lamina-modularity-boundaries

# Modularity Boundaries Divide the product into cohesive parts with clear ownership. Users see simple surfaces; complexity lives behind boundaries that match domain seams. ## Decision frameworks - **Deep boundary**: Small public surface (what actors can do) hiding substantial internal rules (how state changes are validated). - When to use: Complex domains exposed through simple operations ("download ticket" hides payment, venue, timing rules). - How: One user-facing operation per coherent outcome. - **Information hiding**: Actors see only what their role requires (student sees own ticket; invigilator sees roster for assigned venue only). - When to use: Multi-actor products, permission-sensitive data. - How: Map actor × entity visibility in `actors`; never leak cross-boundary data in UI. - **Pull complexity downward**: Callers (users, other features) stay simple; hard rules live inside the owning boundary (exam cell owns venue assignment logic, not the student app). - When to use: Shared entities touched by multiple workflows. - **General-purpose vs specialized boundaries**: Prefer boundaries that serve one domain concept well over "god" features that expose every field. ## Checklists 1. Assign each entity a single owning actor or subsystem in product terms. 2. List what each actor type can see and do — nothing else. 3. Avoid pass-through screens that expose another boundary's internals. 4. When two features share an entity, define which boundary owns writes....