subagent-driven-development

Delegate implementation plan tasks to fresh subagents to execute each unit with clean context. This workflow orchestration approach prevents state confusion by isolating task execution environments during feature builds.

Enforce spec compliance and code quality using a two-stage review pipeline for implementation tasks. Run a spec compliance review first, followed by a code quality review, looping fixes on failed reviews before proceeding.

Implement test-driven development by delegating tasks that require failing tests first, minimal implementation, and full regression verification. This TDD-aligned execution ensures each independent unit meets quality standards before integration.

Yes, subagent delegation applies to multi-step feature builds where tasks decompose into mostly independent units like models, endpoints, and utilities. It parses the plan into an explicit todo list and dispatches delegate_task with complete context.

Feature implementation may fail integration without a final integration review and test verification step. After completing individual task reviews, perform a final integration review to verify all plan components work together correctly.

Avoid delegate-per-task workflows when implementation tasks cannot be decomposed into small, mostly independent units. This approach requires parseable plans and task isolation to maintain quality and prevent state confusion during execution.