subagent-driven-development

Coordinate sequential subagents with isolated contexts and two-stage reviews.

4|Updated Mar 15, 2024
One-click install
npx skills add https://github.com/izyanrajwani/agent-skills-library --skill subagent-driven-development-izyanrajwani
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: subagent-driven-development
Source: https://github.com/izyanrajwani/agent-skills-library/tree/main/skills/subagent-driven-development
Command: npx skills add https://github.com/izyanrajwani/agent-skills-library --skill subagent-driven-development-izyanrajwani

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Sequential orchestration of subagents to handle multi-task plans with isolated contexts and staged reviews for compliance and quality.

Core Features & Use Cases

  • Fresh subagent per task to prevent context pollution
  • Two-stage review gates (spec compliance, code quality)
  • Diff-based validation and per-task BASE_SHA and HEAD_SHA tracking
  • Safe parallel plan execution via isolated work contexts

Quick Start

Dispatch a fresh subagent per task and perform a two-stage review after each implementation to validate changes.

Frequently Asked Questions about subagent-driven-development

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

FAQPage Schema
How do I orchestrate subagents to execute multi-task plans without context pollution?

Sequential subagent orchestration dispatches a fresh subagent per task to prevent context pollution. Each subagent executes in an isolated context, ensuring sequential tasks remain independent and maintain clean boundaries throughout the workflow.

How does diff-based validation work for multi-task subagent workflows?

Diff-based validation tracks per-task BASE_SHA and HEAD_SHA references to isolate code changes. This mechanism enforces strict validation by verifying only the modifications made within each specific isolated task boundary before proceeding.

What is a two-stage review gate in subagent-driven development?

A two-stage review gate validates each completed task through spec compliance and code quality checks. This staged review process ensures implementations meet requirements and maintain standards before moving to the next sequential subagent.

Can I execute parallel multi-task plans safely using isolated subagents?

Safe parallel plan execution is supported via isolated work contexts. Each parallel plan operates independently with its own fresh subagents, preventing cross-contamination between concurrent workflows while maintaining sequential integrity within each plan.

When should I use sequential subagent orchestration instead of a single agent?

Sequential subagent orchestration is ideal when multi-task plans require fresh context per task and staged reviews for compliance. Use it when complex workflows risk context pollution or need strict diff-based validation between implementation steps.