houseassassin/twenty
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
13,516 Commits 1 Branch 0 Tags
6897fff63239f0fdf0c8caf2f8f2b481a75c00a4
Commit Graph

11201 Commits

Author SHA1 Message Date
Félix Malfait 6897fff632 Rebuild email composer recipient fields as a structured chip input with person resolution and autocomplete (#22668) 
# Why

The To/Cc/Bcc fields reused `FormMultiTextFieldInput`, the workflow
Tiptap tag editor, with recipients stored as a comma-separated string.
That caused every reported issue: duplicates were allowed, the field was
locked to one 32px line with a hidden horizontal scrollbar, chips did
nothing on click, `First Last <email>` could not even be typed (space
committed a tag) and was rejected by the backend when pasted, chips
could not be edited, invalid addresses only failed server-side after
pressing Send, and there was no autocomplete at all.

## The model

A recipient is `{ address, displayName? }`. Person and workspace member
are never stored in composer state; they are resolved live from the
address at render time, mirroring how `MatchParticipantService` links
`messageParticipant.handle` to `personId`/`workspaceMemberId` on the
receive side. Entities appear at the edges (autocomplete in, chip
display out); state, dedupe, validation, and send operate on addresses
only. The send path is unchanged: `SendEmailInput.to/cc/bcc` stay
comma-separated bare addresses.

# What changed

New module `activities/emails/recipients/` (the workflow editor is
untouched; its other consumers are unaffected):

- **`EmailRecipientsFieldInput`**: wrapping chip rows (up to ~3 lines,
then scroll), commit on Enter/Tab/comma/semicolon/blur, space commits
only when the buffer is already a valid email, paste parses RFC 5322
lists (names, quoted commas, semicolons, newlines), case-insensitive
dedupe with a flash on the existing chip, invalid addresses become red
chips that disable Send, double-click or keyboard editing in place with
Escape revert, Backspace select-then-delete, arrow-key chip navigation,
Ctrl/Cmd+Enter commits a pending buffer or sends when the buffer is
empty.
- **Person resolution**: chips resolve against People
(`emails.primaryEmail`, case-insensitive) and workspace members,
rendering avatar + name when known and degrading to a plain address chip
otherwise.
- **Chip menu**: person/member header, Copy email, Edit, Remove, and Add
as person for unknown addresses (creates the Person; the chip upgrades
in place).
- **Autocomplete**: blends context people (company you are composing
from, or the company behind a person/opportunity), ranked people search,
workspace members with a Team member badge, and a literal "Use this
email" row ranked first when the typed buffer is a valid address.
Suggestions exclude addresses already present in any field. Enter picks
the highlighted or top row.
- **Prefill**: replies and drafts preserve participant display names
(`getEmailDraftPrefillFromMessage`, `useReplyContext`).
- `useEmailComposerState` holds `EmailRecipient[]` per field and blocks
send on invalid recipients; the recipient-limit warning is surfaced
again in the composer.
- The Send Email engine command passes the record context so context
suggestions work from the record page action.
- `EmailsFilter` was missing from the shared `LeafFilter` union, so
nothing could filter on `emails.primaryEmail`; added (additive).
- New dependency `addressparser@1.0.1` in twenty-front, the same package
and version the server already uses to parse inbound mail headers, so
both sides parse identically. Tiny, dependency-free, browser-safe.

# Decisions and tradeoffs

- Person resolution matches on `emails.primaryEmail` only,
case-insensitively via per-address `ilike` filters (no `%` wildcards,
`%_\` escaped). `additionalEmails` is a JSONB array and not cleanly
filterable through the GraphQL filter API today; the server-side matcher
checks additional emails too, so a chip may show as a plain address even
though the send still links to the person via participant matching.
- Chip flash-on-duplicate replays its CSS animation by remounting the
chip subtree (nonce in the React key), chosen over animation-restart
hacks; the remount is invisible.
- Keyboard chip selection keeps DOM focus on the input and tracks a
virtual `selectedChipIndex` (`aria-activedescendant`) instead of roving
focus across chips: one focus point, no focus juggling, standard
combobox listbox pattern.
- `flushSync` (precedent: `Dropdown.tsx`) focuses and places the caret
after entering chip-edit mode; the alternative was a useEffect on
editing state.
- Suggestion rows `preventDefault` on mousedown so picking a suggestion
never blurs the input (blur would first commit the half-typed buffer as
a junk chip).
- Cmd/Ctrl+Enter inside a recipient field: with a non-empty buffer it
commits the buffer only; with an empty buffer it sends via an `onSubmit`
prop wired to `handleSend`. Not commit+send in one stroke: `handleSend`
holds a same-render closure over composer state, so sending in the same
event would read the pre-commit recipients. E2E also showed the side
panel's own ctrl+Enter hotkey never fires while any form field is
focused (focus-stack scoping, applies to the old composer too), which is
why the field triggers the submit itself.
- Enter with suggestions open picks the highlighted (or top) suggestion,
Gmail-style. When the typed buffer is itself a valid email, the literal
row is ranked first so Enter keeps meaning "add what I typed".
- Suggestions are disabled while editing a chip (the edit buffer holds
`Name <email>` text, a poor search query).
- Dedupe blocks within a field; across fields typed duplicates are
allowed (sometimes intentional), but suggestions exclude addresses
already present in any of To/Cc/Bcc.
- Chip menu actions never navigate: navigating the side panel (or main
view) unmounts the composer and silently destroys the draft, since
composer state is component-local with no draft persistence. "Add as
person" creates the record and shows a snackbar while the chip upgrades
in place; the person header row is informational. "Open person"
navigation should come back once drafts survive navigation.
- The reply composer gets no context record: its widget target record is
the message thread, not a person/company, and replies already prefill
participants.
- If two people share a primary email, the last fetched match wins for
chip display (no ambiguity UI).
- "Add as person" splits the display name on the first space for
firstName/lastName, the same heuristic the contact-creation manager uses
server-side.

# Deferred

- Display names on the wire (`Name <email>` in outbound headers): needs
`SendEmailInput` / `EmailComposerService.validateEmails` changes
server-side.
- Drag chips between To/Cc/Bcc; collapse-on-blur to one line with a "+N
others" summary.
- Frequency/recency ranking of suggestions from `messageParticipant`
aggregates.
- "Open person" from the chip menu, pending draft persistence across
navigation.

# Verification

Unit tests cover the parser, formatter round-trip, merge/dedupe, and the
field state machine (commit, dedupe flash, edit, cancel, keyboard
selection). Typecheck, lint, and the email module suites pass, plus the
shared and side-panel suites.

Every flow was also driven end to end with Playwright against seeded
data: prefill resolution, context and typed suggestions, keyboard
navigation and picks, dedupe flash, RFC 5322 paste, invalid chips gating
Send, wrapping, in-place editing, chip menus, clipboard copy, Add as
person with live chip upgrade, Cc/Bcc exclusions, and the Ctrl+Enter
send path (the mutation reached the server; it failed only on the seeded
account's missing refresh token, expected outside a real provider
connection).

Screenshots of each verified behavior:
https://claude.ai/code/artifact/1743f05d-422e-43d0-bbea-a34a0470c180

---
_Generated by [Claude
Code](https://claude.ai/code/session_0199wDARiw48GqVTpgWzbXWw)_

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22668?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->
 2026-07-09 13:11:09 +02:00
Félix Malfait 3dd7dfde5a feat(billing): collect credit card in-app when starting subscription from billing page (#22706) 
## Context

On the billing page, a trialing workspace without a payment method that
clicks "Increase" (credits section) or "Subscribe Now" (subscription
card) goes through the end-trial confirmation and then gets redirected
to the Stripe billing portal to add a card. We already have an in-app
card collection flow (`AddCreditCardModal`, Stripe PaymentElement on a
SetupIntent) used by the AI chat credits banner and the global end-trial
information banner. The billing page was the last surface still bouncing
users to the portal.

## What this does

Applies the same pattern as `AIChatNoMoreBillingCreditsBanner` and
`InformationBannerEndTrialPeriod` to the billing page:

- When the workspace is trialing and `billingHasPaymentMethod ===
false`, the `endTrialPeriod` modal now renders `AddCreditCardModal`
(in-app card entry) instead of the confirmation modal. Once the card is
added, the subscription starts via `endTrialPeriod({
skipPaymentMethodRedirect: true })`, and the user stays on the billing
page.
- When a payment method exists (or the flag is unknown), the existing
"Start Your Subscription" confirmation modal is unchanged, including its
portal-redirect fallback if the backend reports no card.
- Both entry points share the same modal instance
(`BILLING_MODAL_IDS.endTrialPeriod`), so the "Increase" button in the
credits section and the "Subscribe Now" header action both get the
in-app flow. Button labels are unchanged: they state the outcome, and
the modal explains the card step.
- `AddCreditCardModal` now closes only after `onPaymentMethodAdded`
resolves, so the form keeps a visible loading state until the
subscription is confirmed instead of closing instantly while activation
runs in the background. This also benefits the AI chat and information
banner flows that share the component.

## What stays the same

- No backend changes. `endSubscriptionTrialPeriod` still verifies the
payment method live against Stripe and calls
`ensureDefaultPaymentMethod` before activating, so the card added
through the SetupIntent becomes the default charged card.
- 3DS or redirect-based payment methods return to the billing page with
the `start-subscription-after-payment-method` param, which the globally
mounted `EndTrialAfterPaymentMethodGater` already handles.
- The trial banner ("Trial ends X, please add card details") keeps its
portal flow on purpose: its semantic is add a card while keeping the
trial, whereas `AddCreditCardModal` starts the subscription immediately.
Converting it would need a variant of the form that skips the auto-start
param. Same for the past-due "Update payment" path. Both are possible
follow-ups.
- All strings reuse existing Lingui msgids, no catalog changes needed.

