How does this differ from the standard 'Executing Plans' skill?

While 'Executing Plans' often uses parallel sessions, this skill stays in your current session and enforces a stricter two-stage review process (spec compliance then code quality) for every task.

Do I need a plan file before using this skill?

Yes, this skill is designed to execute an existing implementation plan typically created by a planning skill or manual brainstorming session.

Can I skip the review stages for faster execution?

Skipping reviews is discouraged as it violates the core principle of the skill. The two-stage review is essential for preventing technical debt and ensuring the code matches your requirements.

What happens if a subagent fails a review?

The workflow automatically enters a fix-and-review loop where the implementer subagent addresses specific feedback and resubmits until both quality gates are passed.

Subagent-Driven Development

Name: Subagent-Driven Development
Author: astrosteveo

byastrosteveo

•

生产力与工作流

Executes complex implementation plans by dispatching specialized subagents for each task with rigorous two-stage quality reviews.

关于

Subagent-Driven Development is an advanced workflow for Claude Code designed to automate the execution of modular implementation plans. By isolating individual tasks into fresh subagent sessions within the same terminal environment, it prevents context pollution and ensures focused, high-quality output. The skill implements a strict 'review-loop' architecture where every code change must pass two distinct checkpoints—specification compliance and code quality—before proceeding. This methodology is ideal for developers who want the speed of AI automation paired with the reliability of a multi-reviewer professional workflow.

主要功能

Automatic task extraction and TodoWrite list creation from existing markdown plans.
Two-stage verification process: spec compliance review followed by code quality review.
1 GitHub stars
Task isolation using fresh subagent sessions to prevent context drift and pollution.
Integrated feedback loops allowing subagents to self-correct based on reviewer feedback.
Seamless session management that maintains progress without needing manual handoffs.

使用场景

Executing multi-step feature implementations from a pre-defined technical specification.
Refactoring modular codebases where tasks are independent and require strict quality gates.
Automating the Test-Driven Development (TDD) cycle across a series of code components.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add astrosteveo/marketplace subagent-driven-development

For use in Claude.ai and ChatGPT

Download Skill

GitHub

关于