Lamina Ux Agent Skill

Lamina UX

Reference-loading protocol

• Match the request's primary interaction risk to one row below.
• Open that linked reference before answering. Add another only when a second

risk materially changes the answer; do not preload the directory.

• Start the response with Using lamina-ux: <topic path(s)> so the selected

interaction lens is auditable.

Topic index

| Interaction signal | Read | Adds |

|---|---|---|

| Need the end-to-end design-to-implementation loop | Design Process | Persona walks, graph expansion, implementation, and verification lifecycle |

| Need to order steps, branches, prerequisites, and recovery for one outcome | Workflow Design | operation-level flow structure |

| Need to group product concepts, hierarchy, labels, or findability | Information Architecture | entity- and task-based organization |

| Users are lost between destinations or cannot maintain orientation | Navigation | wayfinding, current location, and route structure |

| Need labels, headings, instructions, empty copy, or error copy | Content Design | consistent, scan-first interface language |

| Need input semantics, validation timing, field errors, or submission recovery | Forms | accessible validation and data-entry behavior |

| Need action hierarchy, destructive-action policy, menus, confirmation, or undo | Controls and Actions | explicit action semantics and reversibility |

| A collection or account has no data yet | Empty States | scenario-bound explanation and next action |

| New or returning users cannot reach first value | Onboarding | minimal setup, permissions, and resumable progress |

| Users cannot tell whether an action started, succeeded, failed, or is delayed | Feedback and Status | visible transient and terminal states |

| Need actor-visible failure language and a concrete recovery path | Error Handling | error categories, recovery, and blame-free copy |

| The interface exposes too much complexity at once | Progressive Disclosure | essential versus advanced action tiers |

| Users cannot perceive an available action or its consequence | Discoverability | signifiers, disabled reasons, and feedback gulfs |

| Need keyboard, screen-reader, focus, announcement, contrast, or touch behavior | Accessibility | interaction-specific accessible acceptance criteria |

| Behavior changes through invitations, persuasion, groups, or social influence | Multi-Actor Dynamics | consent, autonomy, and group-effect safeguards |

| A high-stakes action needs honest consequence, fee, privacy, or success signals | Trust Signals | observable product honesty |

| Need distinct denied, stale, concurrent, destructive, or dependency-failure cases | Distinct Product Risks | non-duplicative scenarios with observable acceptance |

Working rule

Use the smallest sufficient reference set. Common pairs are forms + error

handling, navigation + information architecture, and feedback + trust for

high-stakes asynchronous actions. Accessibility applies to every critical

interaction but load its full reference when accessibility behavior is a

decision, acceptance target, or stated risk.