skills.rest
FerranGuardia
FerranGuardiaCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

subagent-driven-development

Dispatch independent subagents for plan tasks with two-stage reviews.

Updated Mar 2, 2026
Search Tags:#automation#subagent#task-management#prompt-engineering#ai-agents#plan-execution#two-stage-review
One-click install
npx skills add https://github.com/FerranGuardia/claude-autonomous-setup --skill subagent-driven-development-ferranguardia
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: subagent-driven-development
Source: https://github.com/FerranGuardia/claude-autonomous-setup/tree/main/skills/methodology/subagent-development
Command: npx skills add https://github.com/FerranGuardia/claude-autonomous-setup --skill subagent-driven-development-ferranguardia

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Orchestrates complex plan execution by spawning a fresh subagent per task, with a built-in two-stage review to ensure spec alignment and high code quality in the same session.

Core Features & Use Cases

  • Fresh per-task subagents: each task runs in isolation to avoid context pollution.
  • Two-stage review: spec compliance review first, then code quality review.
  • Guided workflow: prompts and templates enforce a repeatable, auditable process for plan execution.

Quick Start

Initiate plan execution and dispatch a fresh subagent per task, then run a two-stage review for each task.

Frequently Asked Questions about subagent-driven-development

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

FAQPage Schema
How do I execute a development plan with independent AI agents?▼

Executing a development plan with independent AI agents involves dispatching a fresh subagent per task to maintain isolated context, followed by a two-stage review for spec compliance and code quality. This prevents context pollution during rapid iteration.

What is a two-stage review process for AI-assisted code generation?▼

A two-stage review process for AI-assisted code generation enforces spec compliance first, then evaluates code quality. This guided workflow ensures generated code aligns with requirements and meets quality standards before completing the task handoff.

How can I prevent context pollution when running multiple AI agent tasks in a single session?▼

Preventing context pollution when running multiple AI agent tasks requires spawning fresh per-task subagents. Each task runs in isolation to avoid context bleed, enabling rapid iteration and accurate plan execution without degrading agent performance.

Does subagent-driven development work for rapid iteration and spec reviews?▼

Subagent-driven development works for rapid iteration and spec reviews by dispatching independent subagents per task. It enforces a repeatable, auditable process with dedicated implementer and reviewer prompts to ensure high code quality.

When should I use isolated subagents for task management instead of a single agent?▼

Use isolated subagents for task management when tasks are mostly independent within a single session and rapid iteration is needed. This approach maintains isolated context per task, making it ideal for plan execution and code-quality checks.

Related Skills

Explore More Software Engineering Skills→
badaniyabadaniya
1

subagent-driven-development

Dispatch a fresh subagent per task and run two-stage reviews.

Community
Intermediate
ZhiruiFengZhiruiFeng
1

testing

Generate and execute unit, integration, and end-to-end test suites.

Community
Advanced
obraobra
41

Subagent-Driven Development

Dispatch a fresh subagent for each task and trigger code review after completion.

Community
Basic
alffeialffei
69

subagent-driven-development

Dispatch fresh subagents per task with isolated context and two-stage reviews.

Community
Advanced