Conversational Ask, library actions, automatic metered Encores, onboarding, and Settings remain active product work—not a new numbered-goal migration.
Current native checkpoint · August 13
Living product baselineShowNext. Now.
A visual review of the real composed app—what changed, what we chose, and what still needs proof.
iPhone 17 Pro Simulator · retained test-account data · current signed development build · native app remains authoritative
Cape Fear, Food Truck, Married with Children, person detail, and genuine metadata are protected regression references.
Physical voice, gestures, accessibility, notification behavior, onboarding, large-archive Encore cost, and eventual TestFlight remain separate gates.
Current build reel
See the app before reading the report.
Select any current Simulator capture for an uncropped view. Swipe sideways on smaller screens.
Current
Today
Current
Shows
Current
Movies
Current
Discovery
Current
Settings
Current
Cinematic detail
Current
Retained show detail
Current
Episode progress
Current
Person detail
Current
Season Encore
Protected rendered decision
Direction closedCape Fear sets the cinematic reference.
Genuine art reaches behind the floating controls, preserves its subject, transitions into a readable artwork-derived field, and rounds only the lower edge. Related show, movie, episode, person, and Encore changes must be compared against this accepted family.
Read the protected baseline
Current visual identity
The native icon is now selected.
The icon direction has moved into the current app build. Earlier community studies remain preserved as design history rather than an active product gate.
Development archiveEarlier research, choices, and completed work
The Lab now shows the app we are actually testing.
The current-build gallery was recaptured from the retained isolated-test-account iPhone 17 Pro Simulator on August 13. Today, Shows, Movies, Discovery, consumer Settings, Cape Fear, Great Food Truck Race Progress, Nicholas Galitzine, and Married with Children Encore now represent the current native source. The audit caught and corrected one shared routed-navigation no-op before publication. Historical screenshots remain preserved as dated evidence, not current-build imagery. Simulator evidence remains separate from physical iPhone, Trakt settlement, voice, TestFlight, and App Store proof.
completeToday stays anchored while the posters take the lead.
The current native visual baseline pins its 15-day date rail and layers timed episode cards only on busy days with at least three broadcasts. One- and two-item days stay simple, and Reduce Motion or accessibility text uses the normal list. Shows and Movies move Library Care and Backlog Curator into one compact toolbar menu instead of leading with maintenance cards. Discovery moves Viewing Profile into a compact toolbar action and Settings, combines its secondary controls under Refine, and varies poster scale so the app reads as a cinematic composition rather than a repeating grid. Package tests, focused UI journeys, and current iPhone Simulator review pass. Candidate admission is reopened; native haptic strength and the full physical-iPhone visual walkthrough remain separate proof.
Historical updateThe purchase laboratory passed, then the Release stayed free.
Goal 8E proved an optional non-consumable behind a typed, provider-neutral entitlement boundary, including purchase, restore/relaunch, pending, cancellation, refund/revocation, offline last-verified evidence, and Family Sharing fixtures. Thomas later chose a free Release: every retained capability ultimately admitted will require no receipt, and Release compiles StoreKit commerce and paid-product UI out. Goal 8J-A freezes the no-charge promise inventory and marks the advanced StoreKit families deferred. The laboratory remains R&D—not an App Store Connect product, final price, sandbox transaction, Family Sharing service proof, phone purchase, or TestFlight upload.
completeThe AI experiences now pass as one system.
At the July 26 Goal 8C checkpoint, Personal Upcoming, Watch Next, separate show and movie Backlog Curators, Viewing Profile, and Explainable Discovery passed one fail-closed admission contract. Each exposes inspectable evidence, stays useful without ChatGPT, offers the same private feedback choices, remains account scoped, and keeps any enhanced provider away from title selection, source truth, Trakt, and user mutations. The joined review repaired missing feedback and a duplicate-title crash; the sealed Release fixture boundary and Simulator accessibility journey passed separately. Its historical receipt reported 4 of 9 admitted with five blockers. The July 30 release reset supersedes that denominator without erasing the six experience receipts.
completeDiscovery now explains every recommendation.
