Backend exposes account and plan as proper Inertia resources so any
page can read them off shared props:
- AuthAccountResource — id, name, created_at
- AuthPlanResource — id, slug, name, interval (derived from the active
Cashier subscription's stripe_price)
Both wired into HandleInertiaRequests::share so `auth.account` and
`auth.plan` are available everywhere.
Frontend pushes app + identity context to GTM's dataLayer on every
page load via initializeDataLayer (resources/js/datalayer.ts), and
useTracking gets begin_checkout/purchase wired in the billing flow.
The pushes are no-ops without GTM configured (the array still exists
in memory, nothing reads it). Self-hosted instances without GTM_ID
incur zero error noise. The GTM partials in app.blade.php were
already conditional on `config('services.gtm.id')`.
Processing.vue polls `auth` alongside `subscriptionActive` so
auth.plan.interval is fresh once the Stripe webhook creates the local
Subscription row (it doesn't exist yet at the initial render). The
watch fires only on the false → true transition; an onMounted
fallback redirects users that land on the page with an already-active
subscription without re-firing trackPurchase.
Drops the dedicated /settings/authentication/providers/{provider}/callback
route added in the previous refactor — registering a second callback URL
in each OAuth app is more ops cost than the trade is worth.
Back to one callback URL per provider, with a small `Auth::check()`
branch in the auth controllers' callbacks. The check is safe because
the redirects that initiate the round-trip enforce the right
middleware (signup/login is `guest`-only, connect is `auth`-only),
so the auth state at callback time matches the flow's intent.
Splits the OAuth connect flow off entirely from signup/login so each
route has a single responsibility.
- routes/auth.php returns to its original state — redirect + callback
both back inside the `guest` group.
- routes/app.php gains a paired callback route at
/settings/authentication/providers/{provider}/callback.
- AuthenticationController::connectProvider /
connectProviderCallback override Socialite's redirectUrl so the
round-trip stays on the connect-flow URL. The Auth::check() branch
in the auth controllers is gone.
The OAuth apps in Google Cloud and GitHub Developer Settings need the
new callback URL registered alongside the existing one — documented in
.env.example.
The Connect button on /settings/authentication pointed at the
auth.{provider}.redirect routes that live behind `guest` middleware,
so authenticated users were bounced to /app/home before reaching
Socialite. The OAuth callback also needed to handle two flows
(signup/login vs link to current user) but had no branch for the
second case — meaning a different-email GitHub account would have
been registered as a new user, logging the original session out.
Splits the flows by intent:
- New `app.authentication.connect-provider` route in the auth group,
handled by the settings controller (where it sits next to
disconnect-provider). Replaces the OAuth signup link as the
Connect button's target.
- Auth callbacks moved out of the guest group (still one URL per
provider, since OAuth apps only register one) and gain a single
Auth::check() branch that calls connectToCurrentUser().
- connectToCurrentUser() rejects if the provider id already belongs
to a different user; otherwise sets it on the current user and
redirects back to settings with a flash message.
Persists marketing attribution and registration metadata for new users
across the three signup paths (email, Google, GitHub):
- 5 utm_* columns + registration_ip on the users table
- PreservesUtmParameters trait stores incoming utm_* query params on
the register/redirect GET, retrieves them on the POST/callback —
surviving the OAuth round-trip via session
- request()->ip() captured at the controller layer
Adds GitHub as a second OAuth provider:
- GitHubController mirroring the Google one (now renamed from
SocialLoginController for symmetry)
- Settings → Authentication can connect/disconnect GitHub like Google
- Single SocialLogin.vue component replaces the per-provider buttons
on Login/Register, rendering each enabled provider plus a single
"or continue with" divider
UserFactory gains defaults for the new nullable columns so model
strict-mode access in tests doesn't trip.
Adds POST /api/posts/{post}/media for direct file (multipart) upload
and renames the existing URL-based flow to /api/posts/{post}/media/from-url
so the path matches HTTP semantics (POST <resource>/media expects a file
body, not JSON URLs).
