ozerohax/task-use-creator-decomposition-strategy

<when_to_use> <trigger>Need a draft decomposition to accelerate planning</trigger> <trigger>Need a clear task list with scope, dependencies, and sequencing</trigger> <trigger>Need to transform requirements into actionable work items</trigger> <trigger>Need to estimate or plan delivery phases</trigger> </when_to_use>

<delegation_policy> <main_agent>Owns final decomposition, scope decisions, and user alignment</main_agent> <subagent>Produces a draft breakdown, dependencies, risks, and open questions</subagent> <do_delegate>Initial task list, dependency map, risk list, missing inputs</do_delegate> <do_not_delegate>Final scope, priority, acceptance criteria, release commitments</do_not_delegate> </delegation_policy>

<task_request> <principles> <principle>State the goal and expected granularity (epics -> stories -> tasks)</principle> <principle>Provide constraints: timeline, team size, tech stack, non-goals</principle> <principle>Include inputs: requirements, PRD, architecture, use cases</principle> <principle>Limit scope to one epic or feature per request</principle> <principle>Specify depth level when needed: standard, deep, expert</principle> </principles> <examples> <good> <task>Decompose the "User onboarding" epic into tasks with dependencies and acceptance criteria.</task> <why>Clear scope, expected outputs</why> </good> <good> <task>Break down payment integration into tasks, include risk areas and sequencing.</task> <why>Focused feature and planning needs</why> </good> <bad> <task>Plan the whole product</task> <why>Too broad, no granularity or scope</why> </bad> </examples> </task_request>

<decomposition_principles> <principle>Start from user outcomes and required behaviors</principle> <principle>Define deliverables and acceptance criteria per task</principle> <principle>Minimize cross-task coupling; keep tasks independently testable</principle> <principle>Identify dependencies and critical path</principle> <principle>Surface risks and unknowns early</principle> <principle>Separate discovery from implementation work</principle> </decomposition_principles>

<decomposition_strategies> <strategy name="vertical-slices" use_for="End-to-end delivery"> <step order="1">List key user flows and success criteria</step> <step order="2">Split each flow into slices that deliver value</step> <step order="3">Add enabling tasks for platform or shared components</step> </strategy> <strategy name="capability-first" use_for="Platform or infrastructure"> <step order="1">Identify capabilities and system boundaries</step> <step order="2">Define tasks per capability with clear interfaces</step> <step order="3">Add integration and validation tasks</step> </strategy> <strategy name="risk-first" use_for="High uncertainty"> <step order="1">List technical and product risks</step> <step order="2">Create spike tasks to validate assumptions</step> <step order="3">Decompose remaining work based on validated scope</step> </strategy> </decomposition_strategies>

<output_requirements> <requirement>Provide a structured task list with titles (draft)</requirement> <requirement>Include dependencies and sequencing notes</requirement> <requirement>Note risks, unknowns, and required inputs</requirement> <requirement>Flag assumptions that need main-agent validation</requirement> </output_requirements>

<agent_limitations> <cannot>Edit or write files</cannot> <cannot>Make product decisions without user confirmation</cannot> <cannot>Remember previous sessions</cannot> </agent_limitations>

<depth_levels> <level name="standard">Task list + dependencies</level> <level name="deep">Acceptance criteria + risks + sequencing</level> <level name="expert">Phased plan + estimation approach + rollout strategy</level> </depth_levels>