ShowNext now keeps its four agreed Discovery shelves stable while deterministic local rules rank each typed TMDb candidate. Every personalized reason opens its exact evidence; without a Viewing Profile, the app uses honest release and catalog facts instead of fake personalization. Describe turns supported words into visible removable filters. More, Less, Hide, and Undo act immediately, remain private to ShowNext, and never change Trakt. The same local-first private-iCloud overlay carries those preferences across Apple devices, while an optional enhanced provider may rewrite only validated reason text and cannot add or reorder titles. Focused ranking, metadata, cloud-convergence, UI, Accessibility Large, and Reduce Motion checks pass in the iOS 27 Simulator. The Lab demo remains fictional and resets on refresh. No ChatGPT connection, provider transfer, Trakt write, physical-phone install, production CloudKit change, StoreKit product, or TestFlight upload occurred.
completeYour Viewing Profile now follows you—without slowing the tracker.
ShowNext now saves the bounded derived Viewing Profile and its private label choices locally first, then reconciles them through the user's private iCloud database in the background. A real physical iPhone published a fictional profile, the Mac pulled and renamed one label, the iPhone independently dismissed another, both choices merged, and the final deletion converged back to the phone with Pending 0. A 20-second remote limit prevents endless waiting, account changes pause until explicitly resumed, and a confirmed tombstone cannot be recreated by an automatic check. The browser demo remains fictional and never contacts CloudKit. No Trakt write, ChatGPT connection, provider transfer, production CloudKit change, StoreKit product, or TestFlight upload occurred.
completeViewing Profile now explains the patterns it can prove.
Viewing Profile now keeps complete local counts separate from a balanced recent metadata cohort of at most eight shows and eight movies. It can show honestly worded covered-genre, cast-connection, and runtime patterns; replay patterns require an explicit complete lifetime-history receipt. Stable built-in labels are concise, editable, dismissible, deduplicated, and saved to the current local account. A typed derived artifact also converges deterministically between two fictional private-iCloud endpoints without syncing prompts, full responses, credentials, or viewing history. Production app upload/download of that artifact remains deliberately unwired pending explicit authorization. No ChatGPT connection, provider transfer, Trakt write, physical-phone install, production iCloud write, StoreKit product, or TestFlight upload occurred.
Historical updateYour Viewing Profile now says what it knows—and what it does not.
Discovery now opens a private Viewing Profile with instant Last 12 Months or All Time and All, Shows, or Movies controls. The first version shows only current library, verified Continue Watching, private organization, and locally available recent-history counts. A visible coverage receipt refuses to invent favorite genres, cast, runtime, rewatches, completion, or lifetime totals until ShowNext has broader verified metadata and history. The controls are saved in an opaque account-scoped location and cannot transfer to a replacement account. Deterministic tests, the rendered iOS 27 simulator journey, and visual review pass. Richer labels remain a separate Goal 8C4B coverage task. No ChatGPT connection, provider transfer, Trakt write, physical-phone install, production iCloud write, StoreKit product, or TestFlight upload occurred.
Historical updateYour backlog can now get shorter without becoming another endless list.
Shows and Movies now offer separate private reviews of at most 10 evidence-backed titles. A session never appends while it is open. Skip changes no library or Trakt state and waits 30 days unless the evidence changes; closing changes nothing. Keep, Snooze, and Remove all open the established exact-action preview, and removal still uses ShowNext's durable pending, retry, and Trakt verification path. Focused projector tests, the complete modular suite, the final iOS 27 simulator journey, and visual review pass. The visual pass caught and blocked a broken vertical action layout before admission. No ChatGPT connection, provider data transfer, Trakt write, physical-phone install, production iCloud write, StoreKit product, or TestFlight upload occurred.