## Testing

- `oxlint`, `oxfmt` and `nx typecheck twenty-front` pass.
- All 17 `settings/billing` jest suites pass.
- Could not drive the Stripe flow end-to-end in this environment (no
Stripe test keys); the composition is identical to the two surfaces
already shipped, and the state matrix (`hasPaymentMethod` false / true /
unknown, trialing / not trialing, permission gating) was traced through
both entry points.
 2026-07-09 12:53:29 +02:00
github-actions[bot] bd26eb2d99 i18n - translations (#22716) 
Created by Github action

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22716?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->

Co-authored-by: github-actions <github-actions@twenty.com>
 2026-07-09 11:53:00 +02:00
github-actions[bot] ebee7d71b9 i18n - docs translations (#22715) 
Created by Github action

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22715?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->

Co-authored-by: github-actions <github-actions@twenty.com>
 2026-07-09 11:51:54 +02:00
github-actions[bot] a0cf4cc9e1 i18n - translations (#22714) 
Created by Github action

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22714?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->

---------

Co-authored-by: github-actions <github-actions@twenty.com>
 2026-07-09 11:46:19 +02:00
martmull dcccdbd148 Fix flaky "read EINVAL" in integration tests: bump msw to 2.12.14 (#22702) 
# Context

Part of a CI flakiness sweep. Integration shards are intermittently
killed by an uncaught socket error that poisons a whole spec file (e.g.
`sync-failure-lifecycle.integration-spec.ts`, 4 tests, on unrelated PR
branches — example run 28940565117, shard 8):

```
Error: read EINVAL
    at MockHttpSocket.emit (@mswjs/interceptors/.../MockHttpSocket.ts:161)
```

# Root cause

The integration setup routes **all** HTTP (including
supertest/node-fetch calls to the in-process app) through msw's
ClientRequest interceptor, with localhost passthrough. msw `2.12.7` pins
`@mswjs/interceptors` `0.40.0`, where:

- `passthrough()` aliases the real socket's libuv `_handle` onto the
mock socket (two owners of one handle) and forwards the real socket's
`error` events into `MockHttpSocket.emit`,
- `destroy()` never destroys the passthrough socket, so it stays
orphaned with its error-forwarding listener attached.

When the server side closes such a connection later, a read on the stale
shared handle yields `EINVAL`, forwarded into `emit()` with no request
(and no error listener) attached → uncaught exception → jest kills the
in-flight file. This is upstream
[mswjs/interceptors#753](https://github.com/mswjs/interceptors/issues/753);
the stack frames match line-for-line.

# Fix

Bump `msw` to **exactly `2.12.14`** in `twenty-server` and
`twenty-front` (shared lock entry). msw 2.12.9+ requires interceptors
`^0.41.2` → resolves 0.41.9, which ships
[interceptors#755](https://github.com/mswjs/interceptors/pull/755):
passthrough sockets get their listeners removed and are destroyed on
close, severing the exact error path above.

Notes from adversarial review:
- Pinned exact (`2.12.14`, not `^`) because the caret would resolve to
2.15.0 today — a bigger jump than reviewed.
- Compatibility checked: no deep imports of msw/interceptors anywhere;
interceptors 0.41.x externals already covered by
`transformIgnorePatterns`; `msw-storybook-addon@2.0.6` peer range
satisfied; `mockServiceWorker.js` integrity checksum unchanged between
2.12.7 and 2.12.14; 0.41.7+ adds Node 24 compatibility (repo targets
Node 24).
- Residual risk disclosed: #753 is still open upstream and the `_handle`
aliasing remains in 0.41.9 — this removes the observed orphaned-socket
path, but keep an eye out for recurrence.

Dependency-only change: 2 package.json lines + lockfile.

---
_Generated by [Claude
Code](https://claude.ai/code/session_01AtD2wWm3EthV6t3Hs31QyB)_

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22702?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->
 2026-07-09 10:58:26 +02:00
martmull eb9cc1862a Fix flaky e2e tests: CI-realistic timeouts + retry-safe kanban field label (#22701) 
# Context

Part of a CI flakiness sweep. `ci-e2e-main` fails on ~8% of main pushes
(15 failing runs on 2026-07-08 alone), always the same three tests,
always on 5-second `expect` timeouts:

- `onboarding.spec.ts` — `Create your workspace` heading after sign-up
(sign-up mutation → loadCurrentUser → workspace-creation-defaults query,
three sequential round trips)
- `signup_invite_email.spec.ts` — `Create profile` (invite sign-up +
token exchange + full metadata load)
- `create-kanban-view.spec.ts` — `No Value` kanban column (view +
viewFields + viewGroups persistence, no-value group settles last)

The runner hosts the dev-mode NestJS server (`nest start --watch`), the
worker and Chromium simultaneously, so 5s is structurally too tight;
neighboring steps in the same specs already use 30-90s timeouts.

# Fix

- `playwright.config.ts`: `expect.timeout` 5s → 15s and test timeout 30s
→ 60s **on CI only** (the config already branches on `process.env.CI`
for retries/reporter). These are web-first auto-retrying assertions, so
green runs are not slowed — only genuinely failing assertions wait
longer. The 60s test budget also fixes an existing inconsistency: the
kanban spec has a 30s per-assertion timeout inside a 30s test budget.
- `create-kanban-view.spec.ts`: use a per-run unique label for the
Industry select field. The spec is `test.describe.serial`, and
Playwright retries re-run the whole group against the same database (no
reset between in-run retries). The already-created `Industry` field made
the label-uniqueness validation fail permanently, so Save stayed
disabled and **every retry of this group failed deterministically**
("element is not enabled" after 30s) — retries were dead weight for this
spec.

Adversarially reviewed: the alternative (per-assertion timeouts) is the
whack-a-mole pattern already attempted once (`Food` has a 30s patch);
`waitForResponse` on operation names would be more lines and more
brittle.

Test-infra only, 2 files, +11/-6.

Note for the team (out of scope here): `ci-e2e-main.yaml` builds the
server and then discards it — `nx start twenty-server` runs `rimraf dist
&& NODE_ENV=development nest start --watch`. Running the built server
would cut latency and runner load across the whole suite.

---
_Generated by [Claude
Code](https://claude.ai/code/session_01AtD2wWm3EthV6t3Hs31QyB)_

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22701?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->
 2026-07-09 10:58:16 +02:00
martmull 67fa0cc93c Fix flaky webhook delivery integration test (fixed sleep -> poll) (#22699) 
# Context

Part of a CI flakiness sweep. `webhooks.integration-spec.ts` › "should
deliver webhook successfully when safe mode is disabled" intermittently
fails with `expect(receiver.receivedPayloads.length).toBe(1) ...
Received: 0` on unrelated PRs (example: run 28959576750, shard 3).

# Root cause

The test asserted delivery after a fixed 100ms sleep. Delivery actually
crosses: a fire-and-forget `EventEmitter2.emit` (the GraphQL response
returns before the job is even enqueued) → BullMQ hop 1
(`CallWebhookJobsJob`, which also recomputes the just-invalidated
`flatWebhookMaps` cache) → BullMQ hop 2 (`CallWebhookJob`) → HTTP POST
to the in-test receiver. Two Redis round trips plus a cache rebuild
routinely exceed 100ms on loaded CI runners. The global
`waitForAllJobsToFinish` only runs in `afterEach`, after the assertion.

# Fix

- Poll the receiver with the existing `expectEventually` helper (30s
deadline, 100ms interval) instead of sleeping, and give the test an
explicit 60s timeout (suite default is 20s). Worst case the test fails
slower; it can no longer fail while delivery is merely in flight.
- Bonus bug found during adversarial review of this fix: the `finally`
cleanup deleted config key `HTTP_TOOL_SAFE_MODE_ENABLED` while the test
creates `OUTBOUND_HTTP_SAFE_MODE_ENABLED`, silently leaving outbound
safe mode disabled in the DB for every suite that runs after this one.
Fixed the key.

Duplicate-delivery risk was checked: `CallWebhookJob.handle` never
throws (errors swallowed), so `retryLimit: 3` can't produce a second
payload that would break `toBe(1)`.

Test-only change, 1 file, +15/-9.

---
_Generated by [Claude
Code](https://claude.ai/code/session_01AtD2wWm3EthV6t3Hs31QyB)_

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22699?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->
 2026-07-09 10:58:07 +02:00
Raphaël Bosi e38d587602 Welcome animation when a workspace is set up (#22622) 
## Desktop


https://github.com/user-attachments/assets/87999ff5-9257-4fed-90e0-781322b1db98

## Mobile



https://github.com/user-attachments/assets/0d115cb6-7bca-4229-8fde-0f188e862b00


Greets the user the moment they finish onboarding: a halftone Twenty
mark (brand purple) assembles from scattered particles, shimmers, then
bursts outward to reveal the freshly loaded workspace behind it. Plays
once, ~3s, with a reduced-motion fallback.

`PageChangeEffect` sets a transient `isWelcomeAnimationVisibleState`
atom at the onboarding to workspace redirect seam (gated by
`shouldShowWelcomeAnimationOnNavigate`); `WelcomeOverlay`, mounted in
`WorkspaceAppProviders`, plays it and clears it. Transient by design, so
it never replays on reload.

## Why a canvas rendered on another thread

The overlay plays at the exact moment the workspace mounts behind it,
which is one of the heaviest main-thread stretches in the app (metadata,
providers, first data fetch). Anything animating on the main thread
competes with that work:

- framer-motion and plain main-thread `requestAnimationFrame` visibly
stuttered and froze during the mount, because long tasks starve rAF.
- CSS keyframes stay smooth (they run on the compositor), but can't
drive a ~2900-particle halftone with per-particle physics (staggered
assemble, moving shimmer band, directional burst).

So the halftone runs as a Canvas 2D particle system inside a Web Worker,
drawing to an `OffscreenCanvas` transferred from the main thread. The
whole render loop lives off the main thread, so app-mount jank can't
reach it. A main-thread path is kept as a fallback for browsers without
`OffscreenCanvas` / `transferControlToOffscreen`, and Safari workers
(which lack rAF) fall back to a fixed-interval loop. The renderer is a
pure, DOM-free module so it bundles cleanly into the worker; dot colors
come from CSS vars read on the main thread and passed in.

Also fires for every onboarding completion path and skips the billing
`PlanRequired` detour.

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22622?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->
 2026-07-09 08:47:07 +00:00
martmull e0bd4ab732 docs(apps): make the workspace functions URL the primary route-serving story (#22693) 
Part 5 of the app-docs audit series (after #22688–#22691).

## The problem

`front-components.mdx` warns that the legacy `/s/` route is deprecated
and **deactivates on 2026-07-24** (16 days from now), but the rest of
the docs still teach `/s/` as the only serving path:
`logic-functions.mdx` ("Exposes your function ... under the `/s/`
endpoint"), `logic/overview.mdx` ("A request hits your `/s/<path>`
endpoint"), and the document-generator tutorial fetches
`${TWENTY_API_URL}/s/...` from front-component code. A developer
following those pages today ships an app that breaks on Cloud in two
weeks.

