subagent-driven-development

Dispatches fresh per-task subagents to implement plans with two-stage reviews.

11|Updated May 17, 2026
One-click install
npx skills add https://github.com/StarryCod/cogitum --skill subagent-driven-development-starrycod
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: subagent-driven-development
Source: https://github.com/StarryCod/cogitum/tree/main/cogitum/data/skills/software-development/subagent-driven-development
Command: npx skills add https://github.com/StarryCod/cogitum --skill subagent-driven-development-starrycod

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Planning and executing complex, multi-task initiatives often suffer from context leakage and misalignment between sub-tasks. This Skill enables coordinating fresh per-task subagents and two-stage reviews to ensure isolation and quality.

Core Features & Use Cases

  • Fresh subagent per task to isolate work and prevent context bleed.
  • Two-stage review process: spec compliance and code quality checks before completion.
  • Plan-driven delegation for multi-step projects with parallelizable tasks.

Quick Start

Provide an implementation plan and let the system dispatch fresh subagents to implement tasks with two-stage 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 multi-task development workflows to prevent context leakage between sub-tasks?

Multi-task development workflows can be automated by dispatching a fresh subagent per task, ensuring isolated context and preventing bleed between independent work items decomposed from a larger plan.

What is a two-stage review process for subagent-driven code implementation?

A two-stage review process for subagent-driven development applies spec compliance checks and code quality reviews sequentially, creating structured progression gates before a task meets its completion criteria.

How do I orchestrate plan-driven delegation for parallelizable software engineering tasks?

Plan-driven delegation for parallelizable tasks is orchestrated by decomposing a larger implementation plan into independent work items and dispatching fresh subagents to execute them with formal reviews.

Do I need to provide a full implementation plan to start delegating tasks to subagents?

Yes, you need to provide a full implementation plan so the system can decompose it into independent work items, dispatch fresh subagents per task, and apply two-stage reviews for structured quality checks.

When should I use isolated subagents for plan execution instead of a single continuous context?

Isolated subagents for plan execution should be used when complex initiatives suffer from context leakage and misalignment, benefiting from isolated execution and formal progression gates between decomposed tasks.