decompose-goal

To break down free-text goals into atomic subtasks for workflow automation, this skill parses your input and infers the appropriate atomic unit—such as an agent-turn, DAG node, or cron job—then outputs an ordered list with clear verification criteria for each subtask.

Yes, you can output decomposed subtasks as JSON for programmatic orchestration. The skill defaults to a Markdown block suitable for human review but can optionally switch to JSON mode for direct integration with downstream automation systems.

Yes, goal decomposition supports dependency annotations for DAG workflows. The generated subtask list includes optional dependency annotations, allowing you to map the ordered atomic units directly into a directed acyclic graph execution model.

The best way to generate verifiable done-states for automated task orchestration is to use a decomposition skill that validates each atomic subtask with a clear verification criterion, ensuring robust auditing and reliable hand-offs across diverse runtime environments.

Yes, you can create human checklists from vague goals using automated subtask generation. The execution-agnostic design supports human checklists as an atomic unit, converting free-text objectives into discrete items with independent completion criteria.

You should not use automated goal decomposition for subtask generation when your objective is already a precise, single-step action that requires no breakdown, or when your execution environment cannot parse Markdown or JSON outputs with dependency annotations.