Previously this blocked users who only had SMTP configured to send
outbound emails, this fixes it by making messageChannel and persist
layer conditional
## Summary
- Relocates `AddTableWidgetViewTypeFastInstanceCommand` from `1-22/` to
`1-23/` and bumps its `@RegisteredInstanceCommand` version from `1.22.0`
to `1.23.0`. The original timestamp `1775752190522` is preserved so the
command slots chronologically into the existing 1.23 sequence;
auto-discovered via `@RegisteredInstanceCommand`, no module wiring
change needed.
- Same pattern as #19792 (move
`pageLayoutWidget.conditionalAvailabilityExpression` to 1.23).
## Context
Picking an aggregate option (Count, Sum, Percentage Not Empty, etc.) in
a dashboard Table widget footer did nothing visually — the value never
appeared or updated
## Fix
RecordTableWidget was missing RecordIndexTableContainerEffect, which
reactively syncs currentView.viewFields[].aggregateOperation from the
Apollo cache into the viewFieldAggregateOperationState jotai atom that
the footer reads
<img width="612" height="261" alt="Screenshot 2026-04-17 at 14 17 47"
src="https://github.com/user-attachments/assets/b4409b0e-82a6-4614-bc09-653be738134a"
/>
# Introduction
Even though this would not possible through API at the moment, from
neither API metadata or manifest ( as manifest `permissionsFlag`
declarations etc are done from within a declared role )
Prevent any app to create permissions entities over another app role
from the validation engine itself
## `isEditable`
We might wanna deprecate this column at some point from the entity it
self as now the grain would rather be `what app owns that role ?`
---------
Co-authored-by: Charles Bochet <charles@twenty.com>
## Summary
- Replaces the standalone TypeORM migration
`1775654781000-addConditionalAvailabilityExpressionToPageLayoutWidget.ts`
with a registered fast instance command under
`packages/twenty-server/src/database/commands/upgrade-version-command/1-23/`,
so the `pageLayoutWidget.conditionalAvailabilityExpression` column is
created through the unified upgrade pipeline.
- Uses `ADD COLUMN IF NOT EXISTS` / `DROP COLUMN IF EXISTS` so the new
instance command is a safe no-op for environments that already applied
the previous TypeORM migration.
- Keeps the original timestamp `1775654781000` so the command slots
chronologically into the existing 1.23 sequence; auto-discovered via
`@RegisteredInstanceCommand`, no module wiring needed.
## Context
Reported error when creating a new workspace on `main`:
> column PageLayoutWidgetEntity.conditionalAvailabilityExpression does
not exist
Aligns this column addition with the rest of the 1.23 schema changes
that already use the instance-command pattern.
## Summary
- Add `WorkspaceMigrationGraphqlApiExceptionInterceptor` to
`MarketplaceResolver` and `ApplicationInstallResolver` so validation
failures during app install return `METADATA_VALIDATION_FAILED` with
structured `extensions.errors` instead of generic
`INTERNAL_SERVER_ERROR`
- Update SDK `installTarballApp()` to pass the full GraphQL error object
(including extensions) through the install flow
- Add `formatInstallValidationErrors` utility to format structured
validation errors for CLI output
- Add integration test verifying structured error responses for invalid
navigation menu items and view fields
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
## Summary
Aligns **workspace member** editing and **onboarding** with how the
product is actually used: profile and other “settings” fields go through
**`updateWorkspaceMemberSettings`**, while **`/graphql`** record APIs
follow **object-level** permissions for the `workspaceMember` object.
## Product behaviour
### Completing “Create profile” onboarding
Users who must create a profile (empty name at sign-up) get
`ONBOARDING_CREATE_PROFILE_PENDING` set. The onboarding UI saves the
name with **`updateWorkspaceMemberSettings`**, not with a workspace
record **`updateOne`**.
**Before:** The server only cleared the pending flag on
**`workspaceMember.updateOne`**, so the flag could stay set and
onboarding appeared stuck.
**After:** Clearing the profile step runs when
**`updateWorkspaceMemberSettings`** persists an update that includes a
**name** (same rules as before: non-empty name parts). Onboarding can
advance normally after **Continue** on Create profile.
### Two ways to change workspace member data
| Path | Typical use | Who can change what |
|------|----------------|---------------------|
| **`updateWorkspaceMemberSettings`** (metadata API) | Standard member
fields the app treats as “my profile / preferences” (name,
avatar-related settings, locale, time zone, etc.) | **Always** your
**own** workspace member. Changing **another** member still requires
**Workspace members** in role settings (`WORKSPACE_MEMBERS`). Custom
fields are **not** allowed on this endpoint (unchanged). |
| **`/graphql`** record mutations on **`workspaceMember`** | Custom
fields, integrations, anything that goes through the generic record API
| **`WorkspaceMember`** is special-cased in permissions: **read** stays
**on** for everyone, but **update / create / delete** require
**`WORKSPACE_MEMBERS`**, including updating **your own** row via
`/graphql`. So a **Member** without that permission cannot fix their
name through **`updateWorkspaceMember`**; they use **Settings** /
**`updateWorkspaceMemberSettings`** instead. |
This matches **`WorkspaceRolesPermissionsCacheService`**: for the
workspace member object, `canReadObjectRecords` is always true;
`canUpdateObjectRecords` (and delete-related flags) follow
**`WORKSPACE_MEMBERS`**.
### Hooks and delete side-effects
- Removed **`workspaceMember.updateOne`** pre-query hook and
**`WorkspaceMemberPreQueryHookService`**: they duplicated the same rules
the permission cache already enforces for `/graphql`.
- **`WorkspaceMember.deleteOne`** pre-hook still tells users to remove
members via the dedicated flow; the post-hook only runs the
**`deleteUserWorkspace`** side-effect when a member row is actually
removed—**no** extra settings-permission check there, since only callers
that already passed **object** delete permission can remove the row.
## Tests
- **`workspace-members.integration-spec.ts`**: clarifies and extends
coverage so **`/graphql`** **`updateOne`** is denied for **own** record
on a **standard** name field and on a **custom** field when the role
lacks **`WORKSPACE_MEMBERS`**.
## Implementation notes
- **`OnboardingService.completeOnboardingProfileStepIfNameProvided`**
centralises the “clear profile pending if name present” logic;
**`UserResolver.updateWorkspaceMemberSettings`** calls it after save,
using the typed update payload’s **`name`** (no cast).