Historical updateOne honest next choice now appears when the evidence is fresh.
Shows and Movies can now surface one built-in Watch Next choice from recent verified Continue Watching state. The finishable-first projector rejects stale, dormant, empty, future-dated, or completed input; makes no story or service-availability claim; opens the normal media detail; and exposes next point, progress, activity, source freshness, and three private feedback choices. Focused module tests, the complete ShowNextKit suite, the rendered iOS 27 simulator journey, the explicit Release simulator build, and a signed generic iPhone build pass. Visual review confirmed one-line titles and caught two accessibility/fixture defects before admission. ChatGPT remains disconnected, and no provider data, Trakt write, physical-phone install, production iCloud write, StoreKit product, or TestFlight upload occurred.
Historical updateThe first useful brief now works without ChatGPT.
Discovery now shows Since your last visit only when verified release facts contain a supported meaningful change. The expanded What changed page links every statement to its source, records source freshness, opens the normal media detail, offers Useful, Not for me, and Incorrect or spoiler, and keeps at most eight briefs for 30 days. The deterministic path is account-scoped and a routine source recheck no longer manufactures another brief. Focused module tests and the rendered iOS 27 simulator journey pass after visual review caught and repaired a clipped title and false History growth. ChatGPT remains disconnected; no provider data, Trakt write, physical-phone install, production iCloud write, StoreKit product, or TestFlight upload occurred. Goal 8C remains open for the rest of the selected experiences and joined proof.
Historical updateShowNext intelligence now works without requiring ChatGPT.
ShowNext now owns a provider-independent AI foundation with deterministic built-in processing, a durable low-priority queue, source-linked provenance, explicit consent, spoiler-safe feedback, and private deletion controls. Settings says exactly what is true: built-in intelligence needs no account, ChatGPT is not connected, and nothing is shared with it. The complete 261-test modular suite, explicit iOS 27 builds, focused Settings UI proof, and deterministic project generation pass. No ChatGPT authorization, provider data transfer, physical-phone install, StoreKit product, TestFlight upload, Trakt change, or production iCloud write occurred. The selected user-facing intelligence experiences begin in Goal 8C; production ChatGPT authorization remains a later release-trust gate.
Historical updateThe testing build now says exactly what is ready—and what is not.
At the July 25 Goal 8A checkpoint, ShowNext produced one typed nine-area release receipt that reported 2 of 9 areas admitted and seven blockers. Debug kept its isolated proof routes, while a Release simulator build given hostile flags opened the normal Trakt gate. The modular tests, project generation, Simulator screens, and signed generic Release build passed. The July 30 release reset supersedes that fixed denominator because later promises were not classified; the proof-route evidence remains valid.
Historical updateTestFlight will test the intended product—not a reduced shell.
This July 25 decision selected the trusted core, useful built-in intelligence, AI experiences, Season Rewards, and a no-charge Release. The July 30 audit superseded its old 8A-through-8I sequence: Goal 8I remains the active Trakt laboratory with learning and cleanup/repeat open, Goal 8J now owns Apple-first product completion, Goal 8K owns integrated iPhone acceptance, and Goal 8L owns an exact TestFlight-installed candidate. TestFlight remains evidence, not a shortcut around finishing the experience.
Historical updateThe iPhone kept a choice safe while iCloud was out of reach.
A dedicated one-screen proof replaced the developer-facing Device A and Device B chooser for the final gate. The complete Simulator fixture first retained Saved offline with Pending 1, then recovered the same fictional choice to Pending 0. On the real iPhone, Airplane Mode made iCloud genuinely unreachable; ShowNext visibly reported Saved on this iPhone and offered synchronization only after iCloud returned. Restoring connectivity produced Recovery verified and Pending 0 without repeating the action. Apple's developer file-copy extension timed out after the wireless interruption, so the recovery result remains direct real-screen evidence. After connectivity stabilized, a separate phone-container receipt verified the fictional cleanup as Deleted with Pending 0, and normal ShowNext relaunched. These checks remain separate from the passing 243-test suite and signed builds. Goal 7F now passes.