## What this changes

- **logic/logic-functions.mdx** — httpRoute triggers are described as
served at the workspace's functions base URL (what the server injects as
`TWENTY_FUNCTIONS_URL`; a dedicated per-workspace domain on Cloud, per
`WorkspaceDomainsService.buildPublicFunctionBaseUrl`), with a warning
box covering the `/s/` deprecation and the self-host fallback.
- **logic/overview.mdx** — trigger table no longer hardcodes
`/s/<path>`.
- **document-generator tutorial** — the `curl
http://localhost:2020/s/...` examples stay (they're correct against the
local dev image, where no isolated functions domain exists), with a note
explaining the Cloud behavior. The front-component code snippets now use
the `TWENTY_FUNCTIONS_URL || TWENTY_API_URL + '/s'` fallback pattern —
the same one Twenty's own published apps use (e.g.
`packages/twenty-apps/public/call-recorder`).

Only English sources were touched; `l/<locale>` copies come from
Crowdin.

---
_Generated by [Claude
Code](https://claude.ai/code/session_01ExboyDAT19khDuKXaYXETT)_

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22693?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->

---------

Co-authored-by: Martin <martin@twenty.com>
 2026-07-09 10:41:12 +02:00
martmull 9e3379cf2d fix(create-twenty-app): working docs URL in final message, complete dev:add table in AGENTS.md (#22695) 
Part 7 of the app-docs audit series — two small fixes in the scaffolder
itself, found while testing it end-to-end.

- The `Documentation:` link printed at the end of `npx
create-twenty-app` points to
`https://docs.twenty.com/developers/extend/capabilities/apps`, which
404s (the apps docs live under `/developers/extend/apps/...`). Now
points at the quick start.
- The generated `AGENTS.md` (mirrored to `CLAUDE.md`) recommends `yarn
twenty dev:add` but its entity table lists only 10 of the 14 supported
entity types — AI agents working from that file can't discover
`pageLayoutTab`, `commandMenuItem`, `viewField`, or
`connectionProvider`. Added the missing rows (paths follow the CLI's
`kebabCase(entity)s` convention).

---
_Generated by [Claude
Code](https://claude.ai/code/session_01ExboyDAT19khDuKXaYXETT)_

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22695?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->

Co-authored-by: Martin <martin@twenty.com>
 2026-07-09 10:38:16 +02:00
martmull 51f8b590ef docs(apps): rewrite install hooks page — real client API, one explanation per concept (#22694) 
Part 6 of the app-docs audit series (after #22688–#22693). This is the
"too verbose / duplicated" pass on the worst offender, plus one accuracy
fix that came out of it.

## Accuracy

The seeding and backup examples imported `createClient` from
`./generated/client` and called an ORM-style API
(`client.postCard.create({ data })`, `client.postCard.findMany({ where:
... })`, `client.postCard.update({ where, data })`). That API doesn't
exist anywhere in the SDK or generated clients — the real pattern is
`new CoreApiClient()` with genql-style `query`/`mutation` calls, as used
by the actual post-install hook in
`packages/twenty-apps/examples/postcard`. Both examples are rewritten
accordingly.

## Verbosity

The pre-install vs post-install distinction was explained four separate
times (intro, inside each accordion's "Key points", a dedicated
comparison accordion, and a rule-of-thumb table), and the shared
behavior (InstallPayload shape, one-per-app limit, manifest attachment,
env vars, dev-mode skip, 300s timeout) was duplicated across both
accordions. The page now has:

- one **at-a-glance comparison table** + the rule-of-thumb table up
front,
- one **shared-behavior list** stated once,
- per-hook accordions that carry only what's unique to each hook
(execution model detail, the pared-down pre-sync, one corrected example
each).

Net: −128/+66 lines with no unique fact removed.

Also stops `operations/publishing.mdx` from enumerating the marketplace
metadata field list a second time in the discovery section.

Only English sources were touched; `l/<locale>` copies come from
Crowdin.

---
_Generated by [Claude
Code](https://claude.ai/code/session_01ExboyDAT19khDuKXaYXETT)_

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22694?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->

Co-authored-by: Martin <martin@twenty.com>
 2026-07-09 10:36:06 +02:00
martmull 3a9f405e6c docs(apps): document missing enum values and complete entity references (#22691) 
Part 4 of the app-docs audit series (after #22688, #22689, #22690).
Focus: values that exist in the SDK but never made it into the docs. All
value lists were extracted from `twenty-shared` / `twenty-sdk` source.

## What this adds/fixes

**data/objects.mdx**
- New "Field types" section with the complete `FieldType` value set (24
values, grouped by category, with the composite/`SELECT` caveats and the
lowercase `universalSettings.dataType` values for `NUMBER`). Previously
no page listed the available field types — readers had to
reverse-engineer them from scattered examples.

**layout/views.mdx**
- `ViewFilterOperand` was imported from `twenty-shared/types` in the
example; it's re-exported from `twenty-sdk/define`, which is the
supported import surface for apps.
- New "Optional properties" table covering what the page omitted: `type`
(`ViewType.TABLE`/`KANBAN`/`CALENDAR`), `visibility`, `openRecordIn`,
`sorts`, kanban aggregate settings, and calendar settings.

**getting-started/scaffolding.mdx**
- The `dev:add` table listed 10 of 14 entity types; added
`pageLayoutTab`, `commandMenuItem`, `viewField`, and
`connectionProvider` (paths follow the CLI's kebab-case convention).

**layout/navigation-menu-items.mdx**
- Note about `NavigationMenuItemType.RECORD`: it exists in the enum but
is internal (user favorites) and has no manifest field to reference a
record, so apps can't use it — documented to prevent confusion about the
"missing" value.

Only English sources were touched; `l/<locale>` copies come from
Crowdin.

---
_Generated by [Claude
Code](https://claude.ai/code/session_01ExboyDAT19khDuKXaYXETT)_

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22691?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->

Co-authored-by: Martin <martin@twenty.com>
 2026-07-09 10:29:22 +02:00
martmull 0970f85cd2 docs(apps): align project structure and testing pages with the actual scaffold (#22690) 
Part 3 of the app-docs audit series (after #22688 and #22689). Verified
by scaffolding a fresh app with `create-twenty-app` and diffing the docs
against the generated files and the template in
`packages/create-twenty-app/src/constants/template`.

## What this fixes

**getting-started/project-structure.mdx**
- The documented tree was missing most of what the scaffolder actually
generates: the starter welcome page (`front-components/`,
`navigation-menu-items/`, `page-layouts/`), the real test files
(`global-setup.ts`, `application-config.test.ts`,
`schema.integration-test.ts` — not `setup-test.ts` /
`app-install.integration-test.ts`), `cd.yml`, `vitest.unit.config.ts`,
and `AGENTS.md`/`CLAUDE.md` (the docs said `LLMS.md`, which isn't
generated).
- Dependency snippet showed `^2.13.0`; the scaffolder pins its own
version (currently 2.20.0) and also adds `twenty-ui`.
- `twenty build` → `twenty dev:build`.

**operations/testing.mdx**
- The Vitest setup section described a config that diverges from the
scaffold (uses `setupFiles` instead of `globalSetup`, writes the SDK
config to `os.tmpdir()/.twenty-sdk-test/config.json` — a path the CLI
never reads). Replaced with the actual pattern: `globalSetup` +
`~/.twenty/config.test.json` (what the CLI reads under `NODE_ENV=test`)
+ `appDevOnce` sync and uninstall-teardown.
- The CI section described a `spawn-twenty-docker-image` action and a
4-step workflow; the scaffolded `ci.yml` uses
`spawn-twenty-app-dev-test` and also runs lint, typecheck, and unit
tests. This section previously contradicted `operations/publishing.mdx`
— it now gives a short accurate summary and links to Publishing for the
full walkthrough of both workflows (de-duplicating the two pages).
- Added `appDevOnce` to the programmatic API table (used by the
scaffolded global setup).

**getting-started/troubleshooting.mdx**
- Node requirement made precise (`^24.5.0`), `twenty build` → `twenty
dev:build`.

Only English sources were touched; `l/<locale>` copies come from
Crowdin.

---
_Generated by [Claude
Code](https://claude.ai/code/session_01ExboyDAT19khDuKXaYXETT)_

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22690?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->

Co-authored-by: Martin <martin@twenty.com>
 2026-07-09 10:28:34 +02:00
martmull 2781a06025 docs(apps): fix nonexistent SDK import paths and unsupported config in layout pages (#22689) 
Part 2 of the app-docs audit series (after #22688). Every fix below was
verified against the `twenty-sdk` source and its `exports` map.

## What this fixes

**Broken import paths (copy-paste would not compile)**
- `twenty-sdk/command` and `twenty-sdk/clients` are not export subpaths
of `twenty-sdk` — 9 code samples across `front-components.mdx` and
`command-menu-items.mdx` used them. `Command`, `CommandModal`,
`CommandLink`, `CommandOpenSidePanelPage` actually live in
`twenty-sdk/front-component`, and `CoreApiClient` in
`twenty-client-sdk/core` (matching every example app in
`packages/twenty-apps`).

**Unsupported config**
- The bulk-export example passed an inline `command: {...}` to
`defineFrontComponent`, but `FrontComponentConfig` has no such property.
Replaced with a separate `defineCommandMenuItem` file, which is the
supported pairing.

**Deprecated API in examples**
- Three examples used `useRecordId()` even though the hooks table on the
same page marks it deprecated. Switched them to
`useSelectedRecordIds()`.
- `defineCommandMenuItem`'s `icon` is deprecated (the build warns "icon
will be ignored in favor of application icon") but the docs listed it as
a normal field and used it in examples. Marked it deprecated in the
table and removed it from examples.

**Missing enum value**
- `availabilityType` supports `'GLOBAL_OBJECT_CONTEXT'`
(`CommandMenuItemManifest` in `twenty-shared`), which the config table
omitted.

**Deduplication**
- The full run-action example (component + command, ~40 lines) appeared
verbatim on both layout pages. `command-menu-items.mdx` now keeps only
the command snippet and links to the component example on the Front
Components page.

