Pinterest video pins failed with "provide cover_image_url ... or
cover_image_key_frame_time" because no cover was sent. Default to
cover_image_key_frame_time 0 (first frame) when the user provides no
cover_image_url — verified against Pinterest's OpenAPI spec (integer
seconds, minimum 0, falls back to the last frame past the duration).
YouTube Shorts derive their required title from the post content, so the
publisher rejects an empty post — but the editor still let it be scheduled.
Gate it in compliance: a youtube_short with empty content is now flagged
(requires_text) so it can't be scheduled, instead of failing at publish.
- publishCarousel now caps images at the platform max so the API/MCP paths
can't bypass the 10-image limit the UI enforces
- waitForProcessing throws on timeout instead of posting against an asset
still being processed; poll attempts/interval extracted to overridable
protected methods for testability
- scope EnsureHasWorkspace to connect entry points + disconnect only;
session-driven OAuth callbacks/selects stay ungated so a missing current
workspace can't HTML-redirect the popup instead of closing it cleanly
- DeleteWorkspace now reassigns affected users to another workspace they
belong to (instead of nulling current_workspace_id), so deleting the
current workspace while owning others no longer bounces to "create".
- Remove the dead LINKEDIN_PAGE_CLIENT_REDIRECT env (the unified flow uses
a single callback; config no longer reads a page redirect).
- useOAuthPopup uses the Wayfinder connect helpers per platform instead of
hardcoding /connect/{platform}.
- Unify the Inertia\Response alias in Mastodon/Bluesky controllers to
InertiaResponse, matching the other connect controllers.
- Add tests: workspace reassignment on delete, and independent refresh of
a linkedin row vs its linkedin-page sibling.
EnsureAccountReady bundled a subscription gate (redirects to onboarding,
SaaS only) with a workspace gate (redirects to workspace creation). The
connect routes can't sit behind it because connecting/disconnecting
happens during onboarding, before a subscription exists.
Split the workspace gate into a standalone EnsureHasWorkspace middleware:
- EnsureAccountReady is now subscription-only.
- EnsureHasWorkspace redirects to workspace creation when there is no
current workspace, in both SaaS and self-hosted modes.
- The social connect group gains EnsureHasWorkspace; the main app group
gains it alongside EnsureAccountReady (listed after it, so the
subscription gate still runs first — no custom middleware priority).
- The repeated `if (! $workspace) redirect()` guard is removed from the
connect/store/authorize/disconnect/index/toggle handlers, and their
return types are tightened (no more dangling RedirectResponse).
LinkedIn connect's no-workspace path changes from a popup callback to the
same redirect as the other platforms.
Replace the per-platform native form POST + manual CSRF + JSON/Blade
popup callback with a single Inertia mechanism:
- popupCallback() renders the accounts/PopupCallback Inertia page (notifies
the opener + closes the popup) for both the GET OAuth callbacks and the
selection submits. Drops the auth.social-callback Blade view, the
expectsJson JSON branch, and useHttp/useSocialConnect on the frontend.
- Selection/credential pages (LinkedIn, Facebook, Instagram, Bluesky,
Mastodon) use Inertia useForm: automatic CSRF + native validation errors.
- Unify the three selection screens on one row + View/Choose layout; the
LinkedIn company tag now uses a building icon.
- Bluesky auth failures throw ValidationException (422 for XHR, redirect
back with errors otherwise).
Controller tests updated from assertViewIs/assertViewHas to assertInertia.
LinkedIn was the only multi-image platform allowing 20 images; align
maxMediaCount/maxImages/maxFiles to 10, consistent with Facebook,
Instagram and Threads. The AI generation cap (MAX_GENERATED_IMAGES)
was already 10 and is unaffected.
Add direct coverage for the centralized media-type logic: every legacy image/video extension arm of Type::fromExtension, and a new MediaItem DataTransferObject test (fromArray's extension->mime backfill — previously the only zero-coverage branch — plus is* delegation and the null-mime extension fallback). Locks the broad classification lists so a dropped format fails a test.
