skills.rest
aberrantCode
aberrantCodeCOMMUNITY
Explore all skills by this author →
GitHub

subagent-driven-development

Execute implementation plans with fresh subagents and two-stage reviews.

1|Updated Feb 26, 2026
Search Tags:#automation#subagent#software-development#code-review#sequential#spec-compliance
One-click install
npx skills add https://github.com/aberrantCode/llm_skills --skill subagent-driven-development-aberrantcode
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: subagent-driven-development
Source: https://github.com/aberrantCode/llm_skills/tree/main/claude/skills/subagent-driven-development
Command: npx skills add https://github.com/aberrantCode/llm_skills --skill subagent-driven-development-aberrantcode

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It eliminates context pollution and manual coordination when executing implementation plans composed of independent tasks, ensuring each task is completed with high quality and proper review.

Core Features & Use Cases

  • Fresh subagent per task: Each task runs in an isolated subagent to avoid context bleed.
  • Two‑stage review: Automatic spec‑compliance review followed by a code‑quality review for every task.
  • Sequential task handling: The controller extracts all tasks up‑front and processes them one by one, with automatic loops until reviews approve.
  • Integration with superpowers: Works with writing-plans, requesting-code-review, and finishing-a-development-branch to provide a full end‑to‑end development workflow.
  • Use case: Apply to a feature plan where tasks are independent and need consistent quality checks without manual hand‑offs.

Quick Start

Run subagent-driven-development on the plan file 'feature-plan.md' to automatically implement all tasks with built‑in 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 automate implementation plan execution while preventing context bleed between tasks?▼

Automating implementation plan execution while preventing context bleed requires fresh subagents for each task. This Skill isolates context per task, executing sequentially while enforcing mandatory spec and code quality reviews.

What is the two-stage review process for automated software development?▼

The two-stage review process for automated software development consists of an automatic spec-compliance review followed by a code-quality review. This Skill enforces both stages for every task, looping until reviews approve.

How do I execute independent feature plan tasks sequentially with automated quality checks?▼

To execute independent feature plan tasks sequentially with automated quality checks, the controller extracts all tasks up-front and processes them one by one. It enforces dual reviews and loops until approvals pass.

Does subagent-driven development work with code review and plan writing workflows?▼

Subagent-driven development works with code review and plan writing workflows by integrating with writing-plans, requesting-code-review, and finishing-a-development-branch. This provides a full end-to-end development workflow.

When should I use fresh subagents for sequential task execution?▼

You should use fresh subagents for sequential task execution when applying to a feature plan where tasks are independent and need consistent quality checks. It suits projects requiring reliable outcomes without manual hand-offs.

Why does context pollution occur during implementation plan execution and how can subagents fix it?▼

Context pollution occurs during implementation plan execution when task contexts bleed together. Fresh subagents fix it by isolating each task, eliminating manual coordination and ensuring reliable, high-quality outcomes through automated reviews.

Related Skills

Explore More Software Engineering Skills→
pantheon-orgpantheon-org
1

subagent-driven-development

Dispatch fresh subagents for each task with two-stage reviews.

Official
Basic
m16khbm16khb

finishing-a-development-branch

Run the test suite and present four options for merging, PR, keeping, or discarding a branch.

Community
Advanced
nachopalmerinachopalmeri

executing-plans

Load, validate, and execute written plans with checkpoints and final review.

Community
Advanced
CodingCossackCodingCossack
19

subagent-driven-development

Executes sequential subagents with two-stage reviews for multi-task plans.

Community
Advanced