Only English sources were touched; `l/<locale>` copies come from
Crowdin.

---
_Generated by [Claude
Code](https://claude.ai/code/session_01ExboyDAT19khDuKXaYXETT)_

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22689?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->

Co-authored-by: Martin <martin@twenty.com>
 2026-07-09 10:26:12 +02:00
martmull 5c1e7dd559 docs(apps): fix stale CLI commands in getting-started and operations pages (#22688) 
Part 1 of a series of small PRs from a full audit of the app-development
docs (every claim was cross-checked against `twenty-sdk`,
`create-twenty-app`, and a scaffolded app).

## What this fixes

**quick-start.mdx**
- `yarn twenty server` does not exist — replaced with `yarn twenty
docker:start` (the command every other page uses, and what the CLI
actually ships).
- The scaffolder is non-interactive since create-twenty-app 2.x: there
is no "name and description" prompt and no "Would you like to set up a
local Twenty instance?" prompt (that screenshot was removed). It
auto-starts the local Docker server and authenticates with the
pre-seeded dev key; OAuth (browser sign-in + Authorize) only happens for
remote `--url` targets or `--authentication-method oauth`.
- `--debounceMs` default is `1000`, not `2000` (see `twenty dev
--help`).
- The one-shot section now teaches `twenty plan` / `twenty apply`; `dev
--once` / `--dry-run` are marked as the deprecated aliases they are in
the CLI help.
- Node prerequisite tightened to 24.5+ to match `engines.node: ^24.5.0`.

**operations/sync-and-recovery.mdx**
- Command matrix, previewing section, and recovery ladder switched from
the deprecated `dev --once [--dry-run]` to `plan` / `apply` (heading
anchor updated accordingly).

**operations/cli.mdx**
- Added a complete command overview table (the page previously omitted
`plan`, `apply`, `dev:translations-extract`, `dev:catalog-sync`, and the
whole `docker:*` group without pointing anywhere).
- Added `remote:status` and `remote:remove`, and the `--preInstall` exec
flag.

**tutorials/document-generator/publishing.mdx**
- Pre-publish check now uses `yarn twenty plan`.

Only English sources were touched; `l/<locale>` copies come from
Crowdin.

---
_Generated by [Claude
Code](https://claude.ai/code/session_01ExboyDAT19khDuKXaYXETT)_

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22688?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->

---------

Co-authored-by: Martin <martin@twenty.com>
Co-authored-by: Weiko <corentin@twenty.com>
 2026-07-09 10:25:09 +02:00
Marie cdabef6429 fix(sdk): report the connected server's real version in dev version check (#22670) 
## Summary

The `yarn twenty app dev` version row was reading the "local server"
version from the `twenty-app-dev` Docker container's baked-in
`APP_VERSION` env var. When the CLI is actually pointed at a separately
running instance (or a stale `twenty-app-dev` container is lying
around), this reported a version unrelated to the server serving
requests — e.g. showing `Server v2.5.3` and a bogus "days behind"
warning while the real instance was on `2.16.1`.

This changes the version resolution to ask the server the CLI is
connected to for its real version over HTTP, falling back to Docker
inspection only when the server can't be reached.

- Add `getServerVersionFromApi`, which reads the version from the
public, no-auth `/.well-known/mcp/server-card.json` endpoint (its
`version` is the server's `APP_VERSION`). Returns `null` gracefully on
timeout, non-OK responses, or missing/`0.0.0`/non-semver values.
- `getVersionInfo` now uses `getServerVersionFromApi() ??
getLocalServerVersion(containerName)`, running the API call in parallel
with the Docker Hub published-versions fetch. All downstream logic
(`isMinorOrMajorBehind`, `daysBehind`, the dev UI row, and the headless
warning) now operates on the real version.
 2026-07-09 10:24:35 +02:00
Etienne a51c37dae5 feat(ai) - add light AI chat turn instrumentation (metrics + Sentry correlation) (#22692) 
## Summary

Adds minimal server-side observability for AI chat turns: lifecycle
counters to measure success/failure rates, Sentry scope tags to
correlate API and worker traces, and per-LLM-call telemetry metadata for
turn/stream correlation.

- Add turn lifecycle metrics: `ai-chat/turn-started`,
`ai-chat/turn-completed`, `ai-chat/turn-failed` (with `failure_phase`
and `error_code` attributes)
- Emit counters at key points: job start, clean completion, execution
failures, enqueue failures, interrupted streams, and empty completions
- Tag Sentry scope with `streamId`, `turnId`, `threadId`, and
`workspaceId` at API entry points (`sendChatMessage`,
`retryChatMessage`, `answerAgentChatQuestion`) and worker entry
(`StreamAgentChatJob`)
- Enrich LLM `experimental_telemetry` metadata with
stream/turn/thread/workspace IDs
- Return `turnId` from streaming service methods so resolvers can tag
the scope
- Remove granular tool-learned/skill-loaded metrics in favor of the
turn-level counters

## Test plan

- [ ] Send a chat message and verify `ai-chat/turn-started` and
`ai-chat/turn-completed` increment
- [ ] Trigger a stream failure (e.g. interrupted/dead stream) and verify
`ai-chat/turn-failed` with correct `failure_phase`
- [ ] Retry a failed turn and confirm a new `turn-started` is emitted
for the retry attempt
- [ ] Answer an `ask_questions` prompt and confirm Sentry tags include
`streamId` and `turnId`
- [ ] Check Sentry spans for LLM calls include `streamId`, `turnId`,
`threadId`, `workspaceId` in telemetry metadata
- [ ] Run unit tests:
- `npx jest
packages/twenty-server/src/engine/metadata-modules/ai/ai-chat/jobs/__tests__/stream-agent-chat.job.spec.ts`
- `npx jest
packages/twenty-server/src/engine/metadata-modules/ai/ai-chat/services/__tests__/agent-chat-streaming.service.claim.spec.ts`
- `npx jest
packages/twenty-server/src/engine/metadata-modules/ai/ai-chat/services/__tests__/agent-chat-streaming.service.retry.spec.ts`

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22692?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->
 2026-07-09 08:24:27 +00:00
Etienne f2d0a1b90a fix(billing): only reactivate suspended workspaces when subscription is in good standing (#22687) 
## Context

When a subscription's trial ends without a valid payment method, Stripe
emits
`customer.subscription.updated` (`active → past_due`) within seconds of
`trial_end`.
Our webhook handler correctly suspends the workspace for this event,
because it lands
inside the 24h "trial just ended" window checked by
`shouldSuspendWorkspace`.

However, the workspace does not *stay* suspended if an other
subscription update event arrived.