Media-type detection was duplicated across Media, MediaItem (byte-identical copies), HasMedia::getMediaType, the ContentTypeCompatibleWithMedia rule, and six publishers — each hardcoding MIME prefixes and divergent extension lists (Media's video list even had avi/webm/mkv, contradicting the Type enum's mp4/mov).
Add classify(), fromExtension(), and isGif() to Media\Type as the single source for 'what kind is this?' (broad classification), distinct from fromMime()/allowedMimeTypes() (the strict upload allow-list). Every detector now delegates to the enum; the broad extension lists and MIME prefixes live only there. Behavior-preserving (full suite green); adds direct tests for classify/fromExtension/isGif.
The member and company-page publishers were ~95% identical (~1200 lines). Extract AbstractLinkedInPublisher holding the whole pipeline (media-inferred dispatch, post/carousel/document payloads, image/video/document upload, processing polling, error handling); each concrete publisher now only declares its platform, author URN, and (for pages) the post URL. ~640 duplicated lines removed, single source of truth.
Fix: multi-image posts sent each image URN under 'media', which LinkedIn's multiImage schema rejects with a 422 (unrecognized field 'media' / required 'id' missing). Send the URN under 'id'. Add assertions locking the 'id' field on both carousel tests, a company-page multi-image test that was missing, and a fail-fast assertion that a missing org id makes no HTTP call.
The accounts grid and onboarding both build their connect-card list from cases()->filter(isConnectable()) now, leaving the static enabled() helper without any production caller. Remove it and the two unit tests that only exercised it.
The unified LinkedIn card grouped connected accounts by looking the platform up in the rendered card list, but linkedin-page is no longer a card — so an org-only connection fell through to its own network key and the card showed 'Connect' for an already-connected page, with no way to disconnect or reconnect. Expose each account's network on SocialAccountResource and group by account.network instead.
Onboarding still built its connect grid from SocialPlatform::enabled() instead of isConnectable(), so it rendered a standalone LinkedIn Page card with a missing logo whose connect button hit the deleted connect/linkedin-page route (404). Filter by isConnectable() to match the accounts page.
Collapse LinkedIn to one content type per account kind (linkedin_post, linkedin_page_post). Publishers infer the publish format from the attached media — text, single image/video, multi-image carousel, or PDF document — matching how facebook_post/x_post already work; PDF is exclusive of any other attachment. Removes the editor variant picker, keeping only the PDF document title field. Includes a data migration collapsing the retired carousel/document content types.
Replace the two LinkedIn account cards with a single Connect LinkedIn button: one unified OAuth grant (linkedin-openid driver, union of scopes) then a post-callback identity picker to post as the personal profile (linkedin) or a company page the member administers (linkedin-page). The chosen organization is validated against the admin-verified list from the OAuth grant. Per-capability gating via LINKEDIN_ENABLED / LINKEDIN_PAGE_ENABLED supports profile-only or org-only self-hosting. Removes LinkedInPageController, LinkedInTokenSynchronizer, the standalone linkedin-page connect routes, and the unused redirect_page config.
- Add a PublishToSocialPlatform test that runs the real job -> publisher chain (no publisher mock) for a linkedin_document post, asserting the documents API flow and the published platform state
- resolveDocumentTitle selects the document item explicitly (first(isDocument)) to match publishDocument, instead of the first media item
Review follow-ups before QA:
- Editor: getMediaIncompatibilityReason rejects a PDF on non-document content types, so the schedule gate and variant auto-snap match the backend rule (compliance i18n in en/es/pt-BR)
- MCP UpdatePostTool: validate effective content_type vs stored media on schedule, closing the schedule-without-content_type gap; share entriesForUpdate/errorsFor with the API path
- Tests: Platform allowedMediaTypes contains Document, URL-attach of a PDF (LinkedIn ok / TikTok rejected), multi-platform PDF rejection, document init-failure/missing-URN, Page publisher PROCESSING_FAILED
Make the document (PDF) exclusivity validation — previously web-only — also apply when scheduling/publishing via the public API and MCP, so a misconfigured post can't slip through these entry points.
