subagent-driven-development

Automating implementation plans requires dispatching fresh subagents per task to maintain isolated context, followed by conducting task reviews for spec compliance and code quality. This ensures focused, fast iteration for complex software development.

Subagent-driven development is a technique that automates execution of implementation plans by isolating task context within independent subagents. It solves context bloat by dispatching fresh subagents per task, ensuring high-quality iteration.

Code reviews for spec compliance are conducted after each task execution by a dedicated review subagent using a review package. A broad final review is also performed at the end of all tasks to ensure overall software quality.

Yes, subagent task execution is ideal for complex software projects with numerous independent tasks, such as feature development or bug fixing. It isolates task context to maintain focused and efficient development across the implementation plan.

Yes, automating task execution requires both implementation subagents to execute the plan and review subagents to verify spec compliance. Support for task briefs and review packages is also required to conduct proper code reviews.

The best way to isolate task context is dispatching a fresh subagent for each independent task within the implementation plan. This approach prevents context window bloat and allows a review subagent to efficiently verify spec compliance.