The multipart action validates type against the post's enabled platforms
(image rejected on TikTok-only posts), enforces per-type size caps, and
reuses Workspace::addMedia + Post::appendMedia. URL-based attaching is
unchanged behaviorally — only the route name and controller method are
renamed for symmetry. The MCP AttachMediaFromUrlTool was already named
correctly and needs no changes; binary upload via MCP is a host-protocol
limitation that no MCP server (including Postiz) supports.
Three things in one move:
1. Centralize per-type size caps in config/trypost.php under media.max_size_mb.
The MediaType enum now reads from there:
MediaType::Image->maxSizeInMb() // 10 (env: MEDIA_IMAGE_MAX_SIZE_MB)
MediaType::Video->maxSizeInMb() // 1024 (env: MEDIA_VIDEO_MAX_SIZE_MB)
Plus convenience helpers maxSizeInBytes() and maxSizeInKb() so callers
don't have to multiply themselves. StoreAssetRequest now uses
MediaType::Video->maxSizeInKb() in its 'max:' rule and mimes derived
from MediaType::{Image,Video}->allowedMimeTypes(). storeChunked
validation moved to the new StoreChunkedAssetRequest FormRequest
(also reads from the enum). MediaAttacher uses
MediaType::Video->maxSizeInBytes() as the streaming-abort threshold
and enforces the per-type cap after MIME resolution.
2. Drop MediaType::Document. We never accepted PDFs anywhere — the
StoreAssetRequest mimes list excluded them, the storeChunked
extension regex excluded them, MediaAttacher excluded them. The only
places that referenced Document were:
- Platform::allowedMediaTypes for LinkedIn/LinkedInPage (declared
but unreachable)
- HasMedia::getMediaType fallback when MIME wasn't image/video/*
Both now cleaned up. HasMedia::getMediaType throws
InvalidArgumentException for unsupported MIMEs instead of silently
returning a fake 'document' type. Platform::LinkedIn now matches
every other social platform: [Image, Video].
3. Add MediaType::fromMime($mime): ?self — replaces the inline mime →
type loop that MediaAttacher used to roll. Returns null for
unsupported MIMEs (caller decides how to react).
Tests:
- MediaTypeTest rewritten for the new shape (no Document, config-driven
sizes, fromMime + size-helper coverage).
- PlatformTest no longer asserts Document on LinkedIn.
- HasMediaTest replaces the 'detects document type' case with one that
asserts the throw on unsupported MIMEs. The 'add media from path'
test now uses real PNG bytes from the fixture.
- AssetControllerTest chunked tests use real PNG bytes and assert 422
(FormRequest unprocessable) for malformed Content-Range headers,
matching the new validation layer.
The MediaAttacher used to roll its own SSRF guard with DNS resolution
and a static fakeUrlSafety() flag for tests. Validating URLs is a
request-layer concern, not a service-layer one. Laravel ships
'active_url' which does the same DNS resolvability check via
dns_get_record — applying it at the FormRequest / MCP validate() level
catches dead URLs upfront with a proper 422 instead of letting the
download silently fail.
- Replace the inline 'urls.*' => ['url:http,https'] rule with
['url:http,https', 'active_url'] in both Api/PostController::attachMedia
and Mcp/Tools/Post/AttachMediaFromUrlTool.
- Drop isUrlSafe(), fakeUrlSafety(), resetUrlSafety(), $skipUrlSafety
from MediaAttacher. The remaining defenses (Http::sink streaming +
progress abort at MAX_BYTES, allow_redirects: false, MIME allowlist)
cover the operational concerns.
- Restore tests/TestCase to the original setUp — no SSRF bypass needed
anymore because active_url is satisfied by the test hosts.
- Swap synthetic test hosts (cdn.example.com / evil.example.com) for
example.com / example.org. Both are RFC-reserved AND have stable A
records, so active_url accepts them while Http::fake() still
intercepts the actual request.
