subagent-driven-development

Dispatch a fresh subagent per task with two-stage spec and code reviews.

1|Updated Nov 22, 2025
One-click install
npx skills add https://github.com/gmliao/swift-state-tree --skill subagent-driven-development-gmliao
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: subagent-driven-development
Source: https://github.com/gmliao/swift-state-tree/tree/main/.agent/skills/Superpowers/subagent-driven-development
Command: npx skills add https://github.com/gmliao/swift-state-tree --skill subagent-driven-development-gmliao

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill orchestrates task work by dispatching a fresh subagent per task within the current session.

Core Features & Use Cases

  • Fresh subagent per task with isolated context.
  • Two-stage review workflow (spec compliance followed by code quality).
  • Seamless planning to execution flow in single-session environments.

Quick Start

Use this skill to process a plan by dispatching a fresh subagent per task, perform spec and quality reviews, and complete tasks within this session.

Frequently Asked Questions about subagent-driven-development

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run independent coding tasks in a single session using subagents?

Dispatching a fresh subagent per task within a single session isolates context for each independent task, enabling fast iteration with structured two-stage reviews for spec compliance and code quality.

What is a two-stage review workflow for code quality and spec compliance?

A two-stage review workflow first evaluates completed task output against the original specification, then separately assesses code quality, enforcing structured reporting and completion before results are collected.

How do I orchestrate a development plan with multiple independent tasks?

A coordinating controller distributes plan tasks to fresh subagents, collects their isolated results, and enforces reporting to ensure seamless planning to execution flow in single-session environments.

When should I use a fresh subagent per task instead of a single continuous context?

Use a fresh subagent per task when your plan contains largely independent tasks and you need fast iteration with isolated context to prevent cross-task interference during execution.

Do I need prompt templates to dispatch subagents for task work?

Yes, the skill relies on prompt templates and a coordinating controller to distribute tasks, collect results, and enforce structured reporting and completion for each dispatched subagent.

Can I use subagent-driven development for tasks with heavy dependencies on each other?

This approach is ideal when plans contain largely independent tasks; tasks with heavy cross-dependencies may not benefit from the isolated context and fast iteration this subagent dispatch model provides.