- ContentTypeCompatibleWithMedia: stored-media fallback for partial updates + a stored-state assertStoredPostCompatible(Post)
- MCP PublishPostTool: assert stored-state compatibility before publish (the media-side mirror of assertStoredPostPublishable)
- API UpdatePostRequest: validate each platform's effective content_type against effective media on schedule/publish (covers publishing without resubmitting content_type)
- MCP UpdatePostTool: apply the rule on schedule with stored-media fallback
- Tests: API + MCP happy + rejection paths, plus rule fallback/precedence units
Add LinkedIn document posts — the swipeable PDF carousel — for both personal profiles and company pages. This is the format every major competitor exposes via native PDF upload, and the reason a trial user churned.
- New 'document' media type (application/pdf) across the upload pipeline (Type enum, HasMedia, FormRequests incl. chunked, Platform media types)
- New LinkedInDocument / LinkedInPageDocument content types: PDF-only, single-file, with a supportsDocument() flag
- Publisher flow: documents initializeUpload -> PUT -> poll AVAILABLE -> post with content.media.{id,title}; optional document_title meta (falls back to file name)
- PDF is mutually exclusive with image/video, enforced in ContentTypeCompatibleWithMedia
- Frontend: 'Document (PDF)' variant, media rules (100MB cap), composer/gallery/detail PDF cards, real PDF embed in the LinkedIn editor preview, i18n in en/es/pt-BR
- Tests: publishers (personal + page, incl. processing-failure path), enums, compatibility rule, chunked PDF upload, API + MCP document_title round-trip
LinkedIn caps documents at 100MB / 300 pages (Documents API). The page limit is enforced by LinkedIn at publish, not validated client-side.
Completes the persona grid to a 3x3 with two high-value segments:
Marketer (in-house marketing / social media, a scheduling tool's core user)
and Online store (e-commerce). Labels in all three locales, with
IconSpeakerphone and IconShoppingBag glyphs. Other stays last.
Adds a Developer option to the onboarding persona step (after Freelancer),
with labels in all three locales (Developer / Desarrollador / Desenvolvedor)
and an IconCode glyph in the persona grid. The option flows to the frontend
via Persona::cases() and is accepted by the existing enum validation.
A Bluesky post embed is images XOR video, so a post that carries both can't
be published there. Enforce it per selected platform:
- Add ContentType::supportsMixedMedia() (false only for BlueskyPost) and reject
image+video in ContentTypeCompatibleWithMedia when the type forbids it.
- Mirror it client-side (useMediaRules forbidsMixedMedia + usePostCompliance)
so the editor blocks scheduling with an inline reason before submit, like
every other per-platform compatibility check.
- Add the no_mixed_media message in all three locales (en/es/pt-BR).
- Cover the rule and enum, including the GIF-counts-as-image case.
Builds on the video-upload feature with house-style cleanup and fixes:
- Extract downloadToTempFile / unwrapJobStatus / videoUploadFormat helpers and
guard tempnam/fopen on the image path too (no magic numbers; named consts).
- Send the real content-type and extension for Bluesky's four accepted formats
(mp4, mpeg, webm, mov), falling back to mp4 for anything else.
- Mint the upload and status service-auth tokens once and reuse them.
- Bound the whole upload/poll/retry flow to a wall-clock budget under the queue
job timeout so a stuck transcode degrades to a text post instead of being
killed mid-flight.
- Config-drive the poll interval, max video size, and video-service hosts.
- Expand BlueskyPublisherTest to cover every branch: 409 with and without a
blob, upload/status token failures, did:plc fallback and did:web resolution,
the format mapping, retries, and timeout.
Address review feedback and a transient failure mode seen in production:
- Guard tempnam()/fopen() returning false (bail cleanly instead of a
TypeError later) and guard the finally-block unlink() with is_string().