- **`UserWorkspaceService.updateUserWorkspaceLocaleForUserWorkspace`**:
drops a redundant **`coreEntityCacheService.invalidate`**;
**`updateWorkspaceMemberSettings`** still invalidates the user-workspace
cache after the mutation.
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.
Co-authored-by: FelixMalfait <6399865+FelixMalfait@users.noreply.github.com>
# Introduction
Gracefully validating that when creating an entity its
`universalIdentifier` is available within the all application metadata
maps context ( current app + twenty standard, currently the only managed
dependencies )
## Summary
Adding `https://api.twenty.com/mcp` as an MCP server in Claude fails
with `Couldn't reach the MCP server` before OAuth can start. Two
independent bugs cause this:
1. **Missing path-aware well-known route.** The latest MCP spec
instructs clients to probe `/.well-known/oauth-protected-resource/mcp`
before `/.well-known/oauth-protected-resource`. Only the root path was
registered, so the path-aware request fell through to
`ServeStaticModule` and returned the SPA's `index.html` with HTTP 200.
Strict clients (Claude.ai) tried to parse it as JSON and gave up. Fixed
by registering both paths on the same handler.
2. **Stale protocol version.** Server advertised `2024-11-05`, which
predates Streamable HTTP. We've implemented Streamable HTTP (SSE
response format was added in #19528), so bumped to `2025-06-18`.
Reproduction before the fix:
```
$ curl -s -o /dev/null -w "%{http_code} %{content_type}\n" https://api.twenty.com/.well-known/oauth-protected-resource/mcp
200 text/html; charset=UTF-8
```
After the fix this returns `application/json` with the RFC 9728 metadata
document.
Note: this is separate from #19755 (host-aware resource URL for
multi-host deployments).
## Test plan
- [x] `npx jest oauth-discovery.controller` — 2/2 tests pass, including
one asserting both routes are registered
- [x] `npx nx lint:diff-with-main twenty-server` passes
- [ ] After deploy, `curl
https://api.twenty.com/.well-known/oauth-protected-resource/mcp` returns
JSON (not HTML)
- [ ] Adding `https://api.twenty.com/mcp` in Claude reaches the OAuth
authorization screen
🤖 Generated with [Claude Code](https://claude.com/claude-code)
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
## Summary
- Replaced the `deep-equal` npm package with the existing
`fastDeepEqual` from `twenty-shared/utils` across 5 files in the server
and shared packages
- `deep-equal` was causing severe CPU overhead in the record update hot
path (`executeMany` → `formatTwentyOrmEventToDatabaseBatchEvent` →
`objectRecordChangedValues` → `deepEqual`, called **per field per
record**)
- `fastDeepEqual` is ~100x faster for plain JSON database records since
it skips unnecessary prototype chain inspection and edge-case handling
- Removed the now-unnecessary `LARGE_JSON_FIELDS` branching in
`objectRecordChangedValues` since all fields now use the fast
implementation
## Summary
Follow-up to #19755. Simplifies `OAuthDiscoveryController` by dropping
the `authorization_endpoint → frontend base URL` branch that was there
to make `api.twenty.com/mcp` paste-able in MCP clients.
We've decided not to support pasting `api.twenty.com/mcp` — users can
paste `app.twenty.com/mcp`, `<workspace>.twenty.com/mcp`, or a custom
domain, all of which serve both frontend and API. On those hosts,
`authorization_endpoint` was already pointed at the same host as
`issuer`, which is what we want.
## Change
- Remove `isApiHost` helper and the `authorizeBase` branch — use
`issuer` for `authorization_endpoint`.
- Drop now-unused `TwentyConfigService` and `DomainServerConfigService`
injections.
- Drop duplicate `DomainServerConfigModule` import from
`application-oauth.module.ts` (the module is no longer needed).
Net diff: +1 / -22 across 2 files.
## Breaking change
MCP clients configured with `https://api.twenty.com/mcp` will stop
working. They should be reconfigured with the host matching the
workspace they're connecting to (`<workspace>.twenty.com/mcp`,
`app.twenty.com/mcp`, or a custom domain).
## Test plan
- [x] `yarn jest --testPathPatterns="mcp-auth.guard"` → 2/2 passing
(unchanged)
- [x] `tsc --noEmit` clean on modified files
- [ ] Manual verification on staging: `app.twenty.com/mcp` and
`<workspace>.twenty.com/mcp` OAuth flow still works end-to-end
Generated with [Claude Code](https://claude.com/claude-code)
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
## Summary
OAuth discovery metadata (RFC 9728 protected-resource, RFC 8414
authorization-server) and the MCP `WWW-Authenticate` header were
hardcoded to `SERVER_URL`. This breaks MCP clients that paste any URL
other than `api.twenty.com/mcp` — the metadata declares `resource:
https://api.twenty.com/mcp`, which doesn't match the URL the client
connected to, so the client rejects it and the OAuth flow never starts.
Reproduced with Claude's MCP integration: pasting
`<workspace>.twenty.com/mcp`, `app.twenty.com/mcp`, or a custom domain
returned *"Couldn't reach the MCP server"* because discovery returned a
resource URL for a different host.
Related memory: MCP clients POST to the URL the user entered, not the
discovered resource URL — so every paste-able hostname has to advertise
`resource` for that same hostname.
## What the server now does
`WorkspaceDomainsService.getValidatedRequestBaseUrl(req)` resolves the
canonical base URL for the host the request came in on, validated
against the set of hosts we actually serve:
- `SERVER_URL` (e.g. `api.twenty.com`) — API host
- default base URL (e.g. `app.twenty.com`) — the `DEFAULT_SUBDOMAIN`
base
- `FRONTEND_URL` bare host
- any `<workspace>.twenty.com` subdomain (DB lookup)
- any workspace `customDomain` where `isCustomDomainEnabled = true`
- any registered `publicDomain`
An unrecognized / spoofed Host falls back to
`DomainServerConfigService.getBaseUrl()`. **We never reflect arbitrary
Host values into the response.**
Callers updated:
- `OAuthDiscoveryController.getProtectedResourceMetadata` — echoes the
validated host into `resource` and `authorization_servers`.