## Problem

Reactivation was gated only on the negation of the suspend heuristic:

```ts
} else if (workspace.activationStatus === WorkspaceActivationStatus.SUSPENDED) {
  await this.workspaceService.reactivateWorkspace(workspaceId);
}

<!-- This is an auto-generated description by cubic. -->
<a href="https://cubic.dev/pr/twentyhq/twenty/pull/22687?utm_source=github" target="_blank" rel="noopener noreferrer" data-no-image-dialog="true"><picture><source media="(prefers-color-scheme: dark)" srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source media="(prefers-color-scheme: light)" srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img alt="Review in cubic" src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->
 2026-07-09 09:38:30 +02:00
github-actions[bot] 9da5383289 i18n - website translations (#22683) 
Created by Github action

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22683?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->

---------

Co-authored-by: github-actions <github-actions@twenty.com>
 2026-07-09 09:12:46 +02:00
Félix Malfait 0755855768 fix(billing) - skip paying upgrade invoices already settled at finalization (#22705) 
## Problem

Upgrading resource credits fails with `Invoice is already paid` whenever
the one-off upgrade invoice resolves to a $0 amount due. That is the
case for internal workspaces on the `Twenty Internal - 100% FREE`
coupon, and for customers whose credit balance covers the price
difference.

Reproduced on an internal workspace (5 to 20 credits upgrade):

1. `createImmediateUpgradeInvoice` creates the invoice for the $20 diff
and finalizes it with `auto_advance: true`
2. The 100% coupon brings the amount due to $0, and Stripe settles
zero-due invoices at finalization ("Invoice was finalised and
automatically marked as paid because the amount due was US$0.00")
3. The explicit `stripe.invoices.pay()` that follows is rejected with a
400 `invalid_request_error`: "Invoice is already paid"
4. The error propagates, so the mutation aborts before
`runSubscriptionUpdate`: the Stripe subscription item stays on the old 5
credits price while the upgrade invoice already exists
5. Every retry creates a new invoice item + invoice and fails the same
way, leaving stray $0 invoices on the customer

## History

Third pass on this code path:

- #21097 treated it as a race with `auto_advance` and switched
finalization to `auto_advance: false`. Zero-due invoices are settled at
finalization regardless of that flag, so the failure remained.
- #21450 restored `auto_advance: true` and swallowed the error when
`error.code === 'invoice_already_paid'`. Stripe does not send that code
for this failure (it is not in its documented error codes; the response
only carries the message), so the guard never matched and the error was
always rethrown.

## Fix

Rely on invoice status instead of error codes:

- `finalizeInvoice` returns the finalized invoice; when it comes back
`paid` (the zero-due case), skip `pay` entirely
- if `pay` still fails (the genuine auto_advance race from #21097),
re-retrieve the invoice and only rethrow when it is actually unpaid

## Tests

Unit tests for `createImmediateUpgradeInvoice`:

- open invoice after finalization gets paid
- invoice settled at finalization skips `pay`
- `pay` failure with a meanwhile-paid invoice is swallowed
- `pay` failure with an unpaid invoice is rethrown

---
_Generated by [Claude
Code](https://claude.ai/code/session_01BwoUffgasLUsXsaGuANsAP)_

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22705?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->
 2026-07-09 09:09:58 +02:00
github-actions[bot] 1129cc7ae0 chore: sync AI model catalog from models.dev (#22710) 
Automated daily sync of `ai-providers.json` from
[models.dev](https://models.dev).

This PR updates pricing, context windows, and model availability based
on the latest data.
New models meeting inclusion criteria (tool calling, pricing data,
context limits) are added automatically.
Deprecated models are detected based on cost-efficiency within the same
model family.

**Please review before merging** — verify no critical models were
incorrectly deprecated.

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22710?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->

Co-authored-by: FelixMalfait <6399865+FelixMalfait@users.noreply.github.com>
 2026-07-09 09:08:22 +02:00
Etienne cc7b41db0e feat(ai-chat): inject current date & per-message timestamps into agent context (#22632) 
## Summary

Gives the AI chat agent temporal awareness by injecting the current date
into
the system prompt and a per-message "sent at" timestamp into each user
message,
formatted in the member's timezone. Also hardens all timezone formatting
against
the `"system"` sentinel value, which was crashing the stream job.

## What changed

**Message timestamps (new)**
- Added `injectMessageTimestamps` util: prepends a
`<message_timestamp>Sent: …</message_timestamp>`
text part to each user message before it's sent to the model, so the
agent can
  reason about "yesterday", "last week", etc.
- `loadMessagesFromDB` now stores the message time in the canonical
`metadata.createdAt` slot (ISO string, JSON-serializable for the BullMQ
job
payload) instead of a non-typed top-level `createdAt` field that nothing
read.
- Migrated the AI chat message pipeline from the generic `UIMessage` to
the
typed `ExtendedUIMessage` (`chat-execution.service`,
`extract-code-interpreter-files`,
`replace-unsupported-file-parts`, and related types), since
`metadata.createdAt`
  is declared on `ExtendedUIMessage`.

**Current date in context**
- System prompt now includes `Current date: …` formatted in the member's
timezone
  (`system-prompt-builder.service`).
- Settings › AI prompt preview mirrors the same `Current date` line.

**Timezone safety (bug fix)**
- Workspace members default `timeZone` to the `"system"` sentinel, which
is only
  resolvable client-side. Passing it (or any invalid IANA zone) to
`Intl.DateTimeFormat` throws `RangeError: Invalid time zone specified:
system`,
  which was failing the stream job.
- Added `getValidTimeZoneOrUndefined`, which returns a valid IANA zone
or
`undefined` (letting the runtime fall back to its default). Used in both
`injectMessageTimestamps` and `formatCurrentDate`. This mirrors the
existing
  `isValidTimeZone` convention in the calendar module.

## Notes / follow-ups

- For members who never changed `timeZone` from `"system"`, timestamps
fall back
to the server's default zone (UTC). To honor their real local time, the
frontend would need to send the browser-detected zone with the chat
request
(the same way calendar/charts already pass a resolved zone). Not
included here.

## Test plan

- [x] `inject-message-timestamps.util.spec.ts` — covers timestamp
injection,
assistant messages untouched, invalid `createdAt`, and the `"system"`
      timezone no longer throwing.
- [ ] Send a chat message and confirm the agent sees the correct
date/time.
- [ ] Verify a member with `timeZone = "system"` no longer crashes the
stream job.

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22632?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->
 2026-07-09 07:02:56 +00:00
neo773 3a5545c753 chore: remove IS_MESSAGING_CALENDAR_WEBHOOK_ENABLED flag (#22680) 
Messaging/calendar webhook subscriptions are now always on; drop the
feature flag gate and its enum/public-flag registration.

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22680?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->
 2026-07-08 23:35:35 +02:00
Weiko 97e21c3403 feat(ai): optionally preselect fast or smart model when opening Ask AI (#22679) 
Allow front components to preselect the AI model when opening the Ask AI
side panel with a preprompt.

- Add an optional `model: 'FAST' | 'SMART'` to the preprompt in the
  openSidePanelPage AskAI params of the front-component SDK.
- openAskAiPageWithPreprompt resolves FAST to the workspace fast model
  and SMART to the workspace default (null selection = pinned default
  model), writing to agentChatUserSelectedModelState.
 2026-07-08 22:52:38 +02:00
nitin 78a0f9ea77 feat(call-recorder): type application and server variables, make summary prompt rich text (#22685) 2026-07-08 22:39:47 +02:00
nitin 545de99476 [Twenty Fireflies] Add calendar event summary and transcript UI components (#22667) 
## Summary
Add comprehensive UI components and hooks for displaying call recording
summaries and transcripts on calendar event pages. This includes
markdown parsing for summaries, diarized transcript rendering, and data
fetching hooks integrated with the Twenty SDK.

## Key Changes

### New Hooks
- `useCalendarEventSummary`: Fetches and manages summary markdown for a
calendar event's call recordings
- `useCalendarEventTranscript`: Fetches and manages transcript data for
a calendar event's call recordings

### Summary Components
- `CalendarEventSummary`: Top-level component that displays summary for
selected calendar event
- `CalendarEventSummaryContent`: Container with header and content frame
- `CalendarEventSummaryBody`: Handles loading, error, and empty states
- `SummaryMarkdown`: Renders parsed markdown with support for headings,
lists, and paragraphs
- `SummaryInlineSegments`: Renders inline text with bold formatting
support

### Transcript Components
- `CalendarEventTranscript`: Top-level component that displays
transcript for selected calendar event
- `CalendarEventTranscriptContent`: Container with header and scrollable
content frame
- `CalendarEventTranscriptBody`: Handles loading, error, and empty
states
- `TranscriptEntryList`: Renders list of transcript entries
- `TranscriptEntryListItem`: Individual transcript entry with speaker
avatar, timestamp, and text
- `TranscriptErrorBox`: Styled error state display

### Utilities
- `parseSummaryMarkdownBlocks`: Parses markdown into structured blocks
(headings, lists, paragraphs)
- `parseSummaryInlineSegments`: Parses inline markdown for bold text
formatting
- `parseTranscriptEntries`: Parses diarized transcript format into
structured entries with speaker info and timestamps
- `formatSecondsAsClockTimestamp`: Formats seconds into HH:MM:SS or
MM:SS format
- `asRecord`: Type guard utility for converting values to records

### Page Layout Configuration
- `calendar-event-summary-tab.ts`: Defines Summary tab for calendar
event page layout
- `calendar-event-transcript-tab.ts`: Defines Transcript tab for
calendar event page layout
- Front component definitions for both summary and transcript

### Types
- `TranscriptEntry` and `TranscriptWord`: Structured transcript data
types
- `SummaryMarkdownBlock`: Block-level markdown structure
- `SummaryInlineSegment`: Inline text segment with formatting

## Implementation Details

- Uses `CoreApiClient` from Twenty SDK to query call recordings filtered
by calendar event ID
- Implements proper cleanup with cancellation tokens to prevent state
updates on unmounted components
- Supports both loading and error states with user-friendly messaging
- Markdown parser handles headings (h1-h6), bullet lists, and paragraphs
with inline bold formatting
- Transcript parser validates diarized format and gracefully handles
malformed entries
- Styled with emotion and theme constants from twenty-ui for consistent
design
- Integrates with `useSelectedRecordIds` hook to track selected calendar
event



https://github.com/user-attachments/assets/0bd63590-bb1e-4f89-8b96-e3fbb659473e


<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22667?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->
 2026-07-09 00:52:20 +05:30
martmull 16e5d2b9d0 feat(website): add public apps marketplace with one-click install (#22611) 
<img width="1335" height="570" alt="image"
src="https://github.com/user-attachments/assets/85bbe656-7525-4800-9ae6-bc59445ccf48"
/>
<img width="1512" height="723" alt="image"
src="https://github.com/user-attachments/assets/6494c539-a3e0-4e95-9df5-ee7559c098ab"
/>

<img width="1512" height="813" alt="image"
src="https://github.com/user-attachments/assets/4cb6adea-d35b-470a-9bfd-d47b2a514e8c"
/>
<img width="1512" height="739" alt="image"
src="https://github.com/user-attachments/assets/757ea3e0-bff9-4482-9a71-3e75949fd7e7"
/>
<img width="1468" height="790" alt="image"
src="https://github.com/user-attachments/assets/37365c56-47ae-4f08-a40f-32e5600a6c0f"
/>


## What

Adds a public `/apps` marketplace on twenty-website listing the vetted,
Twenty-built apps. Each app card and detail page has a one-click Install
button that deep-links to the in-app available-application page
(`app.twenty.com/settings/applications/available/:universalIdentifier`),
where authentication and permission consent are handled before
installing.

- New apps directory (`/apps`) and per-app detail pages
(`/apps/[slug]`), with a category filter
- Routing/sitemap entry for `/apps` and an Apps link in the Resources
menu
- The catalog is fetched from the Twenty GraphQL API (public
`publicMarketplaceApps` / `publicMarketplaceAppDetail` queries),
defaulting to `api.twenty.com`, and degrades to an empty state on
failure. No app info is duplicated in the website; logos and screenshots
come from the catalog's CDN URLs.

## Dependency

This is the website half of the split. It consumes the public queries
added in the server PR #22647, which should merge first.

---------

Signed-off-by: Parship Chowdhury <parshipchowdhury@gmail.com>
Co-authored-by: martmull <martin@twenty.com>
Co-authored-by: Paul Rastoin <45004772+prastoin@users.noreply.github.com>
Co-authored-by: nitin <142569587+ehconitin@users.noreply.github.com>
Co-authored-by: Aressand <97886962+Aressand@users.noreply.github.com>
Co-authored-by: Brahm Lower <bplower@gmail.com>
Co-authored-by: Thomas Trompette <thomas.trompette@sfr.fr>
Co-authored-by: Etienne <45695613+etiennejouan@users.noreply.github.com>
Co-authored-by: Parship Chowdhury <parshipchowdhury@gmail.com>
Co-authored-by: Raphaël Bosi <71827178+bosiraphael@users.noreply.github.com>
Co-authored-by: Charles Bochet <charlesBochet@users.noreply.github.com>
Co-authored-by: Félix Malfait <felix.malfait@gmail.com>
Co-authored-by: Pratik Mahajan <Pratik@mahajan.xyz>
Co-authored-by: neo773 <neo773@protonmail.com>
Co-authored-by: Deepak kumar maharana <100968930+deep231w@users.noreply.github.com>
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Co-authored-by: github-actions <github-actions@twenty.com>
Co-authored-by: Marie <51697796+ijreilly@users.noreply.github.com>