Historical updateThe iPhone and Mac now agree after conflicts, deletion, and a stale return.
A real iPhone 17 Pro Max and the bounded native Mac companion synchronized one fictional ShowNext preference through private development CloudKit. Independent revision-4 edits converged deterministically with zero pending work; a phone deletion reached the Mac without resurrection; a Mac left at revision 5 returned after 100 phone revisions and moved directly to revision 105; and the replacement namespace stayed isolated from the primary account. The explicit CloudKit-cursor resume retained local state, and both namespaces finished as synchronized tombstones with pending zero. The current 243-test suite, signed phone/Mac builds, and deterministic project generation pass separately. This is real cross-device evidence, but Goal 7F remains narrowly open because the run did not literally disable system network or iCloud access.
Historical updateThe Mac is now a real second iCloud test endpoint.
A deliberately basic native Mac companion now shares ShowNext's controlled Device A and Device B iCloud proof instead of pretending that two roles on one phone prove cross-device delivery. It imports only the shared sync, CloudKit, domain, and design modules; it has no Trakt, viewing library, posters, or community identity. Opening it makes no CloudKit request. The signed Release app, safe-default UI test, visual inspection, and complete 221-test modular suite pass. A separately unlocked live test also created, read back, deleted, verified the tombstone, and cleaned up one fictional record in development CloudKit. This proves the Mac endpoint, not iPhone-to-Mac convergence. The full Mac product design remains deferred until the iPhone experience is settled.
Historical updateThe two-device iCloud exit now has an honest, controlled walkthrough.
A debug-only harness now separates fictional Device A and Device B local stores, primary and replacement Trakt namespaces, local staging, and explicit development-iCloud synchronization. Opening it makes no CloudKit request. The focused iOS 27 Simulator workflow proves both roles can stage independently while keeping one pending local record each; the complete 221-test modular suite, deterministic project generation, visual inspection, and Thomas Fraley-signed generic iPhone build pass separately. This does not claim two-device convergence. Conflict, deletion, long-offline return, iCloud-off recovery, replacement-account isolation, zero-pending receipts, and cleanup still require two physical devices.
Historical updateComing Soon reminders now stay useful without becoming another sync engine.
ShowNext now derives local reminders only from confirmed release data and private notification choices. Important covers new seasons, theatrical or streaming releases, and changed dates; All can include ordinary episodes. Snoozed, On Hold, Dropped, canceled, uncertain, stale, and more-than-seven-days-away events are suppressed, and the queue is capped at 32. Five focused tests and the complete 221-test modular suite pass. The real iOS 27 notification center proved three fictional owned requests become one and then zero while an unrelated request remains. Separate Allow and Don't Allow product paths stay useful. No physical-phone delivery, remote push, background guarantee, provider write, or CloudKit write is claimed. Goal 7F is next.
Historical updateShowNext now reaches Siri, Shortcuts, and Spotlight without creating another sync engine.
Five read-only App Shortcuts compile into the signed iPhone artifact: open Today, Shows, Movies, Discovery, or ask what's next. Spotlight keeps four generic destinations and may show at most six Today titles from the existing private snapshot. A real iOS 27 controlled round trip proved six fictional private results become zero after sign-out while the four generic destinations remain. Four focused contracts, the full 215-test modular suite, deterministic project generation, and the Thomas Fraley-signed artifact pass separately. No physical-phone install, Siri invocation, provider write, CloudKit write, notification, or new background sync engine occurred. Goal 7E is next.
Historical updateToday now reaches the Home Screen without becoming another sync engine.
