subagent-driven-development

You prevent context pollution by dispatching a fresh, isolated subagent for each task in the implementation plan, which ensures session contamination is avoided and in-session execution is accelerated.

The two-stage review process enforces a spec compliance review first, followed by a code quality review, triggering rework cycles until both stages explicitly approve the task before committing.

Yes, you can use TDD with isolated subagents because the implementer subagents exhibit TDD-driven behavior for each independent task, handling implementation, testing, and iterative review automatically.

Iterative retrieval handles long-running tasks by supporting progressive context discovery, allowing the subagent to explore the codebase iteratively rather than loading all context at once.

Yes, subagent-driven development supports explicit blocker handling, allowing the workflow to properly identify and manage blockers during the in-session execution of implementation plans.

You should use subagent-driven development when your implementation plan contains mostly independent tasks that require implementation, testing, and iterative review, benefiting from isolated context and structured review gates.