How does this differ from standard plan execution?

Unlike standard execution which might handle tasks sequentially in a single context, this skill dispatches fresh subagents for every task to prevent context drift and uses a multi-layered review process to ensure quality.

Can subagents ask questions during implementation?

Yes. The workflow encourages implementer subagents to surface questions before and during work, ensuring they have the necessary context before committing code.

What are the two stages of the review process?

The first stage checks for Spec Compliance (did the agent build what was asked?), and the second stage checks for Code Quality (is the code well-written, efficient, and maintainable?).

Does this work in the same Claude Code session?

Yes, it is designed for same-session execution, allowing for faster iteration and continuous progress without the overhead of switching contexts or sessions.

Subagent-Driven Development

Name: Subagent-Driven Development
Author: BbgnsurfTech

byBbgnsurfTech

•

Productivity & Workflow

Orchestrates specialized subagents to execute implementation plans with mandatory two-stage reviews for spec compliance and code quality.

Subagent-Driven Development is a sophisticated workflow skill designed for executing complex technical plans within a single Claude Code session. By dispatching a fresh, task-specific subagent for every item in an implementation plan, it prevents context pollution and ensures modular code. Each task undergoes a rigorous two-stage validation process: first, a spec-compliance reviewer ensures all requirements are met, followed by a code-quality reviewer who audits the implementation for best practices. This automated feedback loop results in high-integrity code, rapid iteration, and minimized human intervention during the development cycle.

Key Features

01Centralized task tracking and progress management via TodoWrite

02Seamless integration with Test-Driven Development (TDD) workflows

03Automated feedback loops between implementers and reviewers

04Fresh subagent dispatching per task to maintain clean context

053 GitHub stars

06Mandatory two-stage review process (Spec Compliance and Code Quality)

Use Cases

01Scaling development output by automating high-quality code reviews and fixes

02Executing multi-feature implementation plans within a single terminal session

03Enforcing strict adherence to technical specifications during complex refactors

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add bbgnsurftech/claude-skills-collection subagent-driven-development

For use in Claude.ai and ChatGPT

Download Skill