What happens if a subagent fails a task?

The workflow dispatches a specific 'fix' subagent with the feedback from the failure or code review, ensuring the issue is resolved within a clean context before proceeding to the next plan item.

How does this differ from the standard 'executing-plans' skill?

While 'executing-plans' often uses parallel sessions, Subagent-Driven Development operates within your current session. This allows for faster iteration and easier monitoring while still preventing context pollution by using fresh subagents for each task.

Why is code review mandatory between tasks?

Reviewing after every task acts as a quality gate, catching critical issues early before they compound in later stages of the development plan, ultimately saving time on debugging.

Does this skill require a pre-existing plan?

Yes, this skill is specifically designed to execute plans created by the 'writing-plans' skill or similar structured Todo lists.

Subagent-Driven Development

Name: Subagent-Driven Development
Author: lifangda

bylifangda

•

生産性とワークフロー

Executes complex development plans by dispatching specialized subagents for individual tasks with integrated code review checkpoints.

概要

Subagent-Driven Development is an advanced workflow skill for Claude Code designed to execute multi-step implementation plans with high precision and quality. By dispatching a fresh subagent for every discrete task and requiring an automated code review before moving to the next, it prevents context pollution and ensures each component meets architectural standards. This approach allows for fast, continuous iteration within a single session while maintaining the rigor of a multi-stage engineering pipeline.

主な機能

Context-isolated task execution using fresh subagent dispatches
Single-session workflow management to avoid unnecessary context switching
16 GitHub stars
Final comprehensive architectural review upon plan completion
Automated code review checkpoints after every task implementation
Enforced Test-Driven Development (TDD) patterns for subagent tasks

ユースケース

Executing multi-file feature implementations from a structured development plan
Refactoring complex systems where task isolation is required to prevent regression
Rapidly building and testing new modules with automated quality gates

概要

主な機能

Context-isolated task execution using fresh subagent dispatches
Single-session workflow management to avoid unnecessary context switching
16 GitHub stars
Final comprehensive architectural review upon plan completion
Automated code review checkpoints after every task implementation
Enforced Test-Driven Development (TDD) patterns for subagent tasks

ユースケース

Executing multi-file feature implementations from a structured development plan
Refactoring complex systems where task isolation is required to prevent regression
Rapidly building and testing new modules with automated quality gates