subagent-driven-development

Dispatch fresh subagents for independent tasks with two-stage spec and quality reviews.

4|Updated Apr 5, 2026
One-click install
npx skills add https://github.com/Mrlyk/superharness --skill subagent-driven-development-mrlyk
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: subagent-driven-development
Source: https://github.com/Mrlyk/superharness/tree/main/skills/subagent-driven-development
Command: npx skills add https://github.com/Mrlyk/superharness --skill subagent-driven-development-mrlyk

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill addresses the challenges of AI agents losing focus across multiple tasks, producing code that drifts from specifications or lacks quality, by enforcing disciplined execution through isolated subagents and mandatory reviews.

Core Features & Use Cases

  • Fresh Subagents per Task: Dispatches specialized agents with precise context to prevent pollution and ensure focus.
  • Two-Stage Review Process: Automates spec compliance checks followed by code quality reviews, with loops for fixes until approved.
  • Automated Hooks: Handles phase tracking, logging, and context injection without manual intervention.
  • Use Case: When implementing a multi-task feature like building an authentication system, use this Skill to break it into independent tasks, execute them sequentially in one session with subagents handling TDD and reviews for reliable, maintainable code.

Quick Start

Announce you're using the subagent-driven-development skill and dispatch the first implementer subagent with the task details from your plan.

Frequently Asked Questions about subagent-driven-development

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

FAQPage Schema
How do I execute multi-task development plans without AI agents losing focus?

Subagent-driven development executes implementation plans by dispatching fresh, isolated subagents for independent tasks. This prevents context pollution and ensures focused execution across multiple tasks within a single session.

How does the two-stage code review process work in subagent-driven development?

The two-stage review process automates spec compliance checks first, followed by code quality reviews. It loops fixes until approved, ensuring generated code meets specifications and maintains quality standards before proceeding.

Can I use subagent-driven development for disciplined TDD workflows?

Yes, subagent-driven development satisfies requirements for disciplined TDD processes. Subagents handle test-driven development tasks sequentially, applying automated phase tracking and question handling to maintain iterative development flow.

What's the best way to prevent context overload when implementing a multi-task feature?

To prevent context overload when implementing features like authentication systems, break them into independent tasks and execute sequentially in one session. Specialized subagents receive precise context, preventing pollution and maintaining focus.

Do I need automated hooks to handle phase tracking during code implementation?

Automated hooks handle phase tracking, logging, and context injection without manual intervention during subagent-driven development. They also manage question handling and escalation for blockers, ensuring disciplined execution.

When should I not use subagent-driven development for software engineering?

You should avoid subagent-driven development for single, simple tasks that do not require spec compliance checks or multi-stage reviews. It is designed for multi-task features needing focused, iterative development without context overload.