skills.rest
snits
snitsCOMMUNITY
Explore all skills by this author →
GitHub

subagent-driven-development

Execute implementation plans task-by-task with subagents and automated code reviews.

1|2|Updated Aug 7, 2025
Search Tags:#automation#quality assurance#code review#development workflow#TDD#subagents#plan execution
One-click install
npx skills add https://github.com/snits/claude-files --skill subagent-driven-development-snits
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: subagent-driven-development
Source: https://github.com/snits/claude-files/tree/main/skills/subagent-driven-development
Command: npx skills add https://github.com/snits/claude-files --skill subagent-driven-development-snits

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Executing complex implementation plans can be prone to errors, context drift, and quality issues if not managed carefully. This Skill provides a robust, automated workflow for executing tasks, ensuring each step is implemented correctly and reviewed promptly, saving you from manual oversight and debugging.

Core Features & Use Cases

  • Task-Specific Subagents: Dispatches a fresh subagent for each task, preventing context pollution and ensuring focused, error-free execution.
  • Continuous Code Review: Automatically triggers a code review after each task, catching issues early and maintaining high code quality throughout the development cycle.
  • Iterative Fixes: Manages the cycle of identifying, fixing, and verifying issues until the code meets quality standards, ensuring a polished final product.
  • Use Case: You have a detailed implementation plan. Use this Skill to execute it task-by-task. For each task, a dedicated subagent implements the code and tests, then a code-reviewer subagent automatically reviews the changes. If issues are found, another subagent fixes them, and the review cycle repeats until the code is clean. This ensures high-quality, verified progress, allowing you to rest while AI drives the development.

Quick Start

Example: Executing a task with subagent-driven development

  1. Load Plan: Read plan file, create TodoWrite with all tasks.
  2. Execute Task: Dispatch fresh subagent for Task N.
  3. Review Work: Dispatch code-reviewer subagent.
  4. Apply Feedback: Fix issues, re-verify with code-reviewer.
  5. Mark Complete: Move to next task when issues resolved.
  6. Final Review: After all tasks, dispatch final code-reviewer.
  7. Complete Development: Use finishing-a-development-branch.