- pollVideoJob() now bails on a non-2xx getJobStatus (e.g. expired token)
instead of sleeping to the timeout and masking the real error.
- resolvePdsEndpoint() maps did:web colon segments to host + path per the
did:web spec (did:web:example.com:user:alice -> .../user/alice/did.json)
instead of treating the whole remainder as a host.
- The video service occasionally fails a job transiently
(JOB_STATE_FAILED "Failed to process video") for valid input, so the
upload+poll is retried up to VIDEO_UPLOAD_ATTEMPTS times (split into a
new attemptVideoUpload() helper; the upload service-auth token is minted
once and reused across attempts).
Adds a test covering retry-after-transient-failure. Full suite (23) green.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
BlueskyPublisher only attached images; posts whose media was a video
published as text-only with no embed. Bluesky videos do not use the PDS
uploadBlob path — they go through the separate video service
(video.bsky.app), which transcodes the file and writes the resulting blob
back to the account's PDS.
Add a video path that runs when no image embed was built (a post carries
either images or one video, never both):
1. Resolve the account's real PDS host from its DID document (entryway
accounts store bsky.social as `service`, but the video service-auth
audience must be the actual *.host.bsky.network PDS).
2. Mint a com.atproto.server.getServiceAuth token scoped to uploadBlob.
3. POST the bytes to app.bsky.video.uploadVideo.
4. Poll app.bsky.video.getJobStatus until the blob is ready, then embed
it as app.bsky.embed.video.
Failures (download, oversized >100MB, service-auth, transcode) log and
return null so the post still publishes as text rather than crashing the
job — mirroring uploadBlob(). New NSID constants live in BlueskyLexicon
and the video service / PLC directory hosts are configurable.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- store(): members without connected accounts no longer get redirected into
the now-admin-only /accounts (403); non-managers go to the calendar with
the same flash, admins still go to /accounts
- cover the SyncPostPlatforms can('update') gate (viewer creates no platform
rows; member does) and the store redirect split, in WorkspaceRolePermissions
- cover PostPolicy::duplicate viewer-denied
- docs sidebar link uses the canonical https://docs.trypost.it
- drop orphaned sidebar.support.{discord,last_updates} keys in all locales
- remove the explanatory isLocked comment in Edit.vue
Viewers are typically the client: they need to open a draft in the editor
to use the comments tab, but must not change anything.
- post editor (edit) now authorizes view, so viewers can open it; the
composer + schedule tab render read-only and the comments tab stays
interactive (defaults to the comments tab for viewers)
- all mutations stay member+ (update/delete) — the autosave/save/publish/
schedule/delete affordances are hidden and the PUT is still 403 for
viewers; SyncPostPlatforms only runs for users who can update
- drafts route to the editor for everyone again (reverts the read-only
Show detour); Show stays the published-post view
- /accounts now authorizes manageAccounts (admin+), so viewers and members
get 403; the Connections sidebar item is admin+ only and the connect/
disconnect grid is reverted to main (no per-button gating needed)
Tests: draft→editor redirect for every member, viewer can open the editor,
viewer cannot save, and only admins+ can open /accounts.
A viewer clicking a draft/scheduled post landed on the editor route
(authorizes update) and got a 403. The post list/calendar routed every
editable post to the edit page, and PostController@show redirected
draft/scheduled posts to the editor for everyone.
- show only redirects to the editor when the user can update the post;
viewers get the read-only Show page
- posts index + calendar route to show (not edit) when the user cannot
create posts
- cover viewer-sees-show, member-redirected-to-editor, and
viewer-403-on-direct-edit in WorkspaceRolePermissionsTest
Viewers could mutate posts, automations and trigger AI write endpoints,
and every role saw create/manage affordances that 403'd on click.