For SSRF defense beyond 'active_url' (which doesn't block private IPs),
trypost relies on production network egress controls. Open-source
self-hosters who run without a firewall accept the corresponding risk;
that's a deployment concern, not a request validation concern.
The previous suite asserted happy paths and a couple of basic field
omissions but didn't probe the rules themselves. Adds 26 tests
across 5 files:
REST API (tests/Feature/Api/PostApiTest.php) — 9 new:
- content_type not in the enum
- content_type mismatched with the social account's platform
- label_id from another workspace
- platforms[].id from another post on update (cross-post leak)
- content_type mismatched with the post_platform on update
- status=scheduled requires future scheduled_at
- status=draft works with no scheduled_at
- past scheduled_at on store
MCP create-post-tool (tests/Feature/Mcp/PostToolTest.php) — 5 new:
- inactive social account
- content_type not in the enum
- content_type mismatched with the social account's platform
- label_id from another workspace
- already had: scheduled_at past
MCP update-post-tool (tests/Feature/Mcp/PostPublishToolTest.php) — 2 new:
- platforms[].id from another post (regression for the new
Rule::exists scoping)
- content_type mismatched with the post_platform
MCP attach-media-from-url-tool (tests/Feature/Mcp/AttachMediaFromUrlToolTest.php) — 3 new:
- non-http(s) scheme (ftp://...)
- malformed url string
- more than 10 URLs per call
Custom rules unit tests — 2 new files:
- ContentTypeMatchesPlatformTest covers happy path,
cross-platform mismatch, the Instagram + InstagramFacebook
compatibility bridge, and the no-op cases (missing account_id,
unknown content_type — those are caught by Rule::in elsewhere).
- ContentTypeMatchesPostPlatformTest covers the equivalent shape
for the update flow that pivots through post_platform.id.
Code-review surfaced two correctness bugs and a security gap that
needed to land before merging.
- UpdatePost::execute disabled every platform when called without
a `platforms` key. PublishPostTool relied on that path, so every
publish-via-MCP queued a job whose handler then found nothing
enabled to publish to. Wrap the platform toggle in
`Arr::has($data, 'platforms')` (matches the existing label_ids
guard a few lines up). Add a regression assertion to
`PostPublishToolTest::publish post immediate dispatches PublishPost
job` that the previously-enabled platform stays enabled.
- StorePostRequest declared rules for only `platforms`,
`scheduled_at`, and `status`. `validated()` then stripped
`content`, `media`, and `label_ids`, so REST `POST /api/posts`
silently created empty drafts. Added rules for content / media /
label_ids (with workspace-scoped `Rule::exists` for labels) and
dropped the unused `status` field — REST callers transition state
via `PUT /posts/{id}`. Removed the dead `platforms.*.content`
rule. Added a feature test that asserts content + media + labels
roundtrip on create, plus a regression that an `is_active=false`
social_account is rejected at validation.
- CreatePost::execute now syncs label_ids itself so REST and MCP
share the behavior. Removed the duplicate sync from CreatePostTool.
- MCP UpdatePostTool didn't scope `platforms.*.id` to the post being
updated, drifting from the REST UpdatePostRequest which adds
`Rule::exists('post_platforms','id')->where('post_id', ...)`. Now
it loads the post first (failing fast with `Post not found.` if
the workspace check rejects), then uses the same Rule::exists.
- MediaAttacher fetched any URL the caller passed, including
loopback / link-local / private targets — classic SSRF pivot.
Now `isPublicHttpUrl` rejects non-http(s) schemes, restricted IP
ranges, and DNS hostnames whose A/AAAA records resolve into those
ranges (covers DNS rebinding). Bypassed under
`app()->runningUnitTests()` so `Http::fake()` keeps working.
Streaming the response body lets us abort early once we exceed
MAX_BYTES instead of buffering the full payload first; redirects
are disabled so a 200→302 trick can't bypass the host check.
- The `media[]` JSON column had a lost-update race in
`attachFromUrls`: read `$post->media`, mutate in PHP, write back.