- `OAuthDiscoveryController.getAuthorizationServerMetadata` — uses the
validated host for `issuer` and `*_endpoint`, **except**
`authorization_endpoint`: when the request came in via `SERVER_URL`
(API-only, no `/authorize` route), we keep that one pointed at the
default frontend base URL.
- `McpAuthGuard` — sets `WWW-Authenticate: Bearer
resource_metadata=\"<validatedBase>/.well-known/oauth-protected-resource\"`
on 401s, so the MCP client's follow-up discovery fetch lands on the same
host it started on.
## Security
- Workspace identity is already bound to the JWT via per-workspace
signing secrets (`jwtWrapperService.generateAppSecret(tokenType,
workspaceId)`). Host-aware discovery does not weaken that.
- Custom domains are only accepted once `isCustomDomainEnabled = true`
(i.e. after DNS verification), so an attacker can't register a
custom-domain mapping on a workspace and have discovery reflect it
before it's been proven.
- Unknown / spoofed Hosts fall through to the default base URL.
## Drive-by
Fixed a duplicate `DomainServerConfigModule` import in
`application-oauth.module.ts` while adding `WorkspaceDomainsModule`.
## Companion infra change required for custom domains
Customer custom domains (`crm.acme.com/mcp`) also require an
ingress-level fix to exclude `/mcp`, `/oauth`, and `/.well-known` from
the `/s\$uri` rewrite applied when `X-Twenty-Public-Domain: true`.
Shipping that in a twenty-infra PR (will cross-link here).
## Test plan
- [x] 14 new tests in
`WorkspaceDomainsService.getValidatedRequestBaseUrl` covering: missing
Host, SERVER_URL, base URL, FRONTEND_URL, workspace subdomain, unknown
subdomain fallback, enabled custom domain, disabled custom domain,
public domain, completely unrecognized host, lowercase coercion,
malformed Host, single-workspace mode fallback, DB throwing → fallback
- [x] New `oauth-discovery.controller.spec.ts` covering both endpoints
across api / app / workspace-subdomain / custom-domain hosts, plus
`cli_client_id` propagation
- [x] Rewrote `mcp-auth.guard.spec.ts` to cover `WWW-Authenticate` for
all four host types (api, workspace subdomain, custom domain, spoofed
fallback)
- [x] `yarn jest
--testPathPatterns=\"workspace-domains.service|oauth-discovery.controller|mcp-auth.guard\"`
→ 41/41 passing
- [x] `tsc --noEmit` clean on all modified files
- [ ] Manual verification against staging: connect Claude to
`api.twenty.com/mcp`, `app.twenty.com/mcp`,
`<workspace>.twenty.com/mcp`, and a custom domain and confirm OAuth flow
completes on each
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
- Add a `pageLayoutId` foreign key to `CommandMenuItem`, allowing
command menu items to be scoped to a specific page layout instead of
being globally available
- Filter command menu items by the current page layout on the frontend.
Items with a `pageLayoutId` only appear when viewing that layout, while
items without one remain globally visible
- Create an effect to track the current page layout ID
- Include a seed example: a "Show Notification" command pinned to the
Star history standalone page layout
---------
Co-authored-by: Charles Bochet <charles@twenty.com>
## Summary
- Move record fetching and payload building from the enrichment hook
into `TriggerWorkflowVersionEngineCommand`, following the same
component-based pattern as `DeleteRecordsCommand` and
`RestoreRecordsCommand`
- The enrichment hook now only stores workflow metadata (`trigger`,
`availabilityType`, `availabilityObjectMetadataId`); the component uses
`useLazyFetchAllRecords` for exclusion mode (Select All) with full
pagination
- `buildTriggerWorkflowVersionPayloads` is now a pure function accepting
`selectedRecords: ObjectRecord[]` instead of reading from the Jotai
store
Fixes the issue introduced by #19718 which blocked Select All with a
warning toast instead of implementing it.
## Test plan
- [ ] Select individual records → run workflow trigger from command menu
→ works as before
- [ ] Click Select All → run workflow trigger from command menu →
fetches all matching records and runs the workflow
- [ ] Select All with some records deselected → correctly excludes those
records
- [ ] Global workflows (no object context) → run without payload as
before
- [ ] Bulk record triggers → payload wraps records in `{namePlural:
[records]}`
🤖 Generated with [Claude Code](https://claude.com/claude-code)
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
## Summary
- `SettingsApplicationCustomTab` renders `FrontComponentRenderer` which
calls `useFrontComponentExecutionContext` →
`useLayoutRenderingContext()`, but the settings page never provided a
`LayoutRenderingProvider`
- Every other render site (side panel, command menu, record pages) wraps
`FrontComponentRenderer` with this provider — it was just missed here
- Opening the "Custom" tab in Settings → Applications crashes with:
`LayoutRenderingContext Context not found`
- Fix: wrap with `LayoutRenderingProvider` using `DASHBOARD` layout type
and no target record, matching the pattern used in
`SidePanelFrontComponentPage`
## Test plan
- [ ] Open Settings → Applications → any app with a custom settings tab
- [ ] Click the "Custom" tab — should render the front component without
crashing
🤖 Generated with [Claude Code](https://claude.com/claude-code)
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Co-authored-by: Charles Bochet <charles@twenty.com>
This regressed with the migration work
Type checker should've caught it, but didn't because of `Partial`
changed it to `Pick` instead to avoid future cases
/closes https://github.com/twentyhq/twenty/issues/19745
Add Storybook stories and example components to test event forwarding
through the front component renderer: form events (text input, checkbox,
focus/blur, submit), keyboard events (key/code/modifiers), and host API
calls (navigate, snackbar, progress, close panel)
Resolves the following two issues. The reason we had these issues was
because the maxLimit of renderers was reached and the browser started
losing context before restoring it. Now, we make sure the context that
is lost belongs to visuals that are not currently in the viewport and
when those visuals come back into viewport, they're loaded again.
https://github.com/twentyhq/core-team-issues/issues/2372https://github.com/twentyhq/core-team-issues/issues/2373
## Summary
- Bump `twenty-sdk` from `1.22.0-canary.6` to `1.22.0`
- Bump `twenty-client-sdk` from `1.22.0-canary.6` to `1.22.0`
- Bump `create-twenty-app` from `1.22.0-canary.6` to `1.22.0`
Fix isUnique not sent in field update mutation: Added isUnique and
settings to the Pick type in useUpdateOneFieldMetadataItem, which
previously silently dropped these properties from the GraphQL payload.
