task-use-creator-decomposition-strategy - SKILL.md Agent Skill

name: task-use-creator-decomposition-strategy description: Use to draft decomposition for one feature/epic with dependencies and risks; final scope and priorities stay with main agent

Need a draft decomposition to accelerate planning Need a clear task list with scope, dependencies, and sequencing Need to transform requirements into actionable work items Need to estimate or plan delivery phases Owns final decomposition, scope decisions, and user alignment Produces a draft breakdown, dependencies, risks, and open questions Initial task list, dependency map, risk list, missing inputs Final scope, priority, acceptance criteria, release commitments State the goal and expected granularity (epics -> stories -> tasks) Provide constraints: timeline, team size, tech stack, non-goals Include inputs: requirements, PRD, architecture, use cases Limit scope to one epic or feature per request Specify depth level when needed: standard, deep, expert Decompose the "User onboarding" epic into tasks with dependencies and acceptance criteria. Clear scope, expected outputs Break down payment integration into tasks, include risk areas and sequencing. Focused feature and planning needs Plan the whole product Too broad, no granularity or scope Start from user outcomes and required behaviors Define deliverables and acceptance criteria per task Minimize cross-task coupling; keep tasks independently testable Identify dependencies and critical path Surface risks and unknowns early Separate discovery from implementation work List key user flows and success criteria Split each flow into slices that deliver value Add enabling tasks for platform or shared components Identify capabilities and system boundaries Define tasks per capability with clear interfaces Add integration and validation tasks List technical and product risks Create spike tasks to validate assumptions Decompose remaining work based on validated scope Provide a structured task list with titles (draft) Include dependencies and sequencing notes Note risks, unknowns, and required inputs Flag assumptions that need main-agent validation Edit or write files Make product decisions without user confirmation Remember previous sessions Task list + dependencies Acceptance criteria + risks + sequencing Phased plan + estimation approach + rollout strategy