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
Co-authored-by: neo773 <62795688+neo773@users.noreply.github.com>
 2026-07-08 19:12:59 +00:00
Marie 5f3f734b34 fix(front): page header title overlap, Cmd+K on page layout pages & stable tooltip id (#22678) 
## Summary

Three independent front-end fixes.

### 1. Settings page header title overlaps the breadcrumb

On settings detail pages (e.g. an app's logic function), a long centered
title visually overlapped the breadcrumb.

`PageCardHeader` renders the header as a CSS grid (`minmax(0, 1fr) auto
minmax(0, 1fr)`) and the centered title used `justify-self: center`.
With grid, `justify-self: center` sizes the item to its own content
width (up to its `max-width`) instead of to its grid track, so a long
title grew wider than the center track and spilled sideways over the
breadcrumb column — its `overflow: hidden` only clipped its own children
to that oversized box, not to the track.

Fix: let the centered title fill and shrink to its grid track so it
clips (with ellipsis) inside its own column instead of overflowing into
the breadcrumb.
- Center column: `auto` → `minmax(0, auto)` so it can shrink when space
is tight.
- Centered title: `justify-self: center` → `justify-self: stretch` +
`min-width: 0`.

### 2. Cmd+K does nothing on standalone page layout pages

On standalone page layout pages (`/page/:pageLayoutId`, used to render
app front components), the command menu shortcut (Cmd+K) did nothing.

Cmd+K is a global hotkey with a modifier, so it only runs when the
active focus-stack config has `enableGlobalHotkeysWithModifiers: true`.
`RecordIndexPage` and `RecordShowPage` explicitly reset the focus stack
to enable it, but `PageChangeEffect` had no case for
`AppPath.PageLayoutPage`, so the stack kept a stale config (commonly the
Settings config, which disables modifier hotkeys) and swallowed the
shortcut.

Fix: add a `PageLayoutPage` case in `PageChangeEffect` that resets the
focus stack with modifier hotkeys enabled (mirroring `RecordShowPage`),
plus a new `PageFocusId.PageLayoutPage` value.

### 3. Ever-changing / unstable tooltip element id

`OverflowingTextWithTooltip` built its element id from `title-id-${+new
Date()}`, so a new id (the current epoch time in ms) was generated on
every render — the id visibly kept increasing in the DOM. This is
unstable (the tooltip anchor `#id` churns on each render) and
collision-prone (two tooltips rendering in the same millisecond get the
same id, producing duplicate DOM ids and an ambiguous anchor).

Fix: derive the id from React's `useId()` so it is stable per instance
and unique. The colons `useId()` produces are stripped, since the id is
used inside a CSS selector (`anchorSelect={#${id}}`) where colons are
invalid.

## Test plan

- [ ] Open a settings detail page with a long title (e.g. an app logic
function named `maintain-account-team-member-name-on-created`) and
confirm the title no longer overlaps the breadcrumb, and truncates with
an ellipsis when space is tight.
- [ ] Navigate to a standalone page layout page (an app's
front-component page) and confirm Cmd+K opens the command menu,
including after coming from Settings.
- [ ] Inspect an overflowing title/tooltip in DevTools and confirm its
`id` stays stable across re-renders (no longer increments), and tooltips
still show on hover of truncated text.
 2026-07-08 18:51:03 +02:00
Thomas Trompette ca90a9358f feat(workflow): backfill workspace workflowVersion into core (phase A) (#22663) 
## workflowVersion -> core, Phase A

Follows #21674 (Phase 0, merged). Base: `main`.

Populates core `workflowVersion` and keeps it in sync with the workspace
object, so a later phase can switch reads to core. Reads stay on the
workspace object in this PR.

### 1. Backfill (upgrade command)
`BackfillWorkflowVersionToCoreCommand`, a
`@RegisteredWorkspaceCommand('2.20.0', ...)`. Per workspace, reads all
workspace `workflowVersion` records and upserts them into core,
preserving ids (idempotent), dry-run aware.

### 2. Dual-write (always on, not flag-gated)
`WorkflowVersionCoreDualWriteListener` hooks
`@OnDatabaseBatchEvent('workflowVersion', CREATED/UPDATED/DELETED)`
(same mechanism as the existing workflow-version status listener) and
mirrors every mutation into core. Sync failures are logged, never break
the user's write; drift is repaired by re-running the backfill command.

Dual-write is deliberately not behind a flag: reading from core (next
phase) is only safe if core has been continuously in sync since the
backfill. An always-on mirror makes "core is fresh" an invariant, so the
read switch becomes a plain flag flip. The cost is one extra upsert on
infrequent workflowVersion writes.

`IS_WORKFLOW_VERSION_IN_CORE_ENABLED` is reserved for the read switch
(Phase B): dispatch from the `workflowAutomatedTriggerMaps` cache,
runner and builder reading trigger/steps from core. Until then it gates
nothing.

Both the backfill and the listener go through a single
`WorkflowVersionCoreSyncService` (`upsertToCore`/`deleteFromCore`): the
workspace-to-core mapping (`trigger` -> `triggers[]`, plus `steps`,
`status`, `workflowId`) and `workflowAutomatedTriggerMaps` invalidation
live in one place.

### Rollout plan (following phases)
- **B, read switch (flag per workspace):** reads move to core; writes
keep flowing workspace -> listener -> core. Rollback = flip the flag
back, workspace never stopped being source of truth.
- **C, contract (code change):** write paths write trigger/steps to core
directly; workspace `workflowVersion` stays as a thin shell
(nav/relations/search) but drops the trigger/steps columns; listener and
flag removed.

### Not in this PR
- Reconciliation tooling beyond re-running the backfill.
- The read switch (Phase B).
 2026-07-08 18:45:57 +02:00
martmull 0ea0c23556 refactor(app-marketplace): rename featured to vetted (#22674) 
## What

Renames the application-registration "featured" flag to "vetted" across
the backend, frontend, GraphQL schema/DTOs, and the marketplace UI.

"Vetted" better describes what the flag actually does today: it marks an
app as reviewed and approved by the Twenty team (a trust signal), rather
than "featured" which reads as spotlighting/promotion. The admin toggle
description was already "Mark this app as reviewed and approved".

## How

- Renamed `isFeatured` -> `isVetted` on the `ApplicationRegistration`
entity, DTOs (`MarketplaceApp`, `MarketplaceAppDetail`,
`UpdateApplicationRegistrationPayload`), services, GraphQL fragments,
and the settings/admin UI (labels: "Featured" -> "Vetted", "Featured
only" -> "Vetted only", etc.).
- Renamed the `MARKETPLACE_FEATURED_APPLICATIONS` constant/file to
`MARKETPLACE_VETTED_APPLICATIONS`.
- Regenerated GraphQL client artifacts (`generated-metadata`,
`generated-admin`, `twenty-client-sdk`).

### Database

The `isFeatured` column is renamed in place to `isVetted` via a single
2.20 fast instance command (`ALTER TABLE ... RENAME COLUMN`). No new
column, no data-copy backfill.

- Since all 2.19 commands (including the existing `isFeatured` backfill)
complete before any 2.20 command runs, the rename carries over the
values that backfill set.
- The entity uses `@WasRenamedInUpgrade` so the upgrade-aware layer
queries the old column name until the rename step runs during an
upgrade.

## Testing

- `nx typecheck` and `nx lint:diff-with-main` pass for twenty-server and
twenty-front.
- Ran `database:reset` on a fresh dev DB: the 2.19 `isFeatured` backfill
runs first, then the 2.20 rename; the column ends up as `isVetted` (and
`isFeatured` no longer exists), values preserved.
- Booted the server: the `@WasRenamedInUpgrade` decorator validates
against the upgrade sequence, and GraphQL introspection confirms all
four types expose `isVetted` and none expose `isFeatured`.
- Ran the three `graphql:generate` configs and the SDK metadata client
generator so the committed generated files match the generator output
(field ordering included).

## Notes

- The `api-breaking-changes` check flags the removal of the `isFeatured`
GraphQL field — that is expected and inherent to this rename.
- Translation catalogs (`locales/`) are intentionally not touched here
since they are managed via Crowdin; new English strings render via
Lingui's default-message fallback until translated.
 2026-07-08 16:41:58 +00:00
Paul Rastoin 163c96c2e5 Validate range version app dev sync (#22625) 
# Introduction
Also now validating the workspace version when running a sync manifest

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22625?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->
 2026-07-08 16:37:42 +00:00
neo773 b5a73ad86a Chore/remove messaging mock specs (#22665) 
<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22665?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->
 2026-07-08 18:37:06 +02:00
neo773 674de0056b feat(messaging): message campaign delivery stats + views (#22661) 
Re-land of #22452 (reverted in #22627). Rebuilt on fresh main with
upgrade commands isolated to 2-20 only; no other version's commands
touched.

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22661?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->
 2026-07-08 15:55:20 +00:00
Marie e751c589ca [Billing for self hosts] Add info in description (#22660) 
Clarify how the same enterprise key is reused for one prod + one dev
instance.
Suggest to save / store enterprise key in clear somewhere.

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22660?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->

---------

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
 2026-07-08 15:54:09 +00:00
Charles Bochet 89b6037db2 fix(docker): bump node base to 24.18.0 (all 24.17.0 CVE fixes + http.Agent premature-close fix) (#22677) 
## Context

Follow-up to #22673, which pinned the base image back to
`node:24.16.0-alpine` to stop the prod flood of `Invalid response body
while trying to fetch …: Premature close` failures on Gmail/Calendar
sync and Cloudflare checks introduced by the 24.17.0 bump (#22529).

The regression is confirmed upstream: 24.17.0's response-queue-poisoning
fix (CVE-2026-48931) attaches a public `'data'` listener on idle
keep-alive sockets in the `http.Agent` pool, which false-triggers
node-fetch@2's premature-close detection whenever a server abruptly
resets a keep-alive socket right after a **complete** response —
standard behavior for Google's front end. Reported the day 24.17.0
shipped (nodejs/node#63989, #64098) and fixed by nodejs/node#64004,
released in **Node 24.18.0 (2026-06-23)**.

## What this PR does

Bumps all four stages to `node:24.18.0-alpine3.23` (digest-pinned).
24.18.0 is the current 24 LTS and contains:
- everything from 24.17.0: OpenSSL 3.5.7, CVE-2026-48930 (CVSS 9.8), and
the response-queue-poisoning guard itself — reimplemented via the
socket's internal `onread` hook instead of a public stream listener
(nodejs/node#64004)
- so we get the full security posture back **and** the regression fix.

## Verification

Deterministic repro (complete chunked response over keep-alive, then
abrupt socket destroy — per nodejs/node#64098), run against all three
images with node-fetch v2 and v3:

| Node | node-fetch@2 | node-fetch@3 |
|------|--------------|--------------|
| 24.16.0 | OK | OK |
| 24.17.0 | **`ERR_STREAM_PREMATURE_CLOSE: Invalid response body …
Premature close`** (byte-for-byte the prod Sentry error) | OK |
| 24.18.0 | OK | OK |

node-fetch@2 is what the Gmail batch layer
(`@jrmdayn/googleapis-batcher`) and the Cloudflare client resolve to,
matching the affected prod paths.

## Related

- #22673 — interim rollback to 24.16.0 (shipped as twenty/v2.19.1); this
PR supersedes it
- #22671 — classifies `ERR_STREAM_PREMATURE_CLOSE` as a transient
retryable network error; still worth landing since servers legitimately
reset keep-alive sockets

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22677?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->
 2026-07-08 17:44:04 +02:00
martmull 9423af7f67 feat(server): add public marketplace resolver for vetted app catalog (#22647) 
## What

Adds a public GraphQL resolver so unauthenticated clients (the public
website) can read the listed/vetted marketplace catalog without a
workspace token.

- `MarketplacePublicResolver` (metadata schema) exposes two public
queries guarded by `PublicEndpointGuard` + `NoPermissionGuard`:
  - `publicMarketplaceApps`
  - `publicMarketplaceAppDetail(universalIdentifier)`
  
Both delegate to the existing `MarketplaceQueryService` (no new logic,
no new REST routing). The existing workspace-guarded
`findManyMarketplaceApps` / `findMarketplaceAppDetail` queries are
untouched.
- Adds a shared `ApplicationCategory` type in `twenty-shared` (known
values plus `string` for backward compatibility) used to type
`ApplicationManifest.category`. A warning is logged server-side when an
app declares a category outside the known set.

## Why

This is the backend half of the public apps marketplace on the website.
Splitting it out so the server-side catalog exposure can be reviewed
independently from the website UI.

## Follow-up

The website PR (the `/apps` marketplace UI) consumes
`publicMarketplaceApps` and should merge after this one.

---
_Generated by [Claude
Code](https://claude.ai/code/session_01GBfegArtJcoiTLSsnWPH8R)_

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22647?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->

---------

Co-authored-by: martmull <martin@twenty.com>
 2026-07-08 15:43:17 +00:00
github-actions[bot] b1dee7cf26 i18n - docs translations (#22676) 
Created by Github action

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22676?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->

Co-authored-by: github-actions <github-actions@twenty.com>
 2026-07-08 17:24:00 +02:00
Charles Bochet 840c41e6b7 fix(docker): pin node base back to 24.16.0 to stop premature-close fetch failures (#22673) 
## Context

Since v2.19.0 rolled out to prod (2026-07-07), workers are flooded with
mid-body fetch failures — `Invalid response body while trying to fetch
…: Premature close` — on Gmail message import (Sentry TWENTY-SERVER-D3X,
~22k events/day, ~1k users) and, simultaneously, on Cloudflare
custom-domain checks (TWENTY-SERVER-HXW/HXT). Both code paths were
unchanged between v2.18.5 and v2.19.0; their only shared layer is the
runtime HTTP stack.

The one relevant change in v2.19.0 is #22529: the base image bump