Two concurrent calls clobbered each other. Now wrapped in a
transaction with `lockForUpdate()`.
- ESLint: `resources/js/actions/**` and `resources/js/routes/**`
are auto-generated by Wayfinder on every build. Their import
order matches PHP scan order, not alphabetical, so import/order
fought eslint-fix forever. Added them to ignores.
Lets ChatGPT (MCP) and external clients (REST API) drive the full lifecycle of
a post — create with platform selection, attach media from URLs, schedule or
publish immediately, and fetch engagement metrics — without touching the web UI.
MCP tools added: UpdatePostTool, PublishPostTool, AttachMediaFromUrlTool,
ListContentTypesTool, GetPostMetricsTool, PreviewPostTool. CreatePostTool now
accepts platforms[] + scheduled_at + label_ids; ListPostsTool gains
status/search/limit filters.
REST endpoints added: POST /api/posts/{post}/media, GET /api/posts/{post}/metrics,
GET /api/posts/{post}/preview, GET /api/content-types.
Also fixes a silent CreatePost::execute bug — the action validated platforms[]
but ignored it, so REST callers never saw their selection persisted. Adds cross
validation rules (ContentTypeMatchesPlatform / ContentTypeMatchesPostPlatform)
so a LinkedIn account can't be saddled with x_post, and rejects inactive social
accounts during validation instead of failing silently downstream.
Shared services (PostMetricsFetcher, PostPreviewer, MediaAttacher) back both
MCP tools and REST controllers so behaviour stays aligned. New Resources
(PlatformContentTypesResource, PostMetricsResource, PostPreviewResource,
PostMediaAttachResource) keep controllers free of inline model mapping.
Suite: 1.332 passing, 0 failing — covers web (PostControllerTest), REST
(PostApiTest, PlatformApiTest, PostMediaApiTest), MCP (66 tool tests), and
the publish job (PublishToSocialPlatformTest).
Removes /docs from git tracking and TIKTOK_REVIEW_VIDEO_SCRIPT.md.
Mentions in post comments
- @mention autocomplete (workspace members, current user excluded) with
marker syntax @[uuid] persisted, display names rendered via CommentBody
chips; live edit replaces markers with names and converts back on save.
- NotifyMentions action with workspace-scoped membership check, dedupes
same user, only newly-added mentions on update.
- Email + in-app via SendNotification job, respecting per-user
notification_preferences.mentioned_in_comment.
- Heartbeat-based presence (Cache, 60s TTL, 30s ping) so online recipients
get only the in-app notification — no email noise.
- Real-time bell on workspace.{id}.user.{id} private channel
(NotificationCreated event), scoped channel name avoids client-side
filtering and lays out a convention for future workspace channels.
- Mailable localized via lang/{en,es,pt-BR}/mail.php; Maizzle source
template for the email is committed and built into resources/views/mail.
AI generation refactor (Action layer + MCP)
- Extracted Actions/Ai/Generate{Image,Video} with QuotaExhaustedException
so agent tools and MCP tools share a single domain entry point.
- Mcp/Tools/Ai/Generate{Image,Video}Tool registered in TryPostServer; both
return MediaResource payloads.
- Orientation::imageApiSize maps non-OpenAI ratios to 1:1/2:3/3:2.
- config/ai.php is now the single source of truth driven by env, removing
the trypost.ai shim. Default text/image providers flipped to OpenAI.
Settings/UX
- /settings/workspace split into shadcn Tabs (Workspace / Brand / Users)
with three components.
- /assets and the in-editor MediaPicker open the ImagePreviewDialog
lightbox on image click while preserving action button behaviour.
- Comments tab landed via ?tab=comments&comment=<id> from notification
click (scroll-to + temporary highlight).
- Mention autocomplete popover flips above when near the viewport bottom.
- Real social platform PNGs replace Tabler brand glyphs in schedule
pills and post list, with hover tooltip carrying display_name + handle.
