- createClips takes the row of the clipping before looking for clips, so a
timed out attempt and its retry can no longer both store a set
- a clip is claimed (draftedAt off null) before its draft is created and
released when creation fails, so a retry cannot draft it twice
- urls are stripped from the logged processor failure
- Turkish billing label reads correctly after the number
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Turns a YouTube video into captioned vertical clips that land in the media
library and as draft posts. Adds the Clipping / ClippingClip models, the
clipping_minutes credit type, clippingWorkflow + per-clip child workflows,
the /clipping REST routes, the MCP clipping tools and the ui://postiz/clipping
status widget.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Backend traces were sent at 100% (about 77M spans/day) and every
browser session was recorded. Sample /public/v1/analytics/* at 1% and
all other backend traces at 20%, frontend traces at 20% and session
replays at 40%. Error events and error-session replays stay at 100%.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
After the managed-signer migration, clicking Farcaster on the login and
signup pages opened a new tab to the approval link, which only redirects
to the farcaster:// scheme and stays blank on desktop, while the QR was
squeezed into the narrow provider button slot.
The approval view now opens straight in a modal on the login and signup
pages and replaces the intermediate button step in the add/reconnect
channel modal. No tab opens; a Copy Farcaster link button replaces the
Open in Farcaster link. The auth layout mounts the shared modal manager
and the toaster, so errors on those pages are visible. A response
without an approval link (for example the rate limit) now shows the
failure toaster instead of leaving the modal empty, and the signer
creation limit is raised from 10 to 30 per hour per address.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01R8gvBNgPqqayC9zGPB5R5a
Neynar retired Sign In With Neynar, so the hosted popup used for Farcaster
login and channel connect no longer completes. Replace it with Neynar managed
signers: the backend creates a signer, signs the EIP-712 signed key request
with the app custody mnemonic (viem), registers it with Neynar and renders the
approval deep link as a QR (qrcode) since the link is a bare 302 to the
farcaster:// scheme. The frontend polls the new public /auth/farcaster/signer
routes and builds the same base64 blob the popup produced, so authenticate(),
the auth provider, the stored token format and existing users/channels are
unchanged. @neynar/react is removed.
Also thread the oauth_state nonce through the Farcaster login button: since
the Apple login change, checkExists rejects logins without a matching state
cookie, which broke Farcaster login independently of the SIWN retirement.
New env vars: NEYNAR_APP_FID, NEYNAR_APP_MNEMONIC, NEYNAR_SPONSOR_SIGNERS.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01R8gvBNgPqqayC9zGPB5R5a
The "Report a bug" widget asked the reporter to type their email. It was
prefilled from the Sentry scope user but stayed editable and optional, so a
report could arrive with a typo'd, someone else's, or an empty email — and we
could not tie it back to an account.
Setting showEmail: false does not drop the field: the SDK renders it as a
hidden input whose value is the scope user's email (the one we already set from
GET /user/self), and that hidden value is what the form submits. Nothing on our
side has to read or attach the value.
The Name field is left as-is. No user in the production DB has name/lastName —
we never collect it at signup, never pull it from the OAuth profile, and there
is no UI to set it — so hiding it would submit an empty name forever. Keeping
the input lets the reporter tell us who they are; it has no prefill for the
same reason, and an untouched field is no worse than today.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
A schedule-type save targeting a PUBLISHED post silently requeues it:
state back to QUEUE, releaseURL/releaseId nulled, workflow restarted -
and since the workflow's initial sleep is max(0, publishDate - now), a
save that carries (or leaves) a past date publishes again within
seconds. When editing an existing published post the date field holds
its old (past) date, so confirming a reschedule without touching the
date is an accidental instant republish. The existing protection
(2813ed01) is a frontend-only modal: it doesn't cover the public API or
MCP agents at all, and even in the dashboard it never states that
republishing can mean publishing right now.
- changeDate now defaults action to 'update': clients that don't send an
action can no longer requeue a post by accident
- schedule/now saves (createPost) and action 'schedule' (changeDate)
targeting a PUBLISHED post are rejected with a 400 explaining what
happened and how to proceed, unless the new optional republish flag is
sent - the error body is the confirmation dialog for automation
- the dashboard modals send republish: true on explicit confirmation and
now state the consequence (channel, date, recurring note)
- new modal strings added to the locale files via lingo.dev
No behavior change for DRAFT/QUEUE posts.
Testing: 16 service-layer assertions against a stubbed repository + local
Temporal (400 + message for schedule/now saves, rejection before any
write, republish passthrough requeues and restarts the workflow, update
stays date-only, QUEUE posts unaffected), plus a live end-to-end run in
the dashboard: modal confirm -> republish: true -> requeue -> real
publish to an Instagram channel.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Sentry.setUser(null) does not remove custom tags; without this, events
fired after logout but before reload kept the previous org's tag.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Per review on #1800: send the acting user's email alongside the id, on both
backend (auth middleware paths) and frontend (post-login setSentryUser).
