| name | organize-work |
| description | Partition stories/requirements into MECE clusters ordered by dependency. Use when re-slicing many items on wrong seams or turning planning smear into a build partition. Not closure-check (border gate). Not architecture-design. Not write-story. |
Organize Work
Input: a working set of stories or requirement text.
Output: clusters (each one reason-to-change) + dependency order between clusters.
Do this
- Atomize. Shred the input into irreducible atoms (one requirement, decision, or deliverable each). Conserve content — nothing invented, nothing dropped.
- Type by reason-to-change. Group atoms that change together for the same reason and stakeholder. Never group by execution step or layer.
- Edge the atoms. For each atom, list what other atoms it requires to exist first. Edges may point outside the working set — note them, do not ignore them.
- Cluster. Form clusters of high cohesion / low coupling. Every atom in exactly one cluster. Orphan or double-home → re-cluster.
- Order. Topologically sort clusters from the dependency edges. Cycles mean a missing atom — go back to step 1.
- Emit. List: cluster name, atoms in it, depends-on clusters. Stop.
Do not
- Invent requirements to make clusters prettier.
- Order by “what feels first.”
- Run a closure-check here — hand outside edges to
closure-check if the border matters.
