subagent-driven-development

Executing an implementation plan with AI agents involves breaking the plan into independent tasks and dispatching a dedicated subagent to each one. This Skill orchestrates that dispatch process, followed by sequential spec compliance and code quality reviews, managing iterations within a single session.

Subagent-driven development is an iterative AI coding process where complex implementation plans are decomposed into manageable tasks. A fresh AI subagent executes each task independently, ensuring focused code generation and rapid iteration cycles.

AI code review in iterative development works through a rigorous two-stage process after each task is executed. The system runs sequential checks for spec compliance and code quality, automatically managing task iteration and question handling until the code meets all specifications.

Yes, you can use AI subagents to automate TDD and code reviews by assigning a fresh subagent to each task for focused execution. The system then performs automated spec compliance and code quality reviews, iterating until perfect.

You need a structured implementation plan document, such as a markdown file located at 'docs/plans/feature-plan.md', to start AI-driven development. This plan is automatically decomposed into independent tasks for the subagents to execute.

The limitation of using a single AI session for subagent task execution is that all task decomposition, subagent dispatch, two-stage reviews, and iterations must be managed sequentially within that session's context window, which may constrain very large or complex plans.