loop-until-done

Coordinate iterative engineering tasks through planning, execution, verification, and review until completion criteria pass.

4|Updated Feb 23, 2026
One-click install
npx skills add https://github.com/npow/claude-skills --skill loop-until-done
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: loop-until-done
Source: https://github.com/npow/claude-skills/tree/main/loop-until-done
Command: npx skills add https://github.com/npow/claude-skills --skill loop-until-done

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill prevents recurring, multi-step work from stopping prematurely by coordinating execution, verification, review, and durable progress tracking until the defined completion criteria are satisfied.

Core Features & Use Cases

  • PRD-Driven Execution: Breaks a broad task into stories with independently testable acceptance criteria.
  • Verification and Review: Filters criteria for falsifiability, verifies outcomes, and applies a final reviewer gate before completion.
  • Durable Recovery: Uses a Temporal-backed sagaflow workflow with session persistence, artifact tracking, heartbeat monitoring, and an in-session fallback when the worker is unavailable.
  • Use Case: Use it to keep fixing a failing setup, wait for CI results, or iteratively implement a feature until every acceptance criterion has fresh passing evidence.

Quick Start

Ask the loop-until-done skill to keep working on the task until all tests pass and coverage reaches 80 percent.

Frequently Asked Questions about loop-until-done

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

FAQPage Schema
How do I automate iterative task completion for multi-step engineering work?

Automated iterative task completion coordinates planning, execution, and verification until defined acceptance criteria pass. It breaks broad tasks into stories with independently testable criteria and applies a final reviewer gate before completion.

How does CI polling work with durable recovery across session failures?

CI polling with durable recovery uses a Temporal-backed sagaflow workflow with session persistence, artifact tracking, and heartbeat monitoring. It supports an in-session fallback when the worker is unavailable, ensuring progress persists across attempts.

Can I use PRD-based implementation to verify outcomes with falsifiable acceptance criteria?

Yes, PRD-based implementation breaks a broad task into stories and filters criteria for falsifiability. It verifies outcomes with fresh passing evidence and applies reviewer validation before marking the task complete.

Do I need a Temporal-backed sagaflow worker to drive recurring work to verified completion?

A Temporal-backed sagaflow worker is required when available for durable recovery and session persistence. However, the Skill supports an in-session fallback when the worker is unavailable.

Why does my setup recovery stop prematurely before all tests pass?

Setup recovery stops prematurely when execution lacks durable progress tracking and verification gates. Coordinating execution, verification, and review until defined completion criteria are satisfied prevents recurring work from stopping early.

What is the best way to handle iterative bug fixing with structured acceptance criteria?

Iterative bug fixing with structured acceptance criteria requires filtering criteria for falsifiability, verifying outcomes with fresh passing evidence, and applying a final reviewer gate. Durable progress tracking ensures tasks persist across attempts until completion.