What happens when a task fails validation?

If a tick fails validation, failure feedback is appended to the plan, its status is reset to 'ready', and the next execution tick will attempt a retry based on the new feedback.

How does Autodev Loop ensure code quality?

The skill enforces a strict rule where the agent that implements a feature never validates its own work; validation is performed by a subsequent tick using specific fitness functions and gates.

Can I schedule development tasks to run in the background?

Yes, Autodev Loop includes a Time mode that uses CronCreate to schedule recurring ticks at specific intervals, allowing for autonomous progress over hours or days.

What is a 'tick' in the Autodev Loop architecture?

A tick is a single execution cycle where the agent performs exactly one action from a priority list: sweeping completed items, validating previous work, executing a new task, or performing codebase hygiene.

What is the purpose of the Hygiene mode?

Hygiene mode runs when there is no active work to do; it scans for architectural drift, reviews draft plans for readiness, and proposes updates for stale project artifacts.

Autodev Loop

Name: Autodev Loop
Author: lev-os

bylev-os

•

生産性とワークフロー

Automates the software development lifecycle by orchestrating task execution, validation, and codebase hygiene through scheduled autonomous cycles.

Autodev Loop is a sophisticated workflow automation skill designed to manage the end-to-end development process autonomously within Claude Code. It operates on a priority-based 'tick' architecture that handles sweeping orphaned entities, validating completed work, executing active development plans, and performing proactive codebase hygiene. By utilizing standardized flow definitions and strictly separating implementation from validation, it ensures high-quality code generation and continuous project momentum without requiring constant human oversight.

主な機能

01Separation of concerns between implementation and validation agents

021 GitHub stars

03Standardized report contracts for auditable agent progress

04Scheduled background task execution via integrated cron support

05Proactive drift detection between code and architectural specs

06Autonomous multi-step development orchestration

ユースケース

01Managing complex, long-running engineering tasks through autonomous background cycles

02Enforcing rigorous code quality standards via multi-agent validation gates

03Maintaining synchronization between documentation, design specs, and active codebases

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add lev-os/agents autodev-loop

For use in Claude.ai and ChatGPT