skills.rest
chuangkevin
chuangkevinCOMMUNITY
Explore all skills by this author →
GitHub

subagent-driven-development

Dispatch fresh subagents to implement, test, and commit each task.

Updated Mar 16, 2026
Search Tags:#automation#tdd#code review#subagents#spec compliance#task orchestration#developer-workflow
One-click install
npx skills add https://github.com/chuangkevin/project-bridge --skill subagent-driven-development-chuangkevin
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: subagent-driven-development
Source: https://github.com/chuangkevin/project-bridge/tree/main/skill/superpowers/subagent-driven-development
Command: npx skills add https://github.com/chuangkevin/project-bridge --skill subagent-driven-development-chuangkevin

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill solves the coordination and quality-control gap when executing implementation plans by running isolated subagents per task and enforcing structured review gates so work is implemented, tested, and verified without polluting controller context.

Core Features & Use Cases

  • Fresh subagent per task: dispatch an implementer with only the explicit task text and scene-setting context to avoid context bleed.
  • Two-stage reviews: mandatory spec compliance review followed by code quality review after each implementer completion.
  • TDD and status-driven workflow: implementers write tests, self-review, and report statuses (DONE, DONE_WITH_CONCERNS, NEEDS_CONTEXT, BLOCKED) with clear escalation rules.
  • Model selection guidance: choose lighter models for mechanical tasks and stronger models for integration or architectural decisions.
  • Use Case: Running a multi-task feature plan where each task is implemented, tested, and approved in sequence within the same session, minimizing context switching and catching spec/quality issues early.

Quick Start

Dispatch subagent-driven-development to run the plan: extract all tasks, dispatch an implementer subagent for Task 1, answer any questions, then run spec and code-quality reviewer subagents until approved.

Frequently Asked Questions about subagent-driven-development

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

FAQPage Schema
How do I automate implementation plan execution with isolated subagents?▼

Automating implementation plans with isolated subagents involves dispatching a fresh subagent for each task to implement, test, and commit changes. This ensures fresh context per task, preventing context bleed while maintaining structured handoffs.

What is the best way to enforce spec compliance and code review in automated task orchestration?▼

Spec compliance and code review in automated task orchestration are enforced through a mandatory two-stage review process. After an implementer subagent completes a task, separate reviewer subagents validate spec compliance first, then assess code quality before approval.

How does test-driven development work with subagent-driven task orchestration?▼

Test-driven development with subagent-driven task orchestration requires implementer subagents to write tests, self-review their code, and report statuses like DONE or BLOCKED. This ensures automated verification and clear escalation rules during the development workflow.

Can I select different AI models for subagents based on task complexity?▼

You can select different AI models for subagents based on task complexity. The workflow provides model selection guidance, recommending lighter models for mechanical tasks and stronger models for integration or architectural decisions.

When should I use subagent-driven development for my developer workflow?▼

Subagent-driven development should be used for mostly-independent development tasks requiring automated, repeatable handoffs and verification. It is ideal for multi-task feature plans where catching spec and quality issues early within the same session is critical.

Why does context pollution happen when running multiple implementation tasks sequentially?▼

Context pollution happens when running multiple implementation tasks sequentially because shared controller context accumulates irrelevant data. Dispatching a fresh subagent per task with only explicit task text and scene-setting context eliminates this context bleed.

Related Skills

Explore More Software Engineering Skills→
jacexhjacexh

subagent-driven-development

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

Community
Advanced
mr-daedaliummr-daedalium
107

land-and-deploy

Merge PRs, trigger deployment, and verify canary health in CI/CD pipelines.

Community
Advanced
HuskyDannyHuskyDanny

requesting-code-review

Automate pre-commit code review with static security scanning and auto-fix loops.

Community
Advanced
micaelmaltamicaelmalta
1

workflow

Coordinate multi-phase software development workflows from plan to GitHub PR.

Community
Advanced