What it does

Tarot Watch turns a daily draw into a sequence the Digital Crown controls. The ritual moves through drawing, card identity, confirmation, and five reading sections: Overview, Love, Work, Inner World, and Spiritual. DO and AVOID beats lead to a final rest on that day’s card.

The native app uses SwiftUI Canvas and procedural motion. Crown input moves between reading stations rather than scrolling one long page. The station gate accounts for a Crown that continues sending events during a turn, so a sustained turn can advance through settled stations without treating every tiny event as a new gesture.

Daily state stores the draw and reading position locally. Returning on the same day resumes the reading instead of requiring a fresh draw. The deck generation and once-per-day state are separate from authored reading coverage.

The documented content is keyed by card, orientation, and section. The Moon and The Sun each have upright and reversed wording, scene states, and closing readings. A missing content field is reported as missing rather than silently borrowed from another orientation or card. Cards outside the authored set are explicitly shown as The Moon and logged; that coverage rule is not a completed reading for the drawn card.

Each supported card has its own opening and aftermath before entering the shared reading. The shared folio keeps today’s card and a small narrator point; section scenes use the state named by the content. The Sun’s native Noon opening changes the ground and shadow, while The Moon keeps its own eclipse language. The older web lab is a prototype and does not represent the current native reading path.

What I worked on

I worked on the interaction design, native Watch implementation, and reading system, using AI-assisted development. The work includes the Crown station grammar, daily persistence, card/orientation content model, procedural motion, reduced-motion reading states, and the handover from each card’s opening into a shared sequence. Reading copy is an editorial rewrite of the documented Tarotoo source fields, not a claim that I authored a full deck from scratch.

What has been checked

The 4 October 2026 handoff records all four supported readings end to end on a physical Series 12 Watch in a Release build. The user held the Watch awake; a developer -walk mode advanced stations through the app’s own director. This is hardware execution with scripted progression, not an unassisted usability study of daily Crown use.

The same handoff distinguishes simulator checks: scripted Crown movement through every station, same-day relaunch/resume at six positions per reading, and reduced-motion end states. Automated web and Swift tests cover content, stations, cards, and motion models. Simulator resume and frame checks are not physical-device proof of every path.

The physical runs exposed an approximately 150 ms reading-layout cost at the handover. Moving that work to the still card beat produced clean Sun handovers in subsequent hardware runs; the corresponding Moon timing fix was checked in the simulator.

What remains unproven

The authored/tested scope is four readings across two cards. There is no supported claim of a complete authored 78-card experience, broad device performance, or independent user validation. The draw ritual still lacks full reduced-motion handling, and section-scene replay is tracked per process rather than once per day.

The native Watch README, current handoff, and 4 October reading-system specification establish this scope. The reading-system update describes the content-as-data milestone. Licensed deck artwork is not reproduced on this record.