subagent-driven-development

Name: subagent-driven-development
Availability: InStock
Author: beyondwen

Delegate each task to a fresh subagent with isolated context and two-stage reviews.

Updated Mar 30, 2026

Search Tags:#subagent #prompt-templates #ai-agents #plan-execution #two-stage-review #task-isolation

Ready to run

npx skills add https://github.com/beyondwen/track-record-new --skill subagent-driven-development-beyondwen

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: subagent-driven-development
Source: https://github.com/beyondwen/track-record-new/tree/main/.gemini/skills/subagent-driven-development
Command: npx skills add https://github.com/beyondwen/track-record-new --skill subagent-driven-development-beyondwen

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Provides a disciplined approach to executing plans by spawning a fresh subagent for each task, ensuring isolation and reducing context leakage across tasks.

Core Features & Use Cases

One subagent per task to maintain isolation and boundary
Two-stage reviews: spec compliance and code quality to ensure quality
Templates and prompts that guide implementers, spec reviewers, and quality reviewers
Supports parallel execution of independent tasks while preserving context boundaries
Transparent audit trail of each task and its reviews

Quick Start

Create a new subagent for each task, supply the full task text and context, and run the two-stage review workflow to ensure specs and quality before proceeding.

Frequently Asked Questions about subagent-driven-development

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

FAQPage Schema

How do I isolate tasks with subagents to prevent context leakage when executing plans?▼

To isolate tasks with subagents, you delegate each task to a fresh subagent that operates with an isolated context. This approach maintains strict boundaries and prevents context leakage across multiple independent tasks during plan execution.

What is the two-stage review process for AI agents in software engineering?▼

The two-stage review process for AI agents consists of a spec compliance review followed by a code quality review. This enforces discipline and transparency, ensuring high quality across tasks before proceeding.

Can I run parallel tasks with AI agents while preserving context boundaries?▼

Yes, you can run parallel tasks with AI agents while preserving context boundaries. By spawning a fresh subagent for each independent task, the workflow supports parallel execution without cross-task interference.

How do I use prompt templates to guide AI agent task execution and reviews?▼

You use prescribed prompt templates to guide implementers, spec reviewers, and quality reviewers through the workflow. Enforcing these templates maintains discipline and ensures transparent audit trails for each task.

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

You should not use subagent-driven development for sessions lacking multiple independent tasks. The workflow is designed specifically for plans requiring parallel work and isolated task execution to reduce context leakage.