subagent-driven-development

Subagent-driven development orchestrates autonomous task execution by assigning the optimal agent per task. It applies a selection matrix to dispatch specialized agents for parallel or sequential work flows, ensuring each discrete task is handled by the right agent.

To ensure code quality during agent dispatch, perform a two-stage review after task completion. This process runs a spec compliance review followed by a code quality review, verifying that autonomous implementation meets requirements before finalizing the work.

The agent selection matrix uses a decision flow to determine the optimal agent for each task before starting execution. It evaluates the task requirements against available specialized agents to match debugging, research, or implementation work with the right executor.

Yes, subagent-driven development applies to plans decomposed into discrete tasks, including debugging, research, implementation, and batch changes. It supports both sequential and parallel execution using operational templates and a TodoWrite format to track progress.

No, you do not need to manually write prompts. The workflow includes operational templates with predefined prompts for agents and reviewers, along with a TodoWrite format to organize work and track progress throughout the task execution lifecycle.