Invalidate cache on failed migration rollback: Added invalidateCache()
call in the migration runner's catch block so that a failed migration
(e.g., unique index creation failing due to duplicate data) regenerates
the Redis hash, preventing stale metadata from persisting in frontend
localStorage indefinitely. Was
## Summary
- import shared company logos and people avatars into
`packages/twenty-website-new/public/images/shared`
- expand the shared asset registry with the new local logo and avatar
paths
- update the home hero data to use local shared assets for matched
people and company icons
- replace synthetic or mismatched hero avatars with better-fitting named
or anonymous local assets
- prefer local shared company logos in hero visual components before
falling back to remote icons
## Testing
- Not run (not requested)
---------
Co-authored-by: Abdullah <125115953+mabdullahabaid@users.noreply.github.com>
## Summary
Route-triggered logic functions were returning empty 500 responses with
zero server-side logging when the Lambda build chain failed. This PR
makes those failures observable and returns meaningful HTTP responses to
API clients.
- **Observability** — Log errors (with stack traces) at each layer of
the execution chain: `LambdaDriver` (deps-layer fetch, SDK-layer fetch,
invocation), `LogicFunctionExecutorService`, and `RouteTriggerService`.
- **Typed exceptions** — Replace raw `throw error` sites with
`LogicFunctionException` carrying an appropriate code and
`userFriendlyMessage` (new codes: `LOGIC_FUNCTION_EXECUTION_FAILED`,
`LOGIC_FUNCTION_LAYER_BUILD_FAILED`).
- **Correct HTTP semantics** — `RouteTriggerService` maps inner
exception codes to the right `RouteTriggerExceptionCode` so
`LOGIC_FUNCTION_NOT_FOUND` returns 404 and `RATE_LIMIT_EXCEEDED` returns
429 (new code + filter case) instead of a generic 500.
- **User-facing messages** — Forward the inner
`CustomException.userFriendlyMessage` when wrapping into
`RouteTriggerException`, without leaking raw internal error text into
the public exception message.
- **Infra** — Bump Lambda ephemeral storage from 2048 to 4096 MB to
prevent `ENOSPC` errors during yarn install layer builds (root cause of
the original silent failures).
## Summary
Fixes error snackbars disappearing too quickly by **disabling
auto-dismiss for error variants by default**. Error snackbars now remain
visible until the user closes them.
Closes#19694
## What changed
- Error snackbars no longer default to a 6s timeout (they only
auto-dismiss if an explicit `duration` is provided).
- Non-error snackbars keep the existing default auto-dismiss behavior
(6s).
- Progress bar animation/visibility is tied to auto-dismiss (no progress
bar when there’s no duration).
**Files**
-
packages/twenty-front/src/modules/ui/feedback/snack-bar-manager/components/SnackBar.tsx
## How to test
1. Trigger a long error snackbar (example: spreadsheet/CSV import error
with a long message).
2. Confirm the snackbar **stays visible** until clicking **Close**.
3. Trigger a success/info snackbar and confirm it **still
auto-dismisses** after ~6s.
## Notes
- Call sites that explicitly pass `options.duration` for error snackbars
will continue to auto-dismiss (intentional).
---------
Co-authored-by: Charles Bochet <charles@twenty.com>
Summary
- Remove the evictViewMetadataForViewIds step from the page-layout reset
flow. It synchronously cleared viewFields/viewFieldGroups rows from the
metadata store, leaving a window where
useFieldsWidgetGroups saw a view with no fields and fell back to
buildDefaultFieldsWidgetGroups, briefly rendering a synthetic "General"
+ "Other" layout before the real reset defaults
arrived.
- invalidateMetadataStore() alone is sufficient: it marks the
collections stale and triggers MinimalMetadataLoadEffect to refetch,
which replaces current atomically. The UI now
transitions old-layout → new-default with no synthetic flash.
- Simplified refreshPageLayoutAfterReset to no longer take a
collectAffectedViewIds callback, and updated both tab/widget reset call
sites plus ObjectLayout.tsx accordingly.
- Deleted the now-unused evictViewMetadataForViewIds and
collectViewIdsFromWidgets utils.
## Summary
### Problem
The upgrade migration system required new workspaces to always start
from a workspace command, which was too rigid. When the system was
mid-upgrade within an instance command (IC) segment, workspace creation
would fail or produce inconsistent state.
### Solution
#### Workspace-scoped instance command rows
Instance commands now write upgrade migration rows for **all
active/suspended workspaces** alongside the global row. This means every
workspace has a complete migration history, including instance command
records.
- `InstanceCommandRunnerService` reloads `activeOrSuspendedWorkspaceIds`
immediately before writing records (both success and failure paths) to
mitigate race conditions with concurrent workspace creation.
- `recordUpgradeMigration` in `UpgradeMigrationService` accepts a
discriminated union over `status`, handles `error: unknown` formatting
internally, and writes global + workspace rows in batch.
#### Flexible initial cursor for new workspaces
`getInitialCursorForNewWorkspace` now accepts the last **attempted**
(not just completed) instance command with its status:
- If the IC is `completed` and the next step is a workspace segment →
cursor is set to the last WC of that segment (existing behavior).
- If the IC is `failed` or not the last of its segment → cursor is set
to that IC itself, preserving its status.
This allows workspaces to be created at any point during the upgrade
lifecycle, including mid-IC-segment and after IC failure.
#### Relaxed workspace segment validation
`validateWorkspaceCursorsAreInWorkspaceSegment` accepts workspaces whose
cursor is:
1. Within the current workspace segment, OR
2. At the immediately preceding instance command with `completed` status
(handles the `-w` single-workspace upgrade scenario).
Workspaces with cursors in a previous segment, ahead of the current
segment, or at a preceding IC with `failed` status are rejected.
### Test plan
created empty workspaces to allow testing upgrade with several active
workspaces
## Summary
- The `AddStandalonePageFastInstanceCommand` migration was failing with:
`default for column "type" cannot be cast automatically to type
core."navigationMenuItem_type_enum"`
- The migration was missing `DROP DEFAULT` / `SET DEFAULT` around the
`ALTER COLUMN TYPE` for `navigationMenuItem.type`. PostgreSQL cannot
automatically cast the existing default (`'VIEW'::old_enum`) to the new
enum type.
