
Command: npx skills add https://github.com/pextacy/stellar --skill ralph-pextacy

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Ralph provides a persistence loop that ensures a task is not considered complete until explicit architect verification, preventing partial outputs and undiscovered failures.

Core Features & Use Cases

  • Maintains session-grounded state across retries and iterations
  • Delegates work to specialist agents in parallel while preserving end-to-end verification
  • Enforces mandatory verification before completion and escalates issues until resolved

Quick Start

Start the Ralph loop to persistently work on a task until it is architect-verified complete.

Frequently Asked Questions about ralph

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

FAQPage Schema
How do I enforce architect verification before a long-running task is marked complete?

Architect verification is enforced through a persistence loop that escalates issues and retries iterations until an explicit architect confirms completion, preventing partial outputs. This ensures tasks spanning multiple iterations are fully verified.

What is agent orchestration for parallel delegation with state management?

Agent orchestration with state management delegates work to specialist agents in parallel while maintaining session-grounded state across retries. This preserves end-to-end verification and context grounding throughout long-running operations.

How do I maintain session state across multiple iterations of a delegated task?

Session state is maintained across iterations by grounding context continuously within the persistence loop. This keeps state intact during parallel delegation, background operations, and retries until architect verification succeeds.

Can I run long operations in the background while waiting for cross-agent verification?

Long operations run in the background while the persistence loop enforces continuous progress and cross-agent verification. This applies to long-running tasks requiring parallel delegation and escalates issues until verification is achieved.

When do I need a persistence loop for cross-agent verification?

A persistence loop is needed for long-running tasks that span multiple iterations and require parallel delegation. It prevents undiscovered failures by enforcing continuous progress and mandatory architect verification before completion.

Why does my multi-iteration task produce partial outputs without verification?

Partial outputs occur when tasks lack a persistence loop with explicit architect verification. Without enforced continuous progress and cross-agent escalation, delegated work completes prematurely without confirming all failures are resolved.