ShowNext now writes one private, versioned App Group snapshot containing at most six render-ready Today items. The read-only widget has no Trakt, TMDb, CloudKit, credential, mutation, or network authority. Seven focused projection tests, the full 212-test modular suite, the last full product UI run's 43 passes, signed host/extension entitlement parity, and one separate real iOS 27 Home Screen add/render/remove proof pass. Visual review caught secondary text that was too dark and the corrected widget now keeps its schedule and status readable. No physical-phone install, Trakt write, CloudKit write, App Intent, Spotlight item, or production notification occurred. Goal 7D is next.
Historical updatePrivate iCloud sync passed its first live round trip.
ShowNext's private-development CloudKit adapter now passes on a signed iPhone. The first attempt failed safely with a missing-zone error and exposed an empty-scope zone-send bug. After that repair, PowerMobile wrote one fictional preference, read back the CloudKit-confirmed record, replaced it with a tombstone, and showed Private round trip verified. Fifteen foundation tests, nine adapter tests, and the full 205-test modular suite pass. This was one-device development proof only: no Trakt write, viewing title, production CloudKit change, widget, or notification was involved. Goal 7C's bounded widget and App Group projection is next.
Historical updateShowNext now knows what iCloud is allowed to synchronize.
The new CloudOverlaySync module accepts only ShowNext-owned organization choices, preferences, discovery feedback, and private notes. It saves locally first, keeps offline work pending across relaunch, resolves concurrent device edits deterministically, converges deletion through tombstones, and rejects records from another Trakt account. Ten focused tests include a complete 5,000-overlay exchange, and the full modular suite passes 191 tests. The real app also builds and launches cleanly in the iOS 27 iPhone Simulator. No CloudKit entitlement, container, database write, widget, notification, Trakt action, or physical-phone change was added; the private CloudKit development adapter is Goal 7B.
Historical updateThe reward mechanics work; the visual payoff needs another pass.
The native Season Rewards collection, import story, and Crest Forged reveal are built with fictional proof data. The production handoff reads an atomically completed, paginated Trakt watched-season snapshot and joins it to complete catalog metadata; partial imports stay Reconciling and cannot create a crest. The complete modular suite contains 181 tests, including a 2,000-show and 10,000-season persistence proof. On PowerMobile, Continue, skip, replay, scrolling, and reduced motion worked. The review also found that repeated generic check-mark crests do not explain their show or season, and the reveal is too understated to deserve the word ceremony. Mechanics pass; final visual admission does not. No new Trakt write was added.
Historical updateGoal 6F’s connected product path is functionally closed.
The real ShowNext app connected to the isolated test account on PowerMobile without adding a diagnostic or test-runner app. Calendar identity, past-day fetching, same-account first-import refresh, Known For artwork, visible movie-poster hydration, and the always-visible movie-detail title are repaired. Thomas confirmed the real Thursday show, detail, trailer, cast, biography, relaunch persistence, and filmography posters on PowerMobile. The later Matrix-only visual delta passed twice in the connected Simulator, then exact Trakt read-back proved the temporary movie removed. Prior human VoiceOver evidence plus a joined-surface accessibility audit closes the functional gate without asking for a redundant walkthrough. The person-page design remains open: the three-column cards feel too square and the title/type overlay needs more left inset.
Historical updateThe beta now stays useful while fresh details arrive.
The joined iPhone path keeps 7,500 confirmed titles browsable while freshness is checking, opens imported details immediately, hydrates artwork, synopsis, trailer, cast, person biography, and filmography through exact TMDb identity, and gives Discovery four real shelves plus local-library-first search. The current modular suite passes 164 tests. Focused Simulator journeys cover yesterday retention, stable details, missing artwork, Accessibility XXXL, Reduce Motion, offline local truth, joined navigation, and the iOS 27 glass tab bar. Connected signed-app, human VoiceOver, and provider cleanup evidence are recorded separately in the Goal 6F closeout.
Historical updateLibrary cleanup now starts with a truthful preview.