Public-API events are unchanged (no user there). Also gives Sentry a proper
avatar/display identity instead of the "?" placeholder.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Sentry events had zero user context, so issues/traces/replays could not be
tied to a customer. Enrich the per-request (backend) and per-tab (frontend)
Sentry scope with the acting identity, so every element emitted afterwards -
errors, spans, profiles, metrics, replays - carries it automatically:
- New setSentryUserContext helper in the shared server sentry lib: sets
Sentry user (id only), organization.id tag, and stripe.customer_id tag -
the latter only when Organization.paymentId starts with "cus_" (legacy
orgs hold non-Stripe values there). Wrapped in try/catch so telemetry can
never break a request; a no-op when Sentry is not initialized (no DSN).
- auth.middleware calls it once req.user/req.org settle, in both the normal
and impersonation branches (the impersonated user is the acting identity).
- public.auth.middleware calls it with org only (no user exists there) -
setUser(null) also guarantees no stale user attaches.
- Frontend: setSentryUser helper next to initializeSentryClient; called from
a useEffect on the /user/self SWR result in the layout, clearing the user
when absent. IDs only - no email/name; paymentId is deliberately not
exposed to the frontend, backend events carry the Stripe id.
Deliberately NOT enriched (no identity exists at emission time):
- Orchestrator/Temporal activities - no HTTP context and no generic hook
without changing workflow/activity signatures, which is forbidden;
publish-time failures therefore still have no org tag.
- Unauthenticated backend routes (login/register/OAuth callbacks), webhook
endpoints (signature-authed, no user), errors thrown inside the auth
middleware itself, and startup/background errors.
- Frontend pre-login pages, errors before /user/self resolves, and Next.js
server/edge events.
Verified e2e: backend events captured locally (envelope capture server)
confirmed user.id + organization.id + stripe.customer_id on authenticated
errors, org-only on public-API errors, no stripe tag for legacy paymentId,
and no cross-request leakage (per-request isolation scopes). Real Sentry
issues then verified in the cloud project for both backend (user, org and
stripe tags) and frontend (user + org, replay attributed to the user id).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
A person can create two accounts with the same email (LOCAL + OAuth)
because User uniqueness is @@unique([email, providerName]). Guard the
subscribe path: if another account with the same email owns an org with
an active subscription (deletedAt: null), block the subscribe flow and
tell them to sign in to that account instead. Excludes the current user
by id, so renewing your own plan is unaffected.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
A production user ticked "Video made with AI" on a TikTok post using the
UPLOAD posting method and the label never appeared: with
content_posting_method=UPLOAD, TikTok's inbox/upload endpoints accept
nothing but the title, so every other setting is silently discarded.
The settings panel still offered all of them, so there was no way to
know the flag would be dropped.
- Frontend: hide all DIRECT_POST-only settings when UPLOAD is selected
(fields stay mounted so their values survive the switch and the DTO
stays valid), and explain in the helper text above the method
selector that the additional settings only apply when posting
directly. Regroup the video-only settings (Allow Duet, Allow Stitch,
Video made with AI) under a "Video features" header and move "Allow
Comments" down next to the settings that apply to both videos and
photos.
- DTO: document each field's DIRECT_POST-only / video-only / photo-only
constraint via @JSONSchema descriptions, surfaced to the agent tools
and the public integration-settings endpoint. Validation unchanged.
- Provider @Rules: tell the agent that UPLOAD drops every setting
except the title and that duet/stitch/video_made_with_ai are
video-only, so it stops promising settings that won't apply.
- Locales: relabel Duet/Stitch/Comments to "Allow ...", add the new
header key, extend the upload helper text (lingo.dev run included).
No change to what is sent to TikTok.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The shortlink question dialog used the generic "No, cancel!" cancel
label, which reads as aborting the whole post instead of declining
shortlinks. Pass a custom cancel label ("No, original URLs") via the
existing deleteDialog cancelButton param.
Adds the no_original_urls key to the English source, translates it to
all 14 lingo.dev locales, and hand-fixes 8 machine translations that
rendered the label as "there are no original URLs" instead of a choice.
Manually tested locally: the dialog now shows "Yes, shortlink it!" /
"No, original URLs", and both choices schedule the post with/without
shortlinked URLs accordingly.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Follow-up to #1666: run lingo.dev to translate
you_have_to_add_x_linkedin_page_threads_or_bluesky in all 14 locales,
replacing the stale you_have_to_add_x_or_linkedin_or_threads entries.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The plugs page empty state said to add 'X or LinkedIn or Threads', but
plugs only exist for LinkedIn Page (not personal LinkedIn), and Bluesky
was missing. Renamed the translation key so non-English locales fall
back to the corrected English text instead of stale translations.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Add a controlled multi-select (scrollable checkbox list) to the shared
react form components, alongside the existing Select. It is purely
presentational - options in, full selected array out via onChange - so
the caller owns selection state (pair it with react-hook-form setValue).
Matches Select's conventions: TranslatedLabel, useFormContext error
display, and the same newBgColorInner / newTableBorder design tokens.
The app had no general multi-select primitive (only single Select and
the npm react-tag-autocomplete based *.tags.tsx inputs), so this fills
that gap for any provider/form that needs to pick several known options.
Testing: exercised end-to-end in the next commit in this PR, which wires
this primitive into the WordPress categories/tags selectors. Selecting
multiple options there persisted correctly and posted the chosen term
IDs to WordPress.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>