- The same 3-step pattern (DROP DEFAULT → ALTER TYPE → SET DEFAULT) was
already correctly applied for `pageLayout.type` in the same migration —
this fix brings `navigationMenuItem.type` in line.
- Add 72 missing HTML and SVG elements to the remote-dom component
registry (48 HTML + 24 SVG), bringing the total from 47 to 119 supported
elements
- HTML additions include semantic inline text (b, i, u, s, mark, sub,
sup, kbd, etc.), description lists, ruby annotations, structural
elements (figure, details, dialog), and form utilities (fieldset,
progress, meter, optgroup)
- SVG additions include containers (svg, g, defs), shapes (path, circle,
rect, line, polygon), text (text, tspan), gradients (linearGradient,
radialGradient, stop), and utilities (clipPath, mask, foreignObject,
marker)
- Add htmlTag override to support SVG elements with camelCase names
(e.g. clipPath, foreignObject) while keeping custom element tags
lowercase per the Web Components spec
## Summary
- Introduce a `activeNavigationMenuItemState` Jotai atom (persisted via
localStorage) to disambiguate active navigation items when multiple
items share the same URL
- Add active item evaluation for record show pages with three scenarios:
1. Navigating from a nav item → parent stays active + dedicated RECORD
item also active
2. Clicking a dedicated RECORD nav item → only that item active
3. Navigating via search/direct link → OBJECT nav item fallback, or
Opened section if none exists
- Extract shared active logic into `isNavigationMenuItemActive` utility
to eliminate duplication between orphan items and folder items
- Support multiple simultaneously active items within folders via
`Set<number>` instead of a single index
---------
Co-authored-by: Etienne <45695613+etiennejouan@users.noreply.github.com>
## Context
Tab duplication was broken after the view creation logic was deferred
and moved to the FE.
- Duplicating a tab or a FIELDS widget now produces a fully independent
copy: new view, new view field groups, new view fields — all with fresh
IDs — while preserving any unsaved edits
from the source widget.
- Removed the backend auto-seed of default view fields / view field
groups in ViewService.createOne for FIELDS_WIDGET views. The frontend
always sends the complete layout via
upsertFieldsWidget, so the auto-seed was both redundant and the source
of potential bugs.
- Extracted a shared useDuplicateFieldsWidgetForPageLayout hook used by
both tab and widget duplication paths, plus a small
useCloneViewInMetadataStore helper that clones the FlatView
in the metadata store and returns the copied flat view fields/groups for
the caller.
Fix a stack overflow (Maximum call stack size exceeded) in
getAllStepIdsInLoop caused by an If/Else branch inside an iterator loop
pointing back to the enclosing iterator. The traversal incorrectly
treated the enclosing iterator as a nested iterator, calling
getAllStepIdsInLoop recursively with fresh visited sets, causing
infinite recursion.
Add the enclosing iterator's own ID to the skip condition in
traverseSteps so back-edges from If/Else branches are handled the same
way as back-edges from regular nextStepIds.
<img width="1054" height="723" alt="Capture d’écran 2026-04-15 à 11 00
42"
src="https://github.com/user-attachments/assets/aee1477b-5059-4552-809e-7c8a34a9ec4a"
/>
## Summary
- replace static home card imagery with code-driven visuals for the
familiar interface, fast path, and live data cards
- refine the live data card interaction details, including hover states,
animated cursors, filter chips, table styling, and inline tag editing
- add shared company logos, people avatars, and updated partner
testimonial assets to support the new marketing visuals
- refresh related home, partner, case study, signoff, testimonials, and
design-system content/components to align with the updated marketing
presentation
## Testing
- `npx tsc -p tsconfig.json --noEmit` in `packages/twenty-website-new`
Co-authored-by: Abdullah <125115953+mabdullahabaid@users.noreply.github.com>
## Summary
Introduces a dedicated **metadata** mutation to update **standard
(non-custom)** workspace member settings, moves profile-related UI to
use it, and aligns **workspace member** record permissions with the rest
of the CRM so users cannot escalate visibility via RLS by editing their
own member record.
## Product behaviour
### Profile and appearance (standard fields)
- Users can still update **their own** standard workspace member fields
that the product exposes in **Settings / Profile** (e.g. name, locale,
color scheme, avatar flow) via the new
**`updateWorkspaceMemberSettings`** mutation.
- The mutation returns a **boolean**; the app **merges** the updated
fields into local state so the UI stays in sync without refetching the
full workspace member record.
- **Locale** changes also keep **`userWorkspace`** in sync when a locale
is present in the payload (including from the workspace `updateOne` path
when applicable).
### Custom fields on workspace members
- The dedicated metadata mutation **rejects** any **custom** workspace
member field (and unknown keys). Those updates must go through the
normal **object** `updateOne` pipeline, which is subject to **object-
and field-level** permissions like other records. But since we don't
have object- and field-level permission configuration for system objects
yet, this permission is derived from Workspace member settings
permission.
- **Workspace member** is no longer exempt from ORM permission
validation for updates merely because it is a **system** object. Users
who **do not** have workspace member access (e.g. no **Workspace
members** settings permission and no equivalent broad settings access on
the role) **cannot** use `updateOne` on `workspaceMember` to change
**custom** (or other) fields on their own row—even though that row is
used for RLS predicates.
- This closes a path where someone could widen what they can see by
writing to fields that drive row-level rules.
### Who can change another member
- Updating **another** user’s workspace member still requires
**Workspace members** (or equivalent) settings permission, consistent
with admin tooling.
## What changed
This refactor fixes AI chat error surfacing by aligning both the backend
and frontend with the existing GraphQL error architecture instead of
adding AI-local error translation.
On the backend:
- add a dedicated GraphQL billing exception path
- register billing GraphQL handling globally for GraphQL requests
- reuse the existing AI GraphQL interceptor path for agent/chat
exceptions
- keep billing status classification shared between REST and GraphQL
- remove the earlier attempt to preserve `CustomException` metadata in
the global GraphQL fallback
On the frontend:
- keep the original Apollo GraphQL error object in AI chat state
- reuse shared Apollo/GraphQL helpers for user-facing messages and
error-type checks
- delete AI-specific error extraction helpers that duplicated generic
GraphQL parsing
- replace a few direct `extensions.subCode` call sites with a shared
predicate
## Why it changed
The original bug was that `BillingException` and AI exceptions thrown
from chat were not being translated into GraphQL errors with the
expected `extensions.subCode` and `extensions.userFriendlyMessage`, so
the AI chat UI had nothing structured to inspect.
