Lamina product behavior

Product Behavior (agent-native)

The represented model in run.json 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.
• Invariant probes: UI never implies illegal state (e.g. two active tickets).
• Walkthrough: represented labels match contract entity names.

Anti-patterns

• Elastic user: Generic actor justifying any design choice.
• UI after coding: Contract written to match accidental implementation.
• Implementation-shaped menus: File/Edit mirroring backend modules.
• Feature lists without goals: Workflows not traceable to actor outcomes.

Related

• Invariants
• User Modeling
• Platform Posture