How does the two-stage review process work?

First, a 'Spec Reviewer' ensures the code perfectly matches the requirements. Once passed, a 'Code Quality Reviewer' checks for best practices, magic numbers, and maintainability before the task is marked complete.

What is the benefit of using fresh subagents per task?

Using a fresh subagent for each task eliminates 'context pollution,' ensuring the AI focuses only on the specific requirements and current code state of the task at hand without being distracted by previous iterations.

Can this skill handle tasks that require multiple iterations?

Yes. If a reviewer identifies issues, the implementer subagent is tasked with fixing them, and the review cycle repeats until both specification and quality standards are met.

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

Unlike standard execution which often happens in parallel or separate sessions, this skill runs within the same session for faster iteration, allowing for immediate questions and answers without a handoff.

Subagent-Driven Development

Name: Subagent-Driven Development
Author: claudiodearaujo

byclaudiodearaujo

0•

Productivity & Workflow

Executes complex implementation plans using specialized subagents for task implementation and rigorous two-stage reviews.

Subagent-Driven Development is a specialized workflow for Claude that automates the execution of multi-task implementation plans within a single session. By dispatching a fresh subagent for every individual task, the skill prevents context pollution and ensures high-focus implementation. Each task undergoes a mandatory two-stage review process: first, a specification reviewer verifies that all requirements are met without over-engineering; second, a code quality reviewer ensures the implementation follows best practices. This structured approach facilitates rapid iteration, minimizes bugs, and maintains high standards without requiring constant human intervention between development steps.

Key Features

01Automated task extraction and TodoWrite progress tracking

02Two-stage automated review: Specification Compliance and Code Quality

03Integrated Test-Driven Development (TDD) enforcement for subagents

04Isolated subagent dispatch per task to prevent context drift

05Same-session execution to eliminate handoff delays and context switching

060 GitHub stars

Use Cases

01Executing detailed feature plans with multiple independent coding tasks

02Managing development branches that require strict adherence to technical specifications

03Automating high-quality code reviews for complex system refactors

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add claudiodearaujo/izacenter subagent-driven-development

For use in Claude.ai and ChatGPT

Download Skill