An intermediate fix worked mechanically but pushed `CustomException`
handling into the global GraphQL fallback, which blurred the intended
layering. This PR moves the behavior back to explicit GraphQL edges.
## Root cause
`AgentChatResolver` could throw `BillingException` and `AgentException`,
but:
- billing had a REST exception filter and no shared GraphQL equivalent
- AI chat was not consistently using the same GraphQL exception
translation path as the sibling AI resolver
- the frontend chat UI had drifted into AI-specific error parsing
instead of consuming the same structured Apollo errors as the rest of
the app
## Impact
- `BILLING_CREDITS_EXHAUSTED` is now preserved through GraphQL and can
render the existing credits-exhausted UI in chat
- `API_KEY_NOT_CONFIGURED` is preserved through the AI GraphQL path
- AI chat now follows the same general GraphQL error consumption pattern
as the rest of the frontend
- billing GraphQL handling is less dependent on individual resolver
authors remembering to add a filter
## Validation
- `yarn jest --config packages/twenty-server/jest.config.mjs
packages/twenty-server/src/engine/core-modules/billing/utils/__tests__/billing-graphql-api-exception-handler.util.spec.ts
packages/twenty-server/src/engine/metadata-modules/ai/ai-agent/utils/__tests__/agent-graphql-api-exception-handler.util.spec.ts`
- `yarn jest --config packages/twenty-front/jest.config.mjs
packages/twenty-front/src/utils/__tests__/is-graphql-error-of-type.util.test.ts`
- `npx oxlint --type-aware ...` on touched backend/frontend files
- `npx prettier --check ...` on touched backend/frontend files
## Follow-up ideas
- consolidate frontend GraphQL error helpers further so more existing
direct `extensions.subCode` checks move to shared utilities
- consider whether common GraphQL exception filter registration should
live in a more explicit GraphQL-specific module instead of
`CoreEngineModule`
- add an end-to-end test for a real `sendChatMessage` GraphQL failure
path in AI chat
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Add support for standalone pages: a new `PageLayout` type
(`STANDALONE_PAGE`) that can be rendered independently at
`/page/:pageLayoutId`, not tied to any record or object context.
- New `STANDALONE_PAGE` page layout type
- New `PAGE_LAYOUT` navigation menu item type: adds a `pageLayoutId`
foreign key to `NavigationMenuItemEntity`, allowing sidebar items to
link directly to standalone pages
- New `GLOBAL_OBJECT_CONTEXT` command menu availability type: separates
object-context-dependent commands (Create Record, Import, Export, See
Deleted, Create View, Hide Deleted) from truly global ones, so
standalone pages only show relevant commands
- Frontend routing & rendering: adds a `/page/:pageLayoutId` route with
its own page component, header, and command menu
- Widget rendering refactor
- Instance commands: two fast 1.22 migrations: `pageLayoutId` column +
`STANDALONE_PAGE` enum, and `GLOBAL_OBJECT_CONTEXT` availability type
enum
- Workspace command: backfills existing command menu items from `GLOBAL`
to `GLOBAL_OBJECT_CONTEXT` where appropriate
- Dev seeds: adds a sample "Star History" standalone page with an iframe
widget for local development
- Add resetPageLayoutToDefault GraphQL mutation that resets an entire
page layout (all tabs, widgets, view field groups, and view fields) to
their default state in a single operation
- Add a "Reset to default" button in the settings Layout tab
(/settings/objects/:object#layout) with a confirmation modal
<img width="673" height="426" alt="Screenshot 2026-04-14 at 13 13 53"
src="https://github.com/user-attachments/assets/002d33c9-9ea1-49f2-bef6-179ce034c126"
/>
## Context
Expending field widget supported types to all field types. They will all
by default fallback to "FIELD" displaymode which is the inline display
mode (same as the one used in FIELDS widget) and can be extended to
other displayMode such as CARD or EDITOR in the future if needed. Most
of the work was already done in the previous PR and was unnecessary
filter out until this was properly tested
<img width="951" height="757" alt="Screenshot 2026-04-14 at 13 24 24"
src="https://github.com/user-attachments/assets/a08a1855-21ea-4e75-8032-4f970b3ff50d"
/>
<img width="915" height="595" alt="Screenshot 2026-04-14 at 13 23 34"
src="https://github.com/user-attachments/assets/8b420c1f-0496-4c1c-91b8-b266c40b3772"
/>
## Summary
- extract and expose `street` from Google place details (`street_number`
+ `route`) on the server DTO
- request and type `street` in front-end geo-map place details query
- use `placeData.street` as the preferred value for `addressStreet1` in
address autofill
- add regression coverage for query fields and street-line precedence
behavior
## Why
Address autocomplete selection currently writes full place text
(including city/state/postcode/country) into `addressStreet1`,
duplicating values already mapped to dedicated fields.
Fixes#18860
---------
Signed-off-by: jeevan6996 <jeevanpawar5890@gmail.com>
Co-authored-by: Charles Bochet <charles@twenty.com>
## Context
Moving isActive filtering to the frontend for page layout tabs and
widgets, hiding inactive entities from the UI while keeping them in
state for future reactivation
Next we will implement deactivated standard tab re-activation during tab
creation (cc @Devessier)
<img width="234" height="303" alt="📋 Menu (Slots)"
src="https://github.com/user-attachments/assets/17a25ac6-55e2-4778-b7f0-e7554ed69704"
/>
fixes https://github.com/twentyhq/twenty/issues/19543
+ bonus bug : when deleting an advanced filter, it triggers a destroy
which cascade-deletes associated view filters. Then, view filters
deletion throws.
