subagent-driven-development

You automate code generation by orchestrating AI subagents to execute implementation plans, dispatching dedicated agents for writing code and conducting a two-stage review process for spec compliance and code quality.

The best way to break down a feature plan is using subagent-driven development, which decomposes complex implementation plans into manageable, independent tasks executed by specialized AI subagents within a single session.

The two-stage code review process works by first dispatching a subagent to verify spec compliance, then dispatching another subagent to assess code quality, managing iterative feedback loops until all issues are fixed.

Yes, you can use subagent orchestration for complex software development workflows by breaking down detailed feature plans into independent tasks and letting dedicated subagents implement and review the code.

Yes, you need a detailed implementation plan. The subagent-driven development workflow requires a pre-existing plan, such as a feature plan document, to decompose into independent tasks for the subagents to execute.

Iterative refinement loops are used in AI code generation to manage feedback between implementer and reviewer subagents, ensuring that spec compliance and code quality issues are automatically fixed before final approval.