skills.rest
withmartian-sandbox-darkside
withmartian-sandbox-darksideOFFICIAL
Explore all skills by this author →
GitHub

subagent-driven-development

Execute tasks with fresh subagents and enforce parallel code reviews.

Updated Aug 23, 2026
Search Tags:#automation#tdd#subagent#code-review#task-execution#parallel-review
One-click install
npx skills add https://github.com/withmartian-sandbox-darkside/ghrc-y-3126672651424eddb640ecc81321a665 --skill subagent-driven-development-withmartian-sandbox-darkside
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: subagent-driven-development
Source: https://github.com/withmartian-sandbox-darkside/ghrc-y-3126672651424eddb640ecc81321a665/tree/main/default/skills/subagent-driven-development
Command: npx skills add https://github.com/withmartian-sandbox-darkside/ghrc-y-3126672651424eddb640ecc81321a665 --skill subagent-driven-development-withmartian-sandbox-darkside

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill automates execution of development plans by dispatching fresh subagents per task and enforcing parallel code review after each task to maintain high quality and continuous progress without human-in-loop pauses.

Core Features & Use Cases

  • Fresh subagent per task: Isolates context to avoid pollution and ensures reproducible task runs.
  • Parallel enforced reviews: Launches code-reviewer, business-logic-reviewer, and security-reviewer simultaneously to surface issues quickly.
  • Severity-driven remediation loop: Aggregates Critical/High/Medium issues to trigger fix subagents and records Low/Cosmetic items as TODO/FIXME for later.
  • TDD & integration: Subagents follow test-driven-development and the workflow culminates in a finishing-a-development-branch step.
  • Use Case: Ideal for same-session iterative implementation where tasks are independent and require strict quality gates and automated review cycles.

Quick Start

Run the subagent-driven-development skill to execute plan tasks sequentially using fresh subagents per task, run three parallel reviewers, and iterate fixes until all reviews pass.

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 task execution with enforced code review in a development workflow?▼

Automated task execution with enforced code review dispatches a fresh, isolated subagent per task and runs parallel code, business-logic, and security reviewers to maintain quality without human-in-loop pauses.

What is the best way to run parallel code reviews on autonomous development tasks?▼

Parallel code reviews on autonomous tasks are best run by simultaneously launching code-reviewer, business-logic-reviewer, and security-reviewer using opus models, then aggregating severity-based issues for remediation.

How does severity-driven remediation work with automated subagent code reviews?▼

Severity-driven remediation aggregates Critical, High, and Medium issues to trigger automated fix subagents, while recording Low and Cosmetic items as TODO or FIXME comments to address later.

Can I use test-driven development with autonomous subagents for continuous implementation?▼

Yes, autonomous subagents can enforce test-driven development during task execution, ensuring continuous progress through strict quality gates before culminating in a finishing-a-development-branch integration step.

Does dispatching a fresh subagent per task prevent context pollution in iterative development?▼

Dispatching a fresh subagent per task isolates context to avoid pollution, ensuring reproducible task runs and maintaining clean execution boundaries throughout same-session iterative implementation.

When should I not use parallel subagent-driven development for task execution?▼

Subagent-driven development is not ideal for tasks that are highly interdependent or require continuous human-in-loop oversight, as it relies on independent tasks and automated parallel quality gates.

Related Skills

Explore More Software Engineering Skills→
Dq666dingerDq666dinger
1

subagent-driven-development

Delegates each plan task to fresh subagents with two-stage spec and quality reviews.

Community
Advanced
costrict-plugins-repocostrict-plugins-repo

executing-plans

Execute written implementation plans with critical review checkpoints.

Official
Advanced
LerianStudioLerianStudio
205

requesting-code-review

Automate sequential code reviews with specialized subagents and structured markdown output.

Official
Advanced
kk20300113-pngkk20300113-png

subagent-driven-development

Dispatch subagents to execute implementation plan tasks with two-stage reviews.

Community
Advanced