- Refactored prefillWorkflowCommandMenuItems and
prefillFrontComponentCommandMenuItems to use
validateBuildAndRunWorkspaceMigration instead of raw TypeORM
createQueryBuilder inserts
- This ensures the flat entity cache is properly updated when seeding
command menu items, fixing the Quick Lead item not appearing after
workspace creation
- Moved command menu item prefill calls outside the transaction since
they now go through the migration pipeline
# Introduction
We were allowing the sequence to be empty in the worker context that was
facing an edge case importing the UpgradeModule through the
WorkspaceModule god module, no commands were discovered and it was
throwing as the sequence must have at least one workspace commands to
allow a workspace creation
Though the issue was also applicable to the twenty-server `AppModule`
too that was not discovering any commands
## Integration tests were passing
The integration test were importing the `CommandModule` at the nest
testing app creating leading to asymmetric testing context
It was a requirement for a legacy commands import and global assignation
## Fix
The `UpgradeModule` now import both `WorkspaceCommandsProviderModule`
and `InstanceCommandProviderModule` which ships the commands directly in
the module
We could consider moving the commands into the `engine/upgrade` folder
## Concern
Bootstrap could become more and more long to load at both server and
worker start
When this becomes a problem we will have to only import the latest
workspace command or whatever
For the moment this is not worth it the risk to import not the latest
workspace command
## Summary
After uploading an image/file to the empty avatar field in the person
tab, the edit icon next to the field would not appear until the browser
was refreshed or another field was clicked.
### Root cause
- When user clicks over the avatar field,
`recordFieldListCellEditModePosition` is set to `globalIndex`
- That is fine when a avatar already exists. But when there is no avatar
already set, the native file picker is opened with no `onClose` handler
attached.
- So after the file upload is completed,
`recordFieldListCellEditModePosition` is never reset to null.
- `FieldsWidgetCellEditModePortal` stays anchored to the avatar file
element
- When the user hovers over the same field again, its hover portal tries
to compete to anchor for the same element
- So, `RecordInlineCellDisplayMode ` (the edit button) doesn't render
### Fix
- Pass the `onClose` function through `openFieldInput` to
`openFilesFieldInput`
- `onClose` resets `recordFieldListCellEditModePosition` back to null,
when the upload completes.
## Before
https://github.com/user-attachments/assets/ac9318e9-5471-434c-8af3-5c20d0112460
## After
https://github.com/user-attachments/assets/0d064a7f-95ad-4b92-a9ee-d9570f360972Fixes#19595
---------
Co-authored-by: Charles Bochet <charles@twenty.com>
## Summary
The `role` @ResolveField on `ApiKeyResolver` calls `getRolesByApiKeys`
with a single-element array per API key. When a query returns N API
keys, this produces N separate DB queries to resolve their roles.
This adds an `apiKeyRoleLoader` to the existing DataLoader
infrastructure. All API key IDs in a single GraphQL request are
collected and resolved in one batched query.
- Before: N queries (one per API key)
- After: 1 query (batched via DataLoader)
## Changes
- `dataloader.service.ts` - new `createApiKeyRoleLoader` method,
delegates to `ApiKeyRoleService.getRolesByApiKeys`
- `dataloader.interface.ts` - `apiKeyRoleLoader` added to `IDataloaders`
- `dataloader.module.ts` - import `ApiKeyModule` so `ApiKeyRoleService`
is available
- `api-key.resolver.ts` - `role()` now uses
`context.loaders.apiKeyRoleLoader.load()` instead of calling the service
directly
## Test plan
- [ ] Verify `apiKeys { id role { label } }` query returns the same
results as before
- [ ] Confirm only 1 role_target query fires regardless of how many API
keys are returned
---------
Co-authored-by: Charles Bochet <charles@twenty.com>
Summary
This PR fixes a bug in the `useColorScheme` test suite and removes a
lingering `FIXME` comment where the color scheme was unexpectedly
unsetting during state updates.
Root Cause
Previously, the Jotai state was being initialized *inside* the
`renderHook` callback using `useSetAtomState`. When the `setColorScheme`
function was called, it triggered a hook re-render, which caused the
callback to execute again and overwrite the new state with the hardcoded
`'System'` initial state.
The Fix
- Removed the state initialization from inside the render cycle.
- Bootstrapped the state on a fresh store using `resetJotaiStore()` and
`store.set()` *before* rendering the hook.
- Updated the mock `workspaceMember` to correctly use the
`CurrentWorkspaceMember` type.
- Removed the `FIXME` comment and successfully asserted that the color
scheme updates to `'Dark'`.
Testing
Ran tests locally to confirm the fix works as expected:
`corepack yarn jest --config packages/twenty-front/jest.config.mjs
--testPathPattern=useColorScheme.test.tsx`
---------
Co-authored-by: Srabani Ghorai <subhojit04ghorai@gmail.com>
## Summary
- refresh pricing page content, plan cards, CTA styling, and Salesforce
comparison visuals
- update partner and hero/testimonial visuals, including pulled
carousel-compatible partner testimonial data
- improve halftone export and illustration mounting flows, plus related
button and hydration fixes
- add updated website illustration and pricing assets
## Testing
- Not run (not requested)
---------
Co-authored-by: Abdullah <125115953+mabdullahabaid@users.noreply.github.com>
Handle late TwentyORM workspace-not-found exceptions in the workflow
webhook REST exception filter so deleted workspaces return a 404 instead
of surfacing as internal errors.
Also add a focused regression spec covering the deleted-workspace ORM
codes and the existing workflow-trigger status mappings.
Closes#15544
---------
Co-authored-by: Charles Bochet <charles@twenty.com>
## Summary
- Removes the `IS_RECORD_TABLE_WIDGET_ENABLED` feature flag, making the
record table widget unconditionally available in dashboard widget type
selection
- The flag was already seeded as `true` for all new workspaces and only
gated UI visibility in one component
(`SidePanelPageLayoutDashboardWidgetTypeSelect`)
- Cleans up the flag from `FeatureFlagKey` enum, dev seeder, and test
mocks
## Analysis
The flag only controlled whether the "View" (Record Table) widget option
appeared in the dashboard widget type selector. The entire record table
widget infrastructure (rendering, creation hooks, GraphQL types,
`RECORD_TABLE` enum in `WidgetType`) is independent of the flag and
fully implemented. No backend logic depends on this flag.
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Co-authored-by: Charles Bochet <charles@twenty.com>
## Summary
Implements a ClickHouse-backed polling system to enforce metered-credit
caps for workflow executions, replacing reliance on Stripe billing
alerts. The system re-evaluates tier caps against live pricing on every
poll cycle, allowing price/tier changes to propagate immediately without
recreating Stripe alert objects.