Backend (security):
- PostPolicy update/delete now require member+ (was tenancy-only), which
also gates the AI write endpoints that authorize('update')
- AutomationPolicy create/update/delete require member+; activate/pause
delegate to update
- AutomationController authorizes index/store/show; AnalyticsController
authorizes view
- Comments stay open to members incl. viewer (by design)
Frontend (UI gating via new useWorkspaceRole composable):
- Sidebar: create post / create workspace / automations / library nav
- Accounts grid: connect / disconnect / reconnect (admin+)
- Members: invite / change role / remove / cancel invite (admin+)
- Account billing tab (owner); posts index + calendar create affordances
Tests: PostPolicyTest, AutomationPolicyTest (all four roles) and an
end-to-end WorkspaceRolePermissionsTest; aligned the automation test
suites' account/workspace setup with role pivots.
All customers were migrated to the single Workspace plan and the old plan rows
were deleted in production, so drop the now-dead legacy tiers from the code:
reduce the Plan Slug enum to Workspace only and default the PlanFactory to it.
Rework StripeEventListenerTest around the single Workspace plan (its monthly and
yearly price ids still exercise plan-by-price mapping, trial clearing, deletion,
and previous-plan propagation), and point the remaining tests that referenced
the starter/pro slugs at the seeded Workspace plan.
The remove/role-change guards only protected the account owner, so a non-owner
admin could change their own role or remove themselves via a crafted request
(the UI hides it, but the backend didn't). Add an explicit self-guard to both
updateRole and removeMember.
Lock the whole role system with tests: accept assigns the exact invited role
(viewer/admin/member), invite requires and persists a role, updateRole supports
viewer and blocks self/owner/invalid, removeMember blocks self/owner, and a
viewer is read-only (view yes; create post / manage team / invite no).
AcceptInviteController attached invited users with a hardcoded member role,
ignoring the invite's role entirely (a viewer invite joined as member). Use the
invite's role on accept, require role on invite creation (no silent default),
and drop the role from the request/CreateInvite defaults. Add the Viewer option
to the member role dropdown (now iterates all roles), hide the dropdown on the
current user's own row, and require the member's email to confirm removal.
Covered by tests asserting the accepted role matches the invited role for
viewer/admin/member, plus invite role validation.
Pennant: all feature flags were replaced by BillingCycle, so remove the package
(composer), the now-empty app/Features discovery, the pennant-development skill,
and replace the create_features_table migration with a drop_features_table.
Timezone: the registration flow collected a user timezone (seeder, request
validation, hidden field) but no timezone column ever existed and CreateUser
discarded it. Remove the dead handling, the orphaned Timezone rule, and the
tests that covered the now-removed validation.
- LinkedInController: catch NetworkAlreadyConnectedException so a profile
colliding with a connected Page shows the specific network_taken message
instead of the generic error (+ test).
- Billing.vue: pass a string to transChoice replacements (vue-tsc TS2322).
- NetworkConnectGrid: drop the now-unused 'connect' emit (no listener after
AddSocialDialog removal).
- Tests: assert the BillingCycle trial window; fix a stale section comment in
StripeEventListenerTest.
Send the user's persona on both the server-side subscription.created event and
the client-side checkout.completed/dataLayer purchase event, for ICP analysis.
Hold the processing screen ~5s before redirecting so PostHog and the ad pixels
(Google/Meta via GTM) reliably flush. Sharpen the annual-upgrade banner copy
(lead with '2 months free') and give its check icon a white tile.
Add a connect step between persona selection and Stripe checkout: persona ->
connect >=1 network (server-enforced) -> checkout. Extract the network grid into
a shared NetworkConnectGrid used by both onboarding and the accounts page, which
is redesigned from a table into the same grid (one account per network, with
per-card connect / connected+disconnect / reconnect states). Drop the openDialog
auto-open flow now that networks are shown inline.
When a Telegram channel connection is rejected (e.g. the workspace already has a
channel for that network), broadcast TelegramConnectFailed so the connect dialog
shows a clear error message instead of waiting until the code silently expires.
