Does this skill run as a background daemon?

No, the /loop command is not a background process. Each invocation handles exactly one task and returns control to the user, ensuring safety and visibility.

How does it handle multiple AI platforms working at once?

It uses a shared state within the .agents/TASKS/ directory. Agents 'claim' a task with a timestamped lock, preventing other platforms from working on the same file for 30 minutes.

What happens when an AI agent hits a rate limit?

The agent saves its current progress to the Agent-Notes field, releases its claim on the task, and suggests switching to a different platform to continue the work.

How do I review and approve the AI's work?

You can use a visual Kanban board via the Kaiban.md extension or manually review the Testing column. You can then move tasks to 'Done' or 'Reject' them with specific feedback.

AI Development Loop

Name: AI Development Loop
Author: shipshitdev

byshipshitdev

•

Productividad y Flujo de Trabajo

Orchestrates autonomous software development by managing AI agents through a structured task queue and human-in-the-loop QA gates.

AI Development Loop turns Claude into a high-efficiency autonomous developer by implementing a task-based workflow within your repository. It allows AI agents to independently claim, implement, and document features from a prioritized backlog while you focus exclusively on architectural review and quality assurance. Designed for high-velocity teams, it supports parallel execution across Claude CLI, Cursor, and Codex, featuring a smart locking system to prevent conflicts and built-in strategies to bypass platform rate limits by seamlessly switching contexts.

Características Principales

01Integrated human-in-the-loop QA gates with automated rejection handling and feedback loops.

02Conflict prevention via a 30-minute claim expiration system for parallel agent workflows.

03Rate limit mitigation that saves progress and allows platform handoffs without losing context.

0410 GitHub stars

05Multi-platform orchestration supporting Claude CLI, Cursor, and Codex integration.

06Autonomous task claiming and execution using a structured markdown-based queue.

Casos de Uso

01Coordinating parallel development sessions across multiple AI IDEs and CLI tools simultaneously.

02Standardizing AI behavior for complex refactoring tasks that require multi-step verification and QA.

03Managing a high-velocity feature backlog where developers act as reviewers rather than typists.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add shipshitdev/library ai-dev-loop

For use in Claude.ai and ChatGPT

Download Skill