Library Care now opens from Shows, Movies, or Settings with All/Shows/Movies scope, calm review reasons, multi-select, Kept Items, and one exact action preview. Keep and Snooze remain private; confirmed watchlist removal uses the existing durable verification path and the title stays visible until that path settles. Settings now explains the one-account boundary, protects pending work at disconnect, requires two stages for ShowNext-data deletion, previews a redacted support report before sharing, and shows Trakt, TMDB, and JustWatch attribution. All 149 package tests and the focused iPhone 17 Pro simulator trust workflow pass with fictional data. No account was contacted and no deletion, disconnect, CloudKit change, or remote write occurred.
completeThe imported library now helps decide what is next.
Shows and Movies now open from the confirmed local projection, then enrich from a separate 15-minute metadata cache. Coming Soon covers the next 31 days; Continue Watching, release changes, recently released titles, and a complete All list remain independently useful. Snooze, On Hold, Dropped, and per-title notification preferences save immediately as private ShowNext choices and never mutate Trakt. Provider buttons perform a title search without claiming availability. The modular suite passes 138 tests, the iPhone simulator build passes, and one focused workflow restores an On Hold title without changing Trakt. The live metadata refresh was not claimed because the simulator product account was disconnected; notification delivery is also not active yet.
completeWatchlist actions passed their live cleanup gate.
After explicit approval, an isolated test account added one allowlisted show and one allowlisted movie, verified both through complete Trakt read-back, removed those exact records, and verified both absent. A separate write-locked GET-only relaunch independently confirmed the restored starting state. The proven remote is now composed into the signed-in iPhone product: each show or movie saves locally first, remains independent of other titles, and changes settled bookmark state only after Trakt confirmation. The modular suite passes 132 tests, and a focused iPhone 17 Pro simulator workflow changed two titles back-to-back without blocking the list. No personal viewing data is published, and collection, ratings, and list membership remain outside this gate.
completeThe real product proved its resumable local-library path.
ShowNext's isolated empty-account OAuth and read-only import completed in the real product target, including the separate history pass and a one-request no-change relaunch that kept confirmed local state available. A debug-only fictional transport then force-quit the product after page one of a 740-show import; relaunch preserved the run, fetched only the remaining seven pages, committed exactly 740 shows, and left no staged records. Sync Center now visibly names the connected account and has an explicit Close action. The real-service account was intentionally empty, so large-library evidence remains faithful controlled transport rather than personal viewing data. No Trakt write, CloudKit change, purchase, AI connection, or physical-iPhone install was added.
completeThe beta now has one honest launch state.
ProductSessionFeature now owns launch routing, briefing state, Sync Center health, last-confirmed time, pending-action count, and fallback permission. Recent local truth opens Today immediately; stale, slow, offline, and failed checks stay labeled without discarding usable local state. One hundred package tests and three focused iPhone simulator launch workflows pass. This is a controlled-data composition checkpoint: the product account and read-only import connect in 6B, and no new Trakt write was added.
completeGamification now has a build gate.
SeasonRewardsFeature is now the non-blocking Goal 6R proof slice after the coherent private-beta core. It derives Season Marks, Crest ceremonies, the first-import mosaic, and the Living Horizon from verified completion truth without owning Trakt writes or creating a second counter. It can enter Goal 8 launch scope only after import, reversal, duplicate, scale, Reduce Motion, VoiceOver, and physical-iPhone evidence passes.
decidedThe real 1 / 5 / 20 Trakt proof passed.
ShowNext's isolated test account accepted 1, then 5, then 20 watched episodes in one batch per pass. Every item was read back with an exact Trakt history ID, removed only by that ID, and each pass finished with zero remaining records; an independent full-account import also ended at watched shows 0 of 0 and history 0 of 0. Local admission took 2 ms, 2 ms, and 7 ms. The live run taught us that Trakt rounds history timestamps to the minute and can cache an identical filtered read after removal, so the engine now matches provider precision and advances the bounded verification window without repeating a destructive request. The completion audit now measures the add and exact-removal request counts rather than assuming them; all 126 current package checks and the six focused iPhone 17 Pro simulator workflows pass. This evidence applies only to the isolated watched-history contract, not personal accounts or other write types.
completeThe living demo now follows the code.
