subagent-development

To prevent context leakage in multi-task plans, dispatch fresh subagents per task to execute in isolation. This approach stops context drift by ensuring each task operates independently without shared historical context.

A two-stage review pipeline for code generation validates spec compliance first, then evaluates code quality. This enforces reliability by catching specification violations before assessing the implementation's structural integrity.

Orchestrate sequential subagent tasks by processing a structured task loop that monitors explicit status handling and rollback guidance. Provide a task plan to start the workflow and track each isolated subagent until completion.

You need fresh subagents per task for automation workflows when managing complex implementation plans across multiple tasks. This isolation technique is required when tasks run independently to prevent context bleed and inconsistent outcomes.

Yes, subagent orchestration works for multi-task planning workflows without dependencies. It sequentially dispatches isolated subagents to execute plans, applying spec compliance and quality reviews to ensure reliable task completion.