lamina-product-behavior

# Product Behavior (agent-native) The **represented model** in the transactional graph must match how actors conceive tasks — simpler than implementation, aligned with domain invariants. ## Contract encoding | Layer | Artifact | |-------|----------| | Domain truth | `entities[]`, `invariants[]` | | What actors can do | `actors.permissions`, `workflows` | | What UI shows | `surfaces[]` — no affordance for forbidden operations | | Illegal states | `scenarios[]` + disabled/hidden actions in screen spec | **Implementation model** stays in external code. Lamina specifies **represented model** only. ## Frameworks - **Goals vs tasks**: Design for stable goals; eliminate tasks technology made obsolete. - **Capability / viability / desirability**: Record trade-offs in `decisions.md` when pillars conflict. - **Design values**: e.g. don't make actors feel stupid — use as filter when reconciling persona panel conflicts. - **Patterns**: Modeless feedback, reversible actions — reference in `implement.md`, don't prescribe UI library. ## Design checklists 1. No screen shows actions the actor cannot perform (or shows why disabled). 2. Entity names in UI match `entities[]` vocabulary. 3. States visible to actors match the entity lifecycles — no mystery modes. 4. Primary actor path optimized; edge cases in scenarios, not driving IA. 5. Deviations from common patterns documented in `decisions.md`. ## Verify checks - Actor walks: forbidden operations blocked with scenario-matching `ux`....