subagent-driven-development

Execute implementation plans by dispatching fresh subagents for each task.

1|Updated Mar 29, 2026
One-click install
npx skills add https://github.com/dmedina-dev/dev-forge --skill subagent-driven-development-dmedina-dev
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: subagent-driven-development
Source: https://github.com/dmedina-dev/dev-forge/tree/main/plugins/forge-superpowers/skills/subagent-driven-development
Command: npx skills add https://github.com/dmedina-dev/dev-forge --skill subagent-driven-development-dmedina-dev

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Manual execution of multi‑task implementation plans often pollutes the session context and skips systematic reviews, leading to lower code quality and missed specifications.

Core Features & Use Cases

  • Fresh subagent per task: Each independent task is dispatched to a brand‑new subagent, ensuring isolated context.
  • Two‑stage reviews: Automatic spec‑compliance review followed by a code‑quality review for every task.
  • Status handling: Recognizes DONE, DONE_WITH_CONCERNS, NEEDS_CONTEXT, and BLOCKED to adapt workflow dynamically.
  • Integration: Works with superpowers such as git worktrees, TDD, and final branch finishing.
  • Use case: Execute a feature‑development plan that contains several loosely coupled tasks, keeping the main session clean while guaranteeing thorough reviews.

Quick Start

Invoke the subagent-driven-development skill on your current implementation plan to start executing tasks with built‑in spec and quality reviews.

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 a multi-task implementation plan without polluting my session context?

To execute a multi-task implementation plan without context pollution, dispatch a fresh subagent for each independent task. This isolates contexts automatically, keeping the main session clean while processing loosely coupled tasks.

Why does context pollution happen when running multiple development tasks in one session?

Context pollution happens when multiple development tasks run sequentially in one session because shared context accumulates irrelevant details. Dispatching a fresh subagent per task isolates the context, preventing this pollution and maintaining execution quality.

Can I automate code reviews and spec compliance checks during feature development?

Yes, you can automate code reviews and spec compliance checks during feature development by using a two-stage review system. It automatically performs a spec-compliance review followed by a code-quality review for every dispatched task.

What's the best way to handle loosely coupled tasks in a software engineering workflow?

The best way to handle loosely coupled tasks in a software engineering workflow is dispatching a fresh subagent for each task. This isolates execution and enables automatic two-stage reviews for spec compliance and code quality.

Does subagent-driven-development work with git worktrees and TDD?

Yes, subagent-driven-development works with git worktrees and TDD. It integrates with these superpowers to execute plans using fresh subagents, providing automatic reviews and final branch finishing capabilities.

How do I manage blocked tasks or tasks needing more context during automated execution?

To manage blocked tasks or tasks needing context during automated execution, the workflow dynamically adapts using status handling. It recognizes DONE, DONE_WITH_CONCERNS, NEEDS_CONTEXT, and BLOCKED statuses to adjust the execution flow accordingly.