Why are there two separate review stages?

The first stage ensures the code strictly matches the requirements (Spec Compliance), while the second stage focuses on technical excellence, style, and maintainability (Code Quality).

Is this skill suitable for tightly coupled tasks?

This skill works best for implementation plans where tasks are mostly independent. For highly interdependent tasks, manual execution or brainstorming is recommended.

What happens if a subagent fails a task?

The workflow encourages dispatching a fix subagent with specific instructions rather than manual intervention, which keeps the main session's context clean.

How does this differ from standard plan execution?

Unlike parallel session execution, this skill stays within your current session, allowing for faster iteration and immediate feedback without the need for manual context handoffs.

Subagent-Driven Development

Name: Subagent-Driven Development
Author: lastdays03

bylastdays03

0•

Productivity & Workflow

Executes implementation plans by dispatching specialized subagents for individual tasks with multi-stage quality reviews.

Subagent-Driven Development is a high-velocity workflow skill designed to execute complex implementation plans within a single Claude session. By delegating individual, independent tasks to fresh subagents, it prevents context pollution and maintains model focus. Each task undergoes a rigorous two-stage review process—first for specification compliance and then for code quality—ensuring that every feature is built exactly to spec and adheres to best practices before the next task begins. This method bridges the gap between manual execution and fully autonomous agents, providing the speed of automation with the control of session-based development.

Key Features

01Two-stage automated review process covering both Spec Compliance and Code Quality

02Built-in support for Test-Driven Development (TDD) within subagent workflows

03Fresh subagent context per task to prevent model confusion and 'context pollution'

04Integrated plan extraction and progress tracking using TodoWrite

05Same-session execution for faster iteration without the overhead of context switching

060 GitHub stars

Use Cases

01Executing multi-step feature implementations from a pre-defined development plan

02Ensuring high code standards in complex projects through automated multi-stage quality gates

03Accelerating development cycles by automating task dispatching and review loops

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add lastdays03/flask_template subagent-driven-development

For use in Claude.ai and ChatGPT

Download Skill