Credit allotment is now derived directly from BillingCycle::for($account)->creditAllotment()
instead of a cached Pennant feature, removing the dynamic cache-invalidation footgun
(forgetPlanFeatureCache) that had to be called from every subscription/workspace mutation.
Bugs (both with regression tests):
- OnboardingController::store now guards already-subscribed accounts (mirrors
index), preventing a second Stripe Checkout / double subscription if a
subscribed user re-POSTs /onboarding.
- SocialAccountObserver: drop the `platform_user_id != …` clause from the
creating-time one-per-network check. On create there is no "self" to exclude,
so it only weakened the rule — the same account connected via two network
variants (e.g. Instagram standalone + via Facebook, same id) could slip a
second account into the network. Now any account in the network blocks.
Robustness:
- CreateWorkspace wraps create + member attach + switchWorkspace in a
transaction (cache-forget / quantity-sync run after), so a partial failure
can't leave an orphan workspace that inflates the Stripe seat count — covers
both the signup and the add-workspace paths.
Test honesty & coverage:
- Scope the ten "connect multiple <platform> accounts" tests to self-hosted
mode (config + name); they only passed because the test env defaults
SELF_HOSTED=true, and in cloud the one-per-network rule blocks them.
- network_taken popup now has controller-level tests on all six OAuth
controllers (added Threads, YouTube, LinkedInPage, and a new
InstagramFacebook test file; LinkedInPage/InstagramFacebook also exercise
variant collapse).
- Wiring tests that creating/deleting a workspace actually calls
syncWorkspaceQuantity (guards per-seat billing against silent breakage).
- Strengthen TrialLengthTest to assert the configured length reaches
trial_ends_at; add a same-id network-variant block test.
- CreateUser: provision the default workspace OUTSIDE the signup DB transaction
so CreateWorkspace's cache-forget / Stripe-quantity sync never runs inside a
transaction (consistent with WorkspaceController::store); user+account stay
atomic and a failed workspace create degrades to /workspaces/create.
- BillingController::swapToYearly: explicit null guard on the resolved
subscription before dereferencing stripe_price.
- UsageController: reuse the already-loaded workspaces collection for the count
instead of issuing a second query.
- Add a GitHub OAuth-callback test covering the default-workspace auto-create
(mirrors the Google one).
- CreateUser now provisions a default "{name}'s Workspace" for non-invite
signups (email, Google, GitHub, seeder), with the owner as Admin and
current_workspace_id set. Invite signups still skip it (they join the
inviter's workspace on acceptance).
- UserSeeder drops its now-redundant explicit CreateWorkspace call.
- After Stripe checkout, billing/processing lands the user on the social
accounts screen with the connect dialog open (was: calendar), so onboarding
flows straight into connecting an account.
Quantity stays correct (1 workspace → checkout quantity 1). The first workspace
is created without brand details; brand setup remains available in Settings and
on the create-workspace screen for additional workspaces.
- Add a controller-level test asserting the Facebook OAuth callback returns the
localized network_taken popup (not a generic error) when the network is
already connected — covers the bug-A fix across the OAuth callbacks.
- WorkspaceController::autofillBrand: use $request->validated('url') instead of
data_get on the validated array (idiomatic FormRequest accessor).
Bug fix
- Surface the localized "network already connected" message on the
Facebook/Instagram/InstagramFacebook/LinkedInPage/Threads/YouTube OAuth
callbacks: catch NetworkAlreadyConnectedException before the generic catch
so the conflict no longer falls through to a generic error + Log::error.
Scope / dead code
- Remove the orphaned BillingController::checkout() + app.billing.checkout route
(onboarding starts checkout directly); delete the now-dead DiscordWidget and
useFeatureAccess composable; prune orphaned i18n keys left by removing the
plan picker / upgrade dialog / count limits (billing.subscribe.*,
accounts.limit_reached, workspaces.limit_reached, common.discord.*).
- Drop the unused `plan` prop from the usage page and the unused `label` from
the onboarding persona payload (labels come from i18n); remove
Persona::options()/label().
