subagent-driven-development

Executing implementation plans in isolation requires dispatching a fresh subagent per task, preventing context leakage. This approach routes results through two-stage spec compliance and code-quality reviews before final integration on the development branch.

Subagent-driven development is a workflow that delegates independent tasks to fresh subagents within a single session. It preserves main session context for coordination while using TodoWrite tracking and two-stage reviews to maintain structured progress.

TodoWrite tracking for per-task subagent execution is set up by extracting the implementation plan and creating todos for each independent task. The workflow dispatches a fresh subagent for each todo item, followed by spec and code-quality reviews.

Yes, you can use subagents for complex plan execution across multiple independent tasks. The workflow is designed for mostly independent tasks within a single session, performing a final integration step to finalize the development branch after all reviews.

Subagent execution works with spec compliance and code-quality reviews by enforcing a two-stage review process between tasks. After a fresh subagent completes a task, the results are routed through both review stages before proceeding to the next handoff.

You should avoid using isolated subagents for development tasks that require heavy interdependent state sharing within the same session. The workflow is suitable for mostly independent tasks and may not fit tightly coupled execution requiring continuous main session context.