Bug fixes
- AcceptInvite: controller now passes workspace + role payload that the
Vue page expects; login/register CTAs preselect the invite email.
- WorkspaceInvite mailable: stopped referencing nonexistent
$invite->workspace and $invite->role; column added to the migration,
Invite model casts role to WorkspaceRole, CreateInvite persists it.
- PostCommentCreated: added broadcastAs so .PostCommentCreated actually
matches the Echo listener; payload now includes mentioned_users so
receivers render the chip correctly without a refetch.
- Preview components for X/Pinterest/Threads/Bluesky/LinkedIn/Mastodon/
TikTok/YouTube switched from item.type === 'image' to
!isVideoMedia(item) so media without a persisted type still renders.
- UpdatePostRequest now accepts media.*.{type,mime_type,size,...} so the
posts.media JSON keeps the metadata that the previews need.
- Removed throttle:6,1 from social connect routes (was 429ing legitimate
OAuth retries).
- Used MediaType enum cases instead of literal 'image'/'video' strings
when creating media rows.
Tests
- MentionParser unit tests, NotifyMentions feature tests including
online/offline channel selection and preference gating, MCP AI tool
happy paths, MentionedInComment mailable rendering, AcceptInvite +
search-members + index mentioned_users path. 1229 passing.
- gallery: extract /assets tabs (uploads, Unsplash, Giphy) into shared
GalleryBrowser used by both /assets and a new MediaPickerDialog inside the
post editor; add JSON search endpoint for workspace assets with tests
- emoji: replace broken emoji-picker-element web component with a custom
EmojiPicker (full Unicode set, search, categories, recently-used,
light/dark, i18n)
- preview tab: platform selector pills, variant tabs (data-driven from
content_types map) so the user can switch Feed/Reel/Story etc. and have
it autosave through the same handler ScheduleTab uses
- platform logos: shared usePlatformLogo composable (logo + label + content
types); replaces inline maps across 5 components, fixes
instagram-facebook falling back to default.png
- tooltips: hover details (display_name · @username + platform label) on
platform avatars across editor, posts list and calendar
- settings cards: show ` · @username` in the title bar so multiple accounts
on the same network are distinguishable
- routes: drop the throttle:6,1 group middleware on social connect routes
(was 429ing legitimate OAuth retries) and rely on the default limiter
When config('trypost.ai.text_provider') points to a provider whose
API key is populated (services.gemini.api_key or
services.openai.api_key), the autofill action pipes the homepage
markdown through a BrandAnalyzer agent with structured output to
produce higher-quality values for:
- brand_description — 2-3 sentences summarizing the company based on
the actual page content, not the raw meta description (which is
often generic SEO boilerplate)
- brand_tone — classified into one of our seven enum values from the
writing style on the page
- content_language — detected from the actual content (more reliable
than <html lang> which is often wrong)
- brand_voice_notes — concrete writing guidelines inferred from the
site's style, written in the detected language
When the LLM provider is NOT configured (open-source self-hosted
deploys without API keys), the action keeps the existing
deterministic meta-tag-only flow — no crash, no noise, no LLM cost.
When the LLM fails mid-request, we log a warning and fall back to
the meta-tag values so the user still gets something useful.
Stack additions:
- league/html-to-markdown ^5.1 converts the main body to clean
markdown for the LLM input (truncated to 4000 chars).
- BrandAnalyzer agent (Agent + HasStructuredOutput) with schema
enums matching our allowed tones and languages.
- resources/views/prompts/brand_analyzer.blade.php holds the
instructions, including explicit enum lists and examples of good
voice_notes.
Frontend: Brand.vue now also fills brand_tone and brand_voice_notes
from the response when present.
Tests (+3): LLM-configured happy path, no-credentials fallback
(asserts BrandAnalyzer is never prompted via preventStrayPrompts),
and LLM-exception fallback (meta tags win, brand_tone stays null).
