subagent-driven-development

Orchestrate subagents for plan execution with two-stage spec and quality reviews.

66|20|Updated Dec 30, 2025
One-click install
npx skills add https://github.com/cooker/wxHm --skill subagent-driven-development-cooker
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: subagent-driven-development
Source: https://github.com/cooker/wxHm/tree/main/.cursor/skills/subagent-driven-development
Command: npx skills add https://github.com/cooker/wxHm --skill subagent-driven-development-cooker

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill enables executing complex implementation plans by dividing work into independent subagents within the same session, reducing context-switching and accelerating iteration.

Core Features & Use Cases

  • Fresh subagents per task to prevent context pollution and enable parallel-like progress within a single session.
  • Two-stage review for each task: spec compliance first, then code quality, ensuring correct scope and high-quality output.
  • Structured task orchestration across a multi-step plan, with per-task context kept isolated.
  • Ready-to-use prompt templates for implementer, spec compliance reviewer, and code quality reviewer to standardize workflows.
  • Ideal for teams needing fast, iterative development without leaving the current session or losing context.

Quick Start

Dispatch an initial implementer subagent for Task 1 with the full task text and context, then route through the two-stage review loop (spec compliance, then code quality) before proceeding to subsequent tasks. Example: initiate implementer-prompt.md for Task 1, then trigger spec-reviewer-prompt.md, followed by code-quality-reviewer-prompt.md, and iterate until all tasks are completed.

Frequently Asked Questions about subagent-driven-development

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

FAQPage Schema
What is the best way to scale complex tasks with subagents?

Scale complex tasks by orchestrating independent subagents within a single session. Dispatch a fresh implementer subagent for each task, then route it through a two-stage review loop for spec and quality validation.

How do I execute complex implementation plans without context switching in AI development?

Subagent-driven development executes complex implementation plans by dispatching fresh, independent subagents per task. This prevents context pollution and enables parallel-like progress within a single session, accelerating iteration.

What is two-stage review in AI code generation workflows?

Two-stage review validates each task first for spec compliance to ensure correct scope, then for code quality to ensure high output. This structured validation loop routes implementer output through spec and quality reviewers.

How do I prevent context pollution when orchestrating multi-step plans?

Prevent context pollution by dispatching fresh subagents for each task step. Per-task context is kept isolated, ensuring previous steps do not degrade the reasoning quality of subsequent plan execution.

Can I standardize spec compliance and code quality checks in a single session?

Yes, you can standardize checks using ready-to-use prompt templates for implementers, spec compliance reviewers, and code quality reviewers. This standardizes the workflow for fast, iterative development without losing session context.

What is the best way to scale complex tasks with subagents?

Scale complex tasks by orchestrating independent subagents within a single session. Dispatch a fresh implementer subagent for each task, then route it through a two-stage review loop for spec and quality validation.