skills.rest
pedropaulovc
pedropaulovcCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

subagent-driven-development

Coordinate a persistent agent team executing implementation plans with adversarial testing and review.

3|Updated Feb 2, 2026
Search Tags:#tdd#code review#implementation planning#e2e testing#spec compliance#agent teams#demo validation
One-click install
npx skills add https://github.com/pedropaulovc/agent-plugins --skill subagent-driven-development-pedropaulovc
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: subagent-driven-development
Source: https://github.com/pedropaulovc/agent-plugins/tree/main/plugins/superpowers/skills/subagent-driven-development
Command: npx skills add https://github.com/pedropaulovc/agent-plugins --skill subagent-driven-development-pedropaulovc

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Prevents slow, single-threaded implementation by coordinating an agent team that executes plan tasks with adversarial testing and review to keep spec compliance and code quality from drifting.

Core Features & Use Cases

  • Parallel subagent execution: Runs independent tasks with persistent teammates so context survives across fix rounds.
  • TDD-style RED→GREEN loop per task: Tester writes failing e2e/integration tests first, implementer makes them pass while adding unit tests, then both verify again after any change.
  • Spec + code-quality gate before demo: Code reviewer validates requirements and maintainability, and demo review is judged strictly from the demo artifacts.
  • Best for: Implementation plan execution where tasks are mostly independent, and you want reliable progress through repeated fix cycles without losing context.

Quick Start

Use this skill when you have an execution-ready implementation plan and the tasks can mostly run independently in the current session.

Frequently Asked Questions about subagent-driven-development

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

FAQPage Schema
How do I coordinate parallel TDD cycles for independent implementation tasks?▼

Run parallel TDD with an agent team where a tester writes failing e2e/integration tests first, an implementer makes them pass, and a reviewer checks spec compliance before demo. Persistent teammates retain context across repeated fix rounds.

What is the RED to GREEN testing loop in subagent-driven development?▼

The RED to GREEN TDD loop involves a tester agent writing failing e2e/integration tests, an implementer agent writing code and unit tests to pass them, and both verifying again after any subsequent fix-loop changes.

How do I automate spec compliance and code quality review before feature demo?▼

Automate spec compliance and code-quality review by assigning a code reviewer agent to validate requirements and maintainability. The final demo review is judged strictly from the generated demo artifacts.

Does parallel subagent execution require CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS?▼

Yes, parallel subagent execution requires CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS enabled. It uses TeamCreate, persistent teammates, and coordinates work via TaskCreate, TaskUpdate, TaskList, and SendMessage.

When should I avoid using parallel agent teams for implementation planning?▼

Avoid using parallel agent teams for implementation planning when tasks are highly interdependent. This approach requires mostly independent tasks to execute concurrent TDD cycles and achieve reliable progress through fix-loop coordination.

Related Skills

Explore More Software Engineering Skills→
Driving-TeacherDriving-Teacher
1

team-assemble

Decompose tasks into roles and orchestrate expert teams via TeamCreate.

Official
Advanced
45ck45ck
6

heal-spec

Patch broken demo-machine specs by analyzing failure artifacts and validating with the CLI.

Community
Advanced
wodsmithwodsmith
2

test

Route testing questions to unit, integration, or E2E test types.

Official
Advanced
greglambgreglamb
1

subagent-driven-development

Execute implementation plans via isolated subagents with two-stage review gates.

Community
Advanced