subagent-driven-development

Dispatch fresh subagents to execute independent tasks from an implementation plan.

Updated Aug 23, 2026
One-click install
npx skills add https://github.com/rapyuta-robotics/agent-ai --skill subagent-driven-development-rapyuta-robotics
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: subagent-driven-development
Source: https://github.com/rapyuta-robotics/agent-ai/tree/main/skills/subagent-driven-development
Command: npx skills add https://github.com/rapyuta-robotics/agent-ai --skill subagent-driven-development-rapyuta-robotics

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill orchestrates plan execution by dispatching fresh subagents for each task within the same session, ensuring isolated context and rapid iteration.

Core Features & Use Cases

  • Fresh subagents per task with two-stage reviews (spec compliance, then code quality)
  • Parallel task execution with explicit task tracking
  • Use Case: Running a multi-step feature from a plan with independent tasks and consolidated progress

Quick Start

Run a prepared implementation plan through the subagent workflow to execute parallel tasks with per-task 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 independent tasks from an implementation plan in parallel?

To execute independent tasks in parallel, this Skill dispatches fresh subagents for each task within the same session, ensuring isolated context and rapid iteration. It applies plan-driven sequencing to orchestrate multi-step feature execution.

What is a two-stage review in subagent-driven software engineering?

A two-stage review in subagent-driven software engineering evaluates dispatched subagent output sequentially, first checking spec compliance against the implementation plan, then assessing code quality to ensure structured quality control.

How do I run a multi-step feature plan with isolated context for each task?

You can run a multi-step feature plan with isolated context by dispatching fresh subagents per task. This enforces per-task context isolation while maintaining consolidated progress tracking across the entire workflow.

Does parallel task execution work within the same session for software engineering workflows?

Yes, parallel task execution is applied within the same session. It dispatches fresh subagents to execute independent tasks concurrently, followed by structured reviews to ensure quality after each task completion.

What is the best way to automate plan-driven task sequencing with code reviews?

The best way to automate plan-driven task sequencing is using a subagent workflow that enforces plan-driven sequencing and structured two-stage reviews, ensuring both spec compliance and code quality for every dispatched task.

When should I not use fresh subagents for parallel task execution?

You should not use fresh subagents for parallel task execution if your implementation plan lacks independent tasks. The workflow relies on per-task context isolation and plan-driven sequencing, making it unsuitable for highly interdependent task chains.