subagent-driven-development

Dispatch fresh subagents per task with two-stage reviews.

Updated Jul 8, 2022
One-click install
npx skills add https://github.com/limaon/dotfiles --skill subagent-driven-development-limaon
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: subagent-driven-development
Source: https://github.com/limaon/dotfiles/tree/main/.config/opencode/skills/subagent-driven-development
Command: npx skills add https://github.com/limaon/dotfiles --skill subagent-driven-development-limaon

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill centralizes complex task execution by coordinating multiple lightweight subagents within a single session, reducing context switching and enabling rapid, iterative development.

Core Features & Use Cases

  • Fresh subagent per task: Each task is handled by a new subagent to prevent context pollution and ensure clean state.
  • Two-stage review: After every task, a spec compliance review and a code quality review ensure alignment with requirements and high-quality output.
  • Plan-driven orchestration: Given a multi-task plan, the skill dispatches subagents in sequence (or parallel where appropriate) and tracks progress to completion.
  • Use Case: You have a software feature plan with 5 tasks; the system automatically creates an implementer subagent for Task 1, followed by spec and quality reviewers, before moving to Task 2.

Quick Start

Use the subagent-driven-development skill to execute a plan file (e.g., docs/plans/feature-plan.md), dispatching a fresh implementer subagent per task with two-stage reviews after each task.

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 plan while preventing context pollution?

You can prevent context pollution during multi-task plan execution by deploying a fresh subagent for each task, ensuring isolated state and clean execution before moving to the next item.

What is two-stage review in subagent task orchestration?

Two-stage review in subagent task orchestration dispatches separate spec compliance and code quality reviews after every task, ensuring alignment with requirements and high-quality output.

How do I automate implementer and reviewer dispatch for a feature plan?

You can automate implementer and reviewer dispatch by feeding a feature plan file into subagent-driven-development, which sequentially creates implementer, spec-reviewer, and code-quality-reviewer subagents.

Can I run multiple independent tasks in parallel within the same session?

Yes, you can run multiple independent tasks in parallel within the same session, as the orchestration approach dispatches subagents in parallel where appropriate and tracks progress to completion.

What are the limitations of using fresh subagents for in-session task execution?

The limitation of using fresh subagents for fast in-session tasks is that this approach is restricted to plans with multiple independent tasks, requiring sequential or parallel orchestration within a single session.