Users on the Brand step can type their website URL and click
'Preencher' (Portuguese) / 'Autofill' (English). The backend fetches
their homepage, parses standard meta tags, and returns:
- name ← og:site_name | title (suffix stripped at ' | ', ' - ')
- description ← meta[name=description] | og:description
- language ← <html lang> mapped to en / pt-BR / es
- logo ← apple-touch-icon | largest link[rel*=icon] | og:image
The logo is downloaded, validated (mime whitelist, 2MB max), and
attached to the workspace's 'logo' media collection so the avatar
updates immediately.
Zero LLM calls, zero external APIs. Uses symfony/dom-crawler +
symfony/css-selector (newly required) for meta extraction. Everything
else (Http client, workspace media, Intervention) was already in the
project.
Security:
- SSRF guardrail: resolves the host, rejects private / loopback /
link-local ranges, enforces http(s) scheme on both the initial
page fetch and the logo download.
- Rate limited at 10 req/min per user via the throttle middleware
alias on the route.
- Logo content-type must be one of the allowed image mimes; wrong
types are silently dropped so users never see broken images.
UX:
- 'Autofill' button next to the website input, disabled until there
is a URL; shows a spinner while running.
- If a logo was captured, a small preview appears below the input
so users can see what was pulled before saving.
- Success and error paths both surface as vue-sonner toasts, with
translations in en / pt-BR / es.
- Failures leave the form untouched — nothing is destructively
overwritten if parsing gave us nothing.
Tests (16 new): action-level coverage for happy path, title-suffix
fallback, language code normalization across 6 locales, scheme
rejection, private-range SSRF rejection, implicit https prefixing,
empty sites, upstream errors, and wrong-mime logo rejection. Plus
two controller-level tests for the autofill endpoint.
After users pick their persona (role), they now land on a new Brand
step that collects the same fields available in Settings → Workspace
→ Brand: website, description, tone, voice notes, and content
language. When they continue, every AI-generated post for this
workspace already has sensible defaults — before the user's first
post is ever drafted.
Flow:
Role (persona) → Brand (new) → Connections → Subscription → Completed.
A 'Skip for now' button on the brand step advances to Connections
without touching the workspace (defaults stay at their seed values).
Backend:
- Setup enum gets a new Brand case slotted between Role and
Connections with matching stepNumber updates.
- OnboardingController::brand() renders the form pre-filled from the
current workspace. storeBrand() validates via a new
StoreBrandRequest form request and writes the fields onto the
workspace, then advances setup. skipBrand() just advances.
- storeRole() redirects to brand instead of account. enforceStep()
knows how to redirect users whose setup is Brand.
- Three new routes: GET /onboarding/brand, POST /onboarding/brand,
POST /onboarding/brand/skip.
Frontend:
- New Brand.vue page mirrors the Settings brand form but inside the
onboarding AuthLayout. Tone + language sit side by side, both
selects take full width. Translations added to en, pt-BR, and es.
- Wayfinder regenerated so the page can import storeBrand / skipBrand.
Tests:
- Renamed 'redirects to step2' → 'redirects to brand step' and
assert new setup.
- Added six new tests covering brand step auth, redirects, render,
successful store, validation of tone and content_language, and
skip.
- Updated UserSetupTest for the new enum case + reshuffled step
numbers.
Critical fixes:
- Quota enforcement regressed when the regex flow was replaced with
tools. The LLM was trusted to respect [Session state] quota hints,
which is vulnerable to prompt injection / hallucination. Each tool
now checks Pennant feature limits at the top of handle() and returns
a short quota-exhausted string instead of calling the provider.
Audio shares the video quota (there is no AiAudioLimit by design).
- N+1 on aiMessages: combined the two separate ->get() scans that
produced imagesInThread and videosInThread into a single query.
- postPlatforms lazy load: controller now loadMissing('postPlatforms')
before constructing SocialMediaAssistant, which accesses the relation
in activePlatformRules().
Project conventions:
- \RuntimeException and \Throwable are now imported at the top of the
controller instead of inlined per CLAUDE.md.
- Tool handle() methods use data_get($request, 'prompt') instead of
direct array access, matching the data_get convention for service
classes.
