When should I avoid using Subagent-Driven Development?

Avoid using this skill when tasks are tightly coupled and cannot be separated, or if you do not yet have a formal implementation plan created.

Is it compatible with Test-Driven Development (TDD)?

Yes, subagents naturally follow TDD patterns for each individual task, ensuring that functionality is verified by tests before passing to the review stage.

Why is the two-stage review process necessary?

It ensures the code first meets the functional requirements (spec compliance) before optimizing for maintenance and style (code quality), preventing wasted effort on polishing incorrect features.

How does this differ from standard plan execution?

Unlike standard execution, this skill uses fresh subagents for every task within the same session to prevent context pollution and enforces a mandatory two-stage review process (spec then quality).

Subagent-Driven Development

Name: Subagent-Driven Development
Author: EricGrill

byEricGrill

•

生産性とワークフロー

Executes complex implementation plans by dispatching fresh subagents for individual tasks with a rigorous two-stage review process.

Subagent-Driven Development is a specialized Claude Code skill designed to streamline the execution of multi-step software implementation plans within a single session. By dispatching a dedicated implementer subagent for every task and enforcing a mandatory two-stage review—first for specification compliance and then for code quality—it eliminates context pollution and ensures high-quality outputs. This skill is ideal for executing plans where tasks are relatively independent, providing a high-velocity, automated workflow that maintains rigorous development standards without requiring manual human-in-the-loop coordination between every step.

主な機能

012 GitHub stars

02Pre-implementation clarifying questions to resolve ambiguities before coding begins

03Two-stage verification process ensuring both spec compliance and code quality

04Integrated TodoWrite task tracking for real-time progress monitoring

05Task-specific subagent dispatching to prevent context drift and pollution

06Automatic self-correction loops where implementers fix issues found by reviewers

ユースケース

01Scaling software development tasks while maintaining strict architectural and quality standards

02Automating TDD workflows where individual components require isolated environments

03Executing multi-feature implementation plans within a single development session

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add ericgrill/agents-skills-plugins subagent-driven-development

For use in Claude.ai and ChatGPT

Download Skill