subagent-driven-development

Dispatch fresh subagents per task with two-stage spec and code quality reviews.

2|Updated Apr 8, 2026
One-click install
npx skills add https://github.com/miptah21/skills --skill subagent-driven-development-miptah21
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: subagent-driven-development
Source: https://github.com/miptah21/skills/tree/main/.agents/skills/subagent-driven-development
Command: npx skills add https://github.com/miptah21/skills --skill subagent-driven-development-miptah21

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

This Skill helps you execute an implementation plan efficiently while maintaining high quality by delegating each task to a fresh subagent and validating results with two independent review stages.

Core Features & Use Cases

  • Fresh subagent per task: Prevents context pollution by ensuring each task runs with only the instructions and context you provide.
  • Two-stage review loop: Enforces spec compliance review first, then code quality review, repeating fixes until approvals pass.
  • Continuous execution: Keeps momentum by avoiding “should I continue?” prompts and executing the full plan unless blocked by ambiguity or an unresolvable blocker.
  • Status-driven recovery: Handles DONE, DONE_WITH_CONCERNS, NEEDS_CONTEXT, and BLOCKED with the appropriate escalation and re-dispatch behavior.

Quick Start

Use subagent-driven-development to execute each task in your implementation plan by dispatching an implementer subagent, then a spec compliance reviewer, then a code quality reviewer, repeating fixes until both reviews approve.

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 an implementation plan continuously without manual check-ins?

Subagent-driven development executes an implementation plan continuously by dispatching isolated subagent work per task, avoiding manual check-ins unless blocked by ambiguity or unresolvable blockers.

What is the best way to prevent context pollution when delegating coding tasks?

Preventing context pollution during coding tasks requires dispatching a fresh subagent per task, ensuring each isolated execution runs only with the specific instructions and context provided.

How does two-stage code review work for automated task execution?

Two-stage code review enforces spec compliance verification first, then runs a separate code quality review, looping fixes based on reviewer findings until both reviews approve the task.

Can I use subagent-driven development for tasks that are mostly independent within the same session?

Yes, subagent-driven development is suitable for scenarios where tasks are mostly independent within the same session and need continuous progress without requiring manual check-ins.

How does status-driven recovery handle blocked or incomplete tasks?

Status-driven recovery handles DONE, DONE_WITH_CONCERNS, NEEDS_CONTEXT, and BLOCKED statuses with appropriate escalation and re-dispatch behavior to resolve incomplete tasks.