A native-source audit found that the first Lab demo omitted Launch Briefing, Sync Center, and the Coming Soon sections, while presenting planned or invented Library Care, Smart Views, watchlist actions, and an automatic Synced result as current behavior. Alignment 02 replaces that version with the nine current native surfaces, exact Shows, Movies, Today, and Discovery ordering, source-controlled fictional people and artwork, and the real fake-action boundary: Confirming remains pending. A machine-readable contract and Swift-to-Lab alignment test now fail the site check when those markers drift.
completeThe iPhone spine is ready to try.
Milestone 4 passed separately on PowerMobile with 63 package checks, 11 focused workflows and the manual device walkthrough complete. This first browser-demo release used fictional in-memory state and made no Trakt, account, CloudKit, or community-identity call, but it was superseded the same day after a native-source audit found missing and invented surfaces. Native alignment 02 is the current walkthrough.
completeThe local library now has a complete path.
Shows, Movies, Discovery, Search, media details, trailers, cast, and fictional person profiles now run as separate fixture-first modules. Repeated swipe actions change to Confirming immediately while other rows remain actionable, but the adapter performs no Trakt write. Sixty-one package checks pass, and all 17 iPhone workflows have passing evidence across stable simulator shards, including offline use, preserved library scroll, standard visual captures, and a controlled Accessibility Large run. Direct same-viewport comparisons passed for five native screens. The iOS 27 beta can crash SpringBoard during one long automated clipping audit, so that limitation remains visible instead of being called a monolithic suite pass. Milestone 4 still needs its combined physical-iPhone walkthrough, and Goal 5 real writes remain gated.
completeGlass Horizon now uses real iOS 27 glass.
The first coded Today screen matched the Glass Horizon layout but its custom navigation was too flat. The corrected modular shell now uses SwiftUI's native Tab API and floating Liquid Glass tab bar, with selective glass on sync/account, Back to Today, the selected day, and system sheets. Poster shelves and timed rows stay solid for contrast. Fifty-five package checks and all nine iPhone UI workflows pass, including accessibility, measured scroll restoration, and repeatable top, sheet, and scrolled captures. This strengthens Glass Horizon as a front-runner without making it the final visual decision.
completeToday now works across fifteen days.
Thomas authorized Goal 4B after Goal 4A. The modular TodayFeature now covers seven days behind and ahead, followed timed shows, Available Today, brand-new series premieres, watchlist-first Movies Today, honest current/offline freshness, historical watched styling, local artwork, persistent day and vertical scroll state, and read-only media navigation. Fifty-five package checks and all nine iPhone UI workflows—including the text-clipping accessibility audit, measured scroll restoration, and native-glass capture—pass. The final iPhone 17 Pro screen also passed a direct Glass Horizon comparison. No Trakt write or Goal 4C feature was added.
completeThe first product foundation is built.
Thomas recorded Keep for the measured local-first architecture and authorized Goal 4A. ShowNext now has a separate iPhone product target, four navigable destinations, a useful Launch Briefing, truthful recent/stale/offline routing, Sync Center, shared design primitives, and fixture-only states. Forty-eight package tests, a generic iPhone build, and four rendered UI workflows pass. A controlled iPhone 17 Pro simulator inspection verified the repaired cards at standard size and accessibility XXXL; an automated Dynamic Type, clipping, traits, and description audit also passes. Goal 4A is complete. No Trakt write was added or authorized.
completeInstant now has a stricter meaning.