Conventions
- declare(strict_types=1) on the two new migrations.
- Extract autofill validation into AutofillBrandRequest (FormRequest).
- Drop the unused $plan param from AccountPolicy::swapPlan.
- CreateUser: drop the stale config('cashier.trial_days', 7) fallback (now 8).
- Rename LimitEnforcementTest to InvitePermissionTest; use Pest mock() helper in
WorkspaceQuantitySyncTest; move shared test helpers into Pest.php.
- Memoize BillingCycle window() + subscription lookup.
Pricing
- Bill per workspace ($12/mo or $120/yr each); Stripe quantity tracks the
workspace count and syncs on workspace create/delete.
- 2,500 AI credits per workspace, pooled at the account level; monthly reset
on the billing anniversary, annual granted upfront (no rollover).
- One social account per network per workspace; remove all count-based limits
(workspace/social/member) and the legacy plan tiers (single Workspace plan).
Onboarding (cloud only: SELF_HOSTED=false + PostHog)
- Replace the /subscribe plan picker with /onboarding persona selection
(Creator/Freelancer/Startup/Agency/Small business/Other), saved on the user
(users.persona) and mirrored to PostHog, then Stripe Checkout on the monthly
price. 8-day trial so Stripe displays 7.
Billing screen
- Remove the Change Plan dialog (dead with a single plan); add an annual-upgrade
banner for monthly subscribers (swapToYearly).
- Current-plan card shows the workspace count instead of the plan name.
System AI
- Brand analyzer / workspace autofill is always allowed and never debits credits
(system feature, not the user's usage).
Self-hosted (SELF_HOSTED=true) bypasses all billing, credit, limit, network,
and onboarding logic.
The Generate node data rules had no rule for 'style', so validated()
dropped it on save — the run always fell back to the default image_card.
Add a Rule::in(ContentStyle) rule so the selected style is persisted and
unknown values are rejected.
Replace the 'single'/'carousel' string format threaded through the AI
generation path with a typed GeneratorFormat enum (PostContentGenerator,
PostContentHumanizer, PostContentStreamer, StreamPostCreation,
RunGenerateNode, deriveFormat).
Remove generatorFormat() from AiContentTemplate and the three templates —
it conflated style and format (returning 'tweet_card'). The tweet-card
check now lives on ContentStyle via isTweetCard(), and the humanizer skip
on ContentStyle::humanizes(), so the style declares its own behaviour
instead of callers comparing magic strings.
- TemplateContext gains bool $isCarousel (set from format === CAROUSEL_FORMAT in StreamPostCreation)
- AiContentTemplate interface: promptView() now takes TemplateContext
- ImageCardTemplate handles both single and carousel via context.isCarousel; schema() returns the correct shape per mode; assemble() routes to assembleSingle/assembleCarousel
- TweetCardTemplate handles both single and carousel; promptView() returns tweet_card_carousel view when isCarousel; schema() returns {tweet_text} or {caption, slides[]}; assemble() calls forTweetCard (single) or forTweetCardCarousel (carousel)
- CarouselTemplate deleted — carousel is now a format dimension each style handles, not a standalone style
- AiTemplateRegistry: [ImageCardTemplate, TweetCardTemplate] only
- PostImagePipeline: adds forTweetCardCarousel — renders one renderTweetCard per slide text, returns N media items
- StreamPostCreation: carousel detection restored to $format === CAROUSEL_FORMAT (like main); style dispatched via AiTemplateRegistry::find($this->template); humanize skips tweet_card (single and carousel)
- PostContentGenerator: promptView() call updated to pass templateContext
- New prompt: resources/views/prompts/post_content/tweet_card_carousel.blade.php
- Tests: TemplateContractTest and TweetCardTemplateTest updated for promptView(context) signature; AiTemplateRegistryTest asserts [image_card, tweet_card]; StreamPostCreationTest adds tweet_card carousel test; TweetCardTemplateTest covers both schema shapes and both promptView return values
