subagent-driven-development

To delegate implementation plan tasks, you provide a complete plan to the system, which dispatches each independent unit to a fresh subagent. This ensures task isolation, clear accountability, and structured execution with explicit success criteria.

Executing decomposed planning tasks is best handled by delegating each independent unit to a fresh subagent. This approach provides per-task context isolation, structured delegation prompts, and post-task verification to ensure traceable results.

Yes, you can parallelize independent tasks when possible and handle deterministic, sequential execution when needed. The system manages this by dispatching each decomposed task to a fresh subagent with robust failure handling and retry strategies.

Context isolation and traceability are ensured by assigning one fresh subagent per independent task. Each subagent receives context-rich delegation prompts containing the project root, test commands, and conventions for clear accountability.

To start automating task delegation, you need to provide a complete implementation plan. The system then divides the work into independent tasks and delegates each one to a fresh subagent using the delegate tool.

Subagent execution fails without project root and test commands because these elements are required in context-rich delegation prompts. Providing explicit success criteria and review steps ensures robust failure handling and proper post-task verification.