Do I need a plan before starting this skill?

Yes, this skill is designed to execute an existing implementation plan. It works best when paired with a planning skill like 'superpowers:writing-plans'.

How does the two-stage review process work?

Each task is first reviewed by a spec-reviewer subagent to ensure all requirements are met. Once approved, a code-quality reviewer performs a second audit for best practices and performance.

What is the benefit of using a fresh subagent for every task?

Using a fresh subagent prevents 'context pollution,' ensuring the AI doesn't get confused by previous tasks' code or history. This leads to higher focus and fewer errors.

How is this different from the 'executing-plans' skill?

Subagent-Driven Development happens entirely within the current session for faster iteration, whereas 'executing-plans' typically involves parallel sessions.

Subagent-Driven Development

Name: Subagent-Driven Development
Author: Krosebrook

byKrosebrook

•

生産性とワークフロー

Executes implementation plans by dispatching fresh subagents for each task with a rigorous two-stage review process.

概要

Subagent-Driven Development is a high-efficiency workflow for Claude Code designed to execute complex implementation plans within a single session. By dispatching a dedicated subagent for every task, the skill eliminates context pollution and ensures deep focus on specific requirements. Every task follows a strict two-stage review protocol—validating specification compliance first, followed by a code quality audit—to ensure production-grade output. This systematic approach combines the speed of AI automation with the rigor of professional quality gates, making it ideal for rapid feature development and codebase maintenance within large-scale projects.

主な機能

Mandatory two-stage review (Specification Compliance and Code Quality)
1 GitHub stars
Task-specific subagent dispatching to prevent context drift
Continuous session execution without the need for manual handoffs
Automated task tracking and status updates using TodoWrite
Integrated Test-Driven Development (TDD) via implementer subagents

ユースケース

Enforcing strict adherence to technical specifications in professional monorepos
Executing multi-task feature plans where components are largely independent
Rapidly prototyping complex modules while maintaining high quality standards

概要

主な機能

Mandatory two-stage review (Specification Compliance and Code Quality)
1 GitHub stars
Task-specific subagent dispatching to prevent context drift
Continuous session execution without the need for manual handoffs
Automated task tracking and status updates using TodoWrite
Integrated Test-Driven Development (TDD) via implementer subagents

ユースケース

Enforcing strict adherence to technical specifications in professional monorepos
Executing multi-task feature plans where components are largely independent
Rapidly prototyping complex modules while maintaining high quality standards