subagent-driven-development

To execute an implementation plan with parallel subagents, this skill dispatches a fresh implementer per task and enforces sequential spec compliance and code-quality reviews, looping fixes until both approve before proceeding to a final code review.

Subagent-driven development is a plan execution method that breaks an implementation plan into independent tasks, running each through a tight implementer, spec review, and quality review loop to eliminate slow, error-prone cycles and avoid cross-task context confusion.

You enforce spec compliance during code generation by applying a two-stage review checkpoint that performs a dedicated spec compliance review first to catch missing requirements, followed by a code quality review to prevent unwanted extras.

Yes, you can use this approach for iterative test-driven development improvements because it is explicitly designed for contexts where iterative improvements are expected, looping reviewer feedback into fixes until both spec and quality reviews grant approval.

You should not use subagents for implementation plans when your tasks are heavily interdependent, as this method applies specifically to executing plans where tasks are largely independent and can be dispatched to fresh implementers without cross-task context confusion.

The best way to automate code review in a git workflow here is to run a two-stage review per task and conclude with a final code review for the entire implementation, ensuring both spec compliance and code quality are validated before completion.