Story decomposition

Produce one human-reviewed seed containing a clear parent problem and ordered candidate stories. Do not inspect implementation, perform technical design, or write an executable plan.

Choose the workflow

Use this skill when the beneficiary, problem, desired effect, or value is unclear; a request prescribes a solution without establishing why; several outcomes or story boundaries must be chosen; or the first increment's value or learning priority is disputed.

For selected-story goal, scope, and examples, use dough-story-refinement. Hand off one understood story to the project's execution-planning workflow only when the user explicitly requests planning. A seed or backlog selection does not authorize execution.

Resolve required context

Read the user's instructions and repository guidance to identify:

• The repository root, supplied seed or canonical seed directory, seed ID allocation and filename conventions, and stable story-anchor convention.
• Required seed metadata and lifecycle vocabulary.
• The project's definitions of S, M, and L effort bands.
• The canonical backlog path, only when queueing or reprioritizing is requested.
• The execution-planning workflow, only when that handoff is requested.

Resolve project paths from this project, not this skill's location. If context or a linked dependency needed for the current action is unavailable, name it, ask only for the missing information, and stop the affected activity before writing.

Decompose and record

Read and follow problem decomposition to establish human-owned framing decisions, challenge alternatives, and select, size, and order stories. Then use seed format to update the supplied seed or create a new seed. Leave the seed uncommitted for review unless the user explicitly requests a commit.

Report the seed path, recommended first story, rejected simpler alternative, effort distribution, and open decisions. After writing the seed, end with:

## STORY DECOMPOSITION WRITTEN