## Key Changes
- **BillingUsageCapService**: New service that queries ClickHouse for
current-period credit usage and evaluates whether a subscription has
reached its metered-credit allowance (tier cap + credit balance)
- `isClickHouseEnabled()`: Checks if ClickHouse is configured
- `getCurrentPeriodCreditsUsed()`: Sums creditsUsedMicro from usageEvent
table for a workspace within a billing period
- `evaluateCap()`: Determines if usage has reached the allowance by
reading live pricing from the subscription
- **EnforceUsageCapJob**: Cron job that polls all active subscriptions
and updates `hasReachedCurrentPeriodCap` on metered items
- Runs every 2 minutes to keep cap enforcement in sync with live usage
- Supports shadow mode (log-only) via
`BILLING_USAGE_CAP_CLICKHOUSE_ENABLED` flag for safe rollout
- Continues processing after per-subscription errors with detailed
logging
- **EnforceUsageCapCronCommand**: CLI command to register the
enforcement cron job
- **MeteredCreditService**: Extracted
`extractMeteredPricingInfoFromSubscription()` as a pure function for
callers that already hold the subscription with pricing loaded, avoiding
redundant DB queries
- **Configuration**: Added `BILLING_USAGE_CAP_CLICKHOUSE_ENABLED` flag
to control enforcement mode (active vs. shadow)
- **Constants**: Added `METERED_OPERATION_TYPES` to define which
operation types count toward the metered product's credit cap
## Implementation Details
- The service queries ClickHouse for the sum of `creditsUsedMicro` in
the current billing period, matching Stripe meter semantics
- Pricing is re-read on every evaluation, so tier changes propagate
within one poll cycle without Stripe alert recreation
- The cron job only updates the database when the cap state actually
changes (no-op if already in the correct state)
- Shadow mode allows safe validation before enabling enforcement;
transitions are logged but not persisted
- Comprehensive test coverage for both the service and cron job,
including error handling and state transitions
https://claude.ai/code/session_01VksTSrYLXJVCPVBQhQdBTe
---------
Co-authored-by: Claude <noreply@anthropic.com>
## Summary
- Removes the `workspaceId` column, `@Index()`, and `@ManyToOne`
workspace relation from `BillingSubscriptionItemEntity`
- The entity defined these fields but they don't exist in the actual
database table, causing `column X.workspaceId does not exist` errors at
runtime
- The workspace relationship is already accessible through the parent
`BillingSubscriptionEntity`
## PR description
- The command menu in the side panel now reads directly from
`MAIN_CONTEXT_STORE_INSTANCE_ID` instead of snapshotting the main
context store into a separate side-panel instance when opening. This
keeps the command menu always in sync with the current page state
(selection, filters, view, etc.).
- Removed the broadening/reset-to-selection feature (Backspace to clear
context, "Reset to" button) since the command menu no longer maintains
its own copy of the context.
## Video QA
https://github.com/user-attachments/assets/5d5bc664-b6d4-431d-a271-6ce23d8a4ae0
# Introduction
Index seemed to be missing in production only, as it's blocking the
whole migration release on other env that already implements the index
**Merge records fix:**
selectPriorityFieldValue throws when merging records if the priority
record has no value for a field (e.g., null/empty) but 2+ other records
do. The recordsWithValues array is pre-filtered to only records with
non-empty values, so the priority record isn't in the list. The fix:
instead of throwing, fall back to null since this is the priority record
actual value
**Duplicated IDs fix**
https://github.com/user-attachments/assets/bd6d7d08-d079-49a5-aad4-740b59a3c246
When applying a filter that reduces the record count, the virtualized
table's record ID array keeps stale entries from the previous larger
result set. loadRecordsToVirtualRows clones the old array (e.g., 60
entries) and only overwrites the first N positions (e.g., 9) with the
new filtered results, leaving positions 9-59 with old IDs. If any old ID
matches a new one, it appears twice in the selection, causing "-> 2
selected" for a single click and a duplicate ID in the merge mutation
payload. The fix: clear the record IDs array in
useTriggerInitialRecordTableDataLoad before repopulating it with fresh
data.
---------
Co-authored-by: Charles Bochet <charles@twenty.com>
## Introduction
As the new upgrade sequence engine is released in `1.22` it requires all
workspaces to be in `1.21.0` which mean they will have a cursor on the
sequence
As if if someone upgrades from `1.20` to `1.22` no `upgradeMigration`
will exist and throw a pretty basic `Could not find any cursor, database
might not been initialized correctly`
Here we allow a meaningful error
Cleans up the code quality by migrating from Raw SQL to TypeORM
entities. The previous implementation was necessary to do cross‑schema
table joins but since we've migrated to the core schema we don't need it
anymore.
- Also extracted `toIsoStringOrNull` to a utility it was duplicated
several times
- Moved `isThrottled` logic from job handler to cron enqueuer
## Introduction
In the same validate build and run we should be able to delete a view
field targetting a label identifier and at the same create one that
repoints to it again without failing any validation
Leading for this valdiation rule to be moved in the cross entity
validation steps
## Summary
- refresh the partner hero visual and testimonial presentation,
including the partner-specific carousel and illustration assets
- switch the testimonials top notch to the masked rendering approach
used elsewhere for more precise shape control
- extend halftone studio/export support and related geometry/state
handling used by the updated partner visuals
- include supporting website UI adjustments across navigation, pricing,
plans, and Salesforce-related sections
## Testing
- Not run (not requested)
This PR introduces more updates to the website, such as real
testimonials, case studies, copy of pricing plans table. It also adds
modals for "Talk to Us" and "Become a Partner".
## Summary
- fix the halftone studio image-switch behavior so image mode uses a
sane default preview distance instead of rendering nearly off-screen
- add shared preview-distance handling for shape and image modes, and
tune the default 3D idle auto-rotate speed
- update halftone controls/export plumbing to support the latest studio
settings changes
- refresh website UI/content in pricing, Salesforce, menu, and
billing-related sections
## Testing
- Ran targeted Jest tests for halftone state and footprint logic
- Ran TypeScript check for `packages/twenty-website-new`
- Broader app-level/manual testing not