- enrichContent() pluralizes attachment types via Str::plural so
'2 images' reads naturally instead of '2 image'.
Tests:
- Added Storage::fake('public') in tool tests so runs don't pollute
the local disk with fake-generated files.
- Restored quota enforcement coverage (deleted when the flow changed)
as three new tests — one per tool — that fill the monthly bucket
and assert the tool refuses and nothing is generated.
- Removed inline \App\ references in tests in favor of imports.
SocialMediaAssistant::activePlatformRules() reads post_platforms on the
attached Post, maps them to Platform enum values, and resolves rule
classes via Registry::forMany(). The rendered instructions include only
the summaries of platforms actually selected on this post — so when a
user is posting to X only, the agent doesn't waste tokens thinking
about Instagram Reels.
The PLATFORM KNOWLEDGE block in system.blade.php is replaced with
@include('prompts.assistant.platforms') which iterates the passed rules
and renders each summary. When the post has no platforms, the section
renders nothing (a @if guards it).
Each platform has a rule class exposing specs() (char limits, aspect
ratios, media limits, format-specific constraints) and summary() (a
short concise description the agent can render into instructions).
Registered in AppServiceProvider::configurePlatformRules() mapping
Platform enum values to rule classes. Both InstagramFacebook and
LinkedInPage share rules with their non-business siblings.
Contract + Registry follow the lookup-map pattern. Registry is a
static registry seeded at boot — rules are cheap enough to new up
per request.
Tests cover all 12 platforms via Pest dataset and verify the forMany
fan-out, clear/register behavior, and representative summaries.
Controller now delegates to SocialMediaAssistant::prompt() and reads
generated attachments from AttachmentCollector (request-scoped). The
three preg_match branches for [GENERATE_IMAGE/VIDEO/AUDIO] commands
are gone — the LLM invokes tools directly with typed parameters.
Deleted:
- app/Services/Ai/GeminiTextGenerationService.php
- app/Services/Ai/TextGenerationService.php (OpenAI alternative)
- app/Services/Ai/ImageGenerationService.php
- app/Services/Ai/AudioGenerationService.php
- app/Services/Ai/Contracts/TextGenerationInterface.php
Kept: VideoGenerationService (wrapped by GenerateVideo tool since
Veo is not in the SDK's provider matrix) and IntentDetector.
Tests now fake the agent via SocialMediaAssistant::fake() with either
canned text or a callable that simulates tool side-effects by pushing
directly into AttachmentCollector.
Each tool implements Laravel\Ai\Contracts\Tool with description(),
handle(Request), and schema(JsonSchema). They share a common pattern:
- Constructor receives Workspace, optional Post, userId, and an optional
AttachmentCollector (resolves from container if not injected).
- handle() does the work (Image::of(), Audio::of(), or calls our custom
VideoGenerationService for Veo), persists media + usage log, pushes
the full attachment shape into the collector, and returns a short
text summary to the LLM.
- schema() exposes typed parameters with enum constraints and doc
strings so the LLM selects valid inputs.
Also fix the Ai::extend closure signature — MultipleInstanceManager
passes (app, config) not just (config) to custom creators.
The agent implements Laravel AI SDK's Agent + Conversational contracts:
- instructions() renders the existing Blade system prompt with workspace
brand context (name, description, website, tone, voice notes, locale)
- messages() reads directly from the AiMessage model scoped to the post,
so our existing conversation persistence stays the single source of
truth — no duplicate SDK-managed storage.
- Assistant messages with attachments are enriched inline with a summary
like [This assistant message attached: 2 image] so the model tracks
progress through carousel-style multi-image generations.
- provider() maps the trypost.ai.text_provider config to Lab::Gemini or
Lab::OpenAI, preserving the existing provider-switching behavior.
AiUsageLogTest and UsageController still passed string 'image'/'video'
arguments to monthlyCount() after the enum migration. Update them to
pass UsageType::Image and UsageType::Video.