The July 20 audit closes two gaps before product UI begins. Cold or stale launches open into a useful ShowNext Briefing while Trakt currentness is checked, rather than silently showing an old Today. Watched episodes enter a truthful Confirming state, but every other row remains immediately swipeable; compatible rapid actions can batch behind the interface and leave Today only after per-item read-back. Goal 4 is split into launch/navigation, Today, and library/detail proof slices. Thomas has now recorded Keep and authorized Goal 4A; later goals and every Trakt write remain separately gated.
decidedThe real iPhone import passed.
Version 0.5 completed the opt-in, read-only real-account workflow on the baseline iPhone in under two minutes with zero retries, about 79 MB peak memory, and a roughly 30 MB local projection. The large history phase uses 74 requests instead of the earlier roughly 735-request plan, while Trakt's pagination-total discrepancy is recorded honestly rather than hidden or retried forever. Forty-two package tests and four isolated simulator UI workflows also pass, and a one-request no-change refresh avoided downloading the library again. Thomas reviewed this evidence and recorded Keep. No personal viewing counts, titles, comments, credentials, or provider payloads are published.
decidedLarge libraries now reconcile.
Milestone 2 passed: 1,000 fictional shows, 2,000 fictional movies, and 100,000 fictional history events reconciled exactly across 1,030 page commits and 1,030 database reopen boundaries. Thirty-one named tests cover current endpoint contracts, interruption, replacement, migration, malformed pages, count mismatch, token refresh, rate limits, server errors, timeouts, and no-change refresh. No personal Trakt account, app UI, or live provider was used.
completeThe foundation is real.
Milestones 0 and 1 passed: eight native package products now own domain, SQLite, Trakt transport, sync coordination, diagnostics, and fictional scale behavior. Fifteen focused tests pass, and the standalone check reconciles separate launch and deep-history receipts. No app UI or live Trakt access is being claimed.
completeBuild plan tightened.
The iPhone path now has nine measurable milestones, explicit module owners, enforced dependency direction, and separate evidence for tests, devices, Trakt, iCloud, and release readiness.
decidedFive systems remain open.
The requested five complete systems remain available to rank at true 390 × 844 size. They exposed a structural gap, so their results are evidence for the next round—not a final design lock.
Historical updateGlass Horizon became a structural prototype.
Thomas chose the second structural study to develop next. Cinematic Split Horizon keeps Glass Horizon's materials but changes the detail-page hierarchy and interaction; it remains provisional, not final.
Historical updateFoundation before features.
Twelve technical decisions now define local-first Trakt sync, private iCloud overlays, OS 27, lightweight widgets, measurable speed, layered testing, and honest release gates.
decidedEvery admitted capability stays free.
ShowNext is free for this Release: every retained capability ultimately admitted requires no purchase, subscription, or paid product. Goal 8J-A froze the exact promise inventory. The lifetime-purchase path remains deferred R&D and cannot return without a separate Thomas decision.
decidedDiscovery learns without jumping around
ShowNext will personalize stable discovery lanes with visible reasons, reversible feedback, availability context, exploration picks, and honest large-import boundaries.
decidedLibrary care becomes a core promise
ShowNext will pair visible sync verification with safe Watchlist Review, list organization, previewed batches, and recoverable list deletion for large Trakt libraries.
decidedSync stopped at 100 of 740 shows
Trakt now paginates the watched endpoint. Large-library completeness—not another feature tab—is the first technical proof.
completeThe roadmap now requires evidence
Sync laboratory first, then a read-only iPhone spine, safe writes, iCloud overlays, and one platform expansion at a time.
decidedRound one tested five ideas
Spotlight, poster river, release horizon, library pulse, and taste constellation remain visible as exploration—not a hidden discarded draft.
completeiPhone first. Native shells everywhere.
iPad becomes the couch planner, Mac the power-library workspace, and visionOS the ambient living-poster companion.
decidedCompetitors revealed the opening
The category is strong at polished browsing, widgets, and broad feature lists. ShowNext's defensible angle is a visibly trustworthy large-library sync joined to a personal decision and release horizon.
complete