`node:24.16.0-alpine` → `node:24.17.0-alpine`. Node 24.17.0 patched
exactly the components under these fetches:
- `http`: CVE-2026-48931 — idle keep-alive sockets in the Agent pool now
get `socket.resume()` + a destroy-on-data guard on every free→reuse
cycle
- `deps`: llhttp 9.4.2 (security bump of the HTTP parser that decides
when a chunked body is complete)

The messaging import loop cycles the same keep-alive socket through the
pool once per message fetched, so any per-cycle failure probability is
amplified by prod volume. (Note: undici's equivalent CVE fix needed two
follow-up commits for races of this exact kind — sockets destroyed while
freshly handed to a request.)

## What this PR does

Pins the base image back to `node:24.16.0-alpine3.23` (same digest
v2.18.5 shipped with) on all four stages, and updates the security note
accordingly.

This doubles as the definitive root-cause test: if the premature-close
rate drops back to its historical baseline on the rebuilt image, the
24.17.0 HTTP-stack change is confirmed and we can file a solid upstream
report to nodejs/node.

## Trade-off — please weigh in

This re-exposes what #22529 fixed: 24.16.0 statically links OpenSSL
3.5.6, so the scanner will re-flag CVE-2026-48930 (TLS embedded-nul
hostname authority rebinding, CVSS 9.8) on the node binary. There is no
24.x release newer than 24.17.0 yet. The Dockerfile carries a TODO to
re-bump as soon as a fixed 24.x ships.

## Related

- #22671 classifies `ERR_STREAM_PREMATURE_CLOSE` as a transient
(retryable) network error at the application level — worth landing
regardless of this rollback, since sync channels currently hard-fail to
`FAILED_UNKNOWN` on what is a plain network race.

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22673?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->
 2026-07-08 17:19:09 +02:00
Deepak kumar maharana 34c5054bac Fix email validation for over-length inline edits (#22426) 
## Summary

This PR addresses the inconsistency reported in #22406 where over-length
email values were accepted by the inline editor, optimistically shown as
saved, and then rejected by the backend.

### Changes

- Await `updateOneRecord` before updating the local record store, so the
UI is only updated after a successful mutation. This prevents the
optimistic state from showing values that failed to persist.
- Add a client-side maximum length validation (`255`) to `emailSchema`
so over-length email values are rejected before the GraphQL mutation is
sent.
- Propagate the client-side validation message through
`MultiItemFieldInput` so validation failures are surfaced immediately
instead of silently preventing the save.

### Verification

- Valid email addresses continue to save successfully.
- Over-length email values are rejected on the client without sending a
GraphQL request.

Related to #22406.

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22426?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->
 2026-07-08 17:01:56 +02:00
Pratik Mahajan 72d728ea8e fix(messaging): add sender name to IMAP/SMTP From headers (#22603) 
Manual IMAP/SMTP outbound emails were being composed with a
bare email address in the From header, so recipients did not
see the sender's display name.
Gmail already built a proper sender header from Google profile
data, but manually configured IMAP/SMTP accounts had no
equivalent path and fell back to the raw address only.

Fix this by storing the optional sender display name in the
IMAP/SMTP/CALDAV connection parameters, exposing it through
the settings flow and metadata API, and reusing a shared
From-header formatter when composing outbound messages.
The formatter now builds a properly encoded sender header when
a name is available and falls back to the bare email address
when it is not, keeping the behavior safe for blank or missing names.
Gmail keeps using its existing Google-derived display name source;
this change only brings manual accounts up to the same header
formatting standard and removes duplicated formatting logic between
outbound drivers.

After this change, manual SMTP sends and IMAP draft creation include
the configured sender name in the From header, while blank names are
normalized away instead of producing malformed headers.
Existing manual account names are preserved when updates omit the
field, and edited accounts can still explicitly clear it through the
settings flow.

Add focused utility coverage for the shared From-header formatter.

Fixes: #22608

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22603?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->

---------

Co-authored-by: neo773 <neo773@protonmail.com>
 2026-07-08 20:29:25 +05:30
Félix Malfait b5b6e110b0 Merge app Public URL and Public Domains settings into one App URL section (#22666) 
## Context

The app settings tab showed two sections for the same concept: a "Public
URL" block with a very technical description (Permissions-Policy, COOP,
COEP), a blue info banner about the legacy `/s/` endpoint, and a
separate "Public Domains" block. Confusing, and scary-sounding for apps
whose routes are not really "public".

## Changes

**One merged "App URL" section** (still only rendered for apps that
actually expose HTTP-triggered functions):
- Title renamed to the neutral "App URL" with a one-line description:
"This app's routes are served from this URL. Add a custom domain to use
your own."
- Read-only copyable base URL input, custom domain list card right below
it.
- Removed the info banner and all header/CORS jargon.

**Always show the real URL**: `getFunctionsBaseUrl` now takes
`serverBaseUrl` and falls back to `${serverBaseUrl}/s` instead of
returning `undefined`, so self-hosted instances (no dedicated function
domain) see their actual base URL instead of nothing. This centralizes
the fallback previously duplicated in `FrontComponentRenderer` and
`getLogicFunctionHttpUrl`.

**"Public Domain" renamed to "Custom Domain"** in all user-facing
strings (add card, footer button, detail page, snackbars). Internal
identifiers, GraphQL types and routes keep the `PublicDomain` name to
match the backend entity.

**Polish**:
- Domain rows and the add card use a world icon instead of the mail
icon.
- Row description shows "Added x days ago" instead of a raw ISO
timestamp, via a new shared `useGetAddedRelativeDateDescription` hook
also adopted by the approved access domains card that had the same
inline helper.
- `FrontComponentRenderer` consumes `functionsBaseUrl` from
`useGetLogicFunctionHttpUrl` instead of re-deriving it from the same
atoms.
- User docs updated accordingly.

## Testing

- Ran the app locally (Postgres/Redis + server + front), synced a
fixture app with an HTTP-triggered logic function, and verified the
merged section renders correctly with the copyable URL and Add Custom
Domain card, no banner, no duplicate section.
- `getLogicFunctionHttpUrl` unit tests updated and passing (7/7), `npx
nx typecheck twenty-front` and `lint:diff-with-main` green.
- Locale catalogs are untouched; Crowdin sync regenerates them from the
new source strings.

---
_Generated by [Claude
Code](https://claude.ai/code/session_01NyEWoybMBq62cSNfAFdJAT)_

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22666?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->
 2026-07-08 16:44:26 +02:00
Charles Bochet d746909184 feat(sdk): validate graph page-layout widgets at build time (#22559) 
When an app defines a graph widget (aggregate, pie, bar or line chart),
the built manifest can carry the wrong key and the server rejects it at
sync time with a confusing "aggregate field is required" error.

The SDK type already requires
`aggregateFieldMetadataUniversalIdentifier` and renames the raw
`aggregateFieldMetadataId` at compile time. But the manifest build runs
esbuild with no type checking, so a wrong or missing key slips through
and only fails later on the server.

This adds a build-time check that mirrors the server validator, with a
hint pointing at the right key when the raw one was used. It is
non-breaking since correctly authored apps already use the universal
key.

Tests: unit tests on the validator, plus a real graph widget added to
the rich-app fixture so the integration and e2e suites cover the happy
path.
 2026-07-08 16:25:47 +02:00
Raphaël Bosi ad74f75e6a Make onboarding steps responsive on mobile (#22659) 
https://github.com/user-attachments/assets/5840936c-d9b8-412f-bfec-2d5af768660c



The onboarding flow was built for a fixed 440px design with no
responsive handling, so on a phone the steps were cramped: oversized
padding/gaps, a decorative import-preview illustration that clipped, and
side-by-side name fields.

Adds targeted `@media (max-width: 768px)` rules (using the existing
`MOBILE_VIEWPORT`):
- Shared step page: smaller padding and gap on mobile (cascades to every
step).
- `UpgradeFreeTrial`: its own mobile padding (it overrides the base
padding).
- Header: reduced side padding.
- Import preview: hide the floating calendar cards on mobile (their
fixed offsets are tuned for the 440px card).
- Create profile: stack the avatar + name fields vertically on mobile.

Verified each step at 375px via Storybook; desktop is unchanged (media
queries gated to <=768px).

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22659?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->
 2026-07-08 16:05:26 +02:00
nitin 640e6b8b33 fix(call-recorder): stream Recall media to storage to fix OOM (#22652) 
## Summary

Fixes Call Recorder media ingestion OOMs by streaming Recall media into
Twenty direct uploads instead of buffering the full file in memory.

## Changes

- Opens the Recall media download stream and uses its `Content-Length`
as the direct upload size.
- Creates a Twenty direct upload target, streams the media body to it
with Node `http`/`https` backpressure, then completes the upload.
- Cleans up download/upload streams on target creation, upload, and
storage response failures.
- Keeps the media size cap for now while making it no longer required
for memory safety.
- Bumps `twenty-client-sdk` and `twenty-sdk` to `2.19.0`.

## Tests

- `yarn test:unit
src/logic-functions/flows/__tests__/ingest-call-recording-media.test.ts
src/logic-functions/flows/__tests__/put-media-download-body-to-upload-target.test.ts`
- `yarn typecheck`
 2026-07-08 15:48:49 +02:00
Thomas Trompette 48730df0d2 feat(workflow): scaffold core workflowVersion entity + trigger cache (phase 0) (#21674) 
## What

**Phase 0 (scaffold)** of migrating `workflowVersion` data to **core**.
Gated by `IS_WORKFLOW_VERSION_IN_CORE_ENABLED` with **no behavior
change** — nothing reads or writes the new core entity yet.

## Plan

`workflowVersion` becomes a thin **workspace shell** over a core entity
(the `dashboard`/`pageLayout` pattern), so navigation, the metadata
relations, and the record UI keep working while the heavy data
(`triggers`, `steps`) lives in core. Trigger dispatch will derive from
active core versions via a per-workspace cache, letting us **eliminate**
the denormalized `workflowAutomatedTrigger` object. `workflow` and
`workflowRun` stay as workspace objects.

Phases: **0 — scaffold (this PR)** → A — backfill + dual-write → B —
switch reads to core → C — drop the workspace `trigger`/`steps` columns
+ the `workflowAutomatedTrigger` object.

## Included
- **Core `WorkflowVersionEntity`** (`extends WorkspaceRelatedEntity`) —
stores version data, with triggers as an **array** (`triggers:
WorkflowTrigger[]`), a long-due shape change. Storage only: dispatch
reads the primary trigger, so behavior stays single-trigger for now.
- **Fast create-table instance command** for `core."workflowVersion"`
(v2.19.0).
- **`IS_WORKFLOW_VERSION_IN_CORE_ENABLED`** feature flag.
- **Per-workspace automated-trigger cache provider** deriving
CRON/DATABASE_EVENT dispatch from the active version's trigger —
groundwork for removing `workflowAutomatedTrigger`.

## Notes
- `WorkspaceRelatedEntity`, **not** `SyncableEntity`: this is user
runtime data (like `connectedAccount`/`apiKey`/`file`), not
application-manifest metadata.
- No frontend behavior; the generated `FeatureFlagKey` enums are updated
to include the new flag.
 2026-07-08 12:50:03 +00:00
Parship Chowdhury 6554440bb1 fix: dropping favorites to the last position in a open folder (#22360) 
Fixes #9213 

Before:


https://github.com/user-attachments/assets/0855f9c1-07c7-4080-8b5f-94fbe3c8b505



After:


https://github.com/user-attachments/assets/6ae26e89-9fe5-4ebe-b512-d8287ed2c070



<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22360?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->

Signed-off-by: Parship Chowdhury <parshipchowdhury@gmail.com>
 2026-07-08 14:28:15 +02:00
Etienne 4b9f393167 fix: add workspace command to backfill missing AGENT source enum values (#22593) 
**What**
Adds a 2.19 workspace upgrade command that backfills missing
FieldActorSource enum values (AGENT) into the Postgres enums backing
every ACTOR composite field (createdBy, updatedBy) across all existing
workspaces.

Each ACTOR field stores its source sub-field as a Postgres enum scoped
to its own table and schema (e.g.
workspace_abc.company_createdBySource_enum). Workspaces created before
these enum values were introduced are missing them, which causes runtime
errors when those sources are used.

**How**
buildActorSourceEnumBackfillTargets collects all ACTOR fields from the
workspace metadata cache and maps each enum sub-property to its
(tableName, columnName, enumName, expectedValues) tuple.
Before iterating over all targets, the command performs a single fast
pg_catalog.pg_enum lookup on company.createdBySource as a representative
sentinel. If AGENT is already present there, the workspace is skipped
entirely (idempotency fast-path).
For each remaining target, ALTER TYPE … ADD VALUE IF NOT EXISTS is
issued per missing value via
WorkspaceSchemaEnumManagerService.addEnumValue, making the command fully
idempotent and safe to re-run.

Fixes
https://discord.com/channels/1130383047699738754/1522507190949118003

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22593?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->
 2026-07-08 12:05:15 +00:00
Thomas Trompette 6bc4d8efac fix(workflow): batch staled run reset to avoid Postgres param limit (#22654) 
## Problem

Self-hosters with a large backlog of workflow runs stuck in `ENQUEUED`
see the recovery job (`WorkflowHandleStaledRunsJob`) fail with:

```
Error: Data validation error.
    at computeTwentyORMException ...
    at WorkspaceSelectQueryBuilder.getMany ...
    at WorkspaceUpdateQueryBuilder.execute ...
    at WorkflowHandleStaledRunsWorkspaceService.handleStaledRunsForWorkspace ...
```

So the very job meant to unblock enqueued runs can never complete, and
runs stay stuck.

## Root cause

`handleStaledRunsForWorkspace` fetched **every** staled run unbounded,
then called `repository.update(allIds, ...)`. That builds a `WHERE id IN
($1, $2, ... $N)`. Inside `WorkspaceUpdateQueryBuilder.execute`, a
"before" `SELECT` runs with that same huge `IN` list; with a big enough
backlog the bind-parameter count exceeds Postgres' limit, the `getMany`
throws a `QueryFailedError`, and `computeTwentyORMException` maps the
resulting PG error code to the generic `PostgresException('Data
validation error.')`.

There's also a secondary `before.length > QUERY_MAX_RECORDS` (200) guard
in the update path that would reject anything over 200 rows even if the
param limit weren't hit.

## Fix

Process staled runs in batches of `QUERY_MAX_RECORDS` (200), looping
until a pass finds none left — the same batching pattern the sibling
clean-runs job already uses. Each update flips the batch from `ENQUEUED`
to `NOT_STARTED`, so the find criteria stops matching them and the loop
terminates. The throttling recompute now runs once at the end, and only
if at least one batch was reset.

## Tests

New unit spec covering:
- no staled runs -> no update, no recompute
- single batch -> correct ids/payload, recompute once
- exactly 200 -> `take: 200`, 200 ids per update
- 450 backlog -> 3 update calls (200/200/50), loops until empty,
recompute exactly once

All 4 pass locally.

## Note

This fixes the recovery job. If runs keep re-accumulating as `ENQUEUED`,
there may be a separate producer-side issue worth investigating.

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22654?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->
 2026-07-08 13:53:26 +02:00
Thomas Trompette bfeaaa56a3 fix(workflow): handle IS/IS_NOT operand in text and array filters (#22640) 
## Problem

Sentry `TWENTY-SERVER-G4F` — `Error: Operand IS not supported for this
filter type` (30k+ occurrences, 15 workspaces, ongoing).

A workflow **Filter** step throws when a step filter carries an
`IS`/`IS_NOT` operand on a text/array field type (`TEXT`,
`MULTI_SELECT`, `EMAILS`, `PHONES`, `ADDRESS`, `LINKS`, `FULL_NAME`,
`ARRAY`, `RAW_JSON`). `evaluateTextAndArrayFilter` only handled
`CONTAINS`/`DOES_NOT_CONTAIN`/`IS_EMPTY`/`IS_NOT_EMPTY` and hit
`default:` → `throw`. The throw propagates out of `FilterWorkflowAction`
and **fails the entire workflow run**.

The current frontend no longer offers `IS`/`IS_NOT` for these types, so
these are **legacy persisted step filters** in older (immutable)
workflow versions that keep executing.

## Fix

Handle `IS`/`IS_NOT` in `evaluateTextAndArrayFilter` as
`contains`/`!contains`, consistent with `evaluateSelectFilter` (chosen
over strict equality because the routed types include arrays/composites
where `==` would silently never match). No existing operand behavior
changes.

## Tests

Added coverage for legacy `IS`/`IS_NOT` on `TEXT` and `MULTI_SELECT`.
Note: the pre-existing `date operands` test failures are
timezone-dependent and unrelated to this change (they fail on `main`
too).

Fixes TWENTY-SERVER-G4F

<!-- This is an auto-generated description by cubic. -->
<a
href="https://cubic.dev/pr/twentyhq/twenty/pull/22640?utm_source=github"
target="_blank" rel="noopener noreferrer"
data-no-image-dialog="true"><picture><source
media="(prefers-color-scheme: dark)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"><source
media="(prefers-color-scheme: light)"
srcset="https://www.cubic.dev/buttons/review-in-cubic-light.svg"><img
alt="Review in cubic"
src="https://www.cubic.dev/buttons/review-in-cubic-dark.svg"></picture></a>
<!-- End of auto-generated description by cubic. -->
 2026-07-08 13:12:08 +02:00