subagent-first

Subagent drift in complex multi-step workflows is prevented by enforcing a mandatory 6-section context structure that standardizes delegation prompts and applies strict constraints, ensuring subagents stay on track and deliver consistent output quality.

Wave-based parallel execution planning is a task decomposition method that breaks complex agent workflows into sequential and parallel waves. This optimizes execution time by managing dependencies across multiple subagents for high-reliability processing.

You standardize subagent delegation prompts by applying a mandatory 6-section context structure to your instructions. This enforces strict constraints and prevents scope creep and hallucination during complex task execution.

Yes, context packaging manages dependencies in parallel agent tasks by structuring instructions into a rigorous 6-section format. This enables wave-based parallel planning to execute complex workflows while maintaining high reliability and preventing subagent drift.

Wisdom accumulation improves future agent performance by capturing and recalling successful execution patterns. This protocol ensures that iterative refinement from previous tasks continuously enhances the reliability of subagent delegation.

You should use subagent delegation for workflow control when handling complex, multi-step agent workflows that require high reliability, dependency management, and iterative refinement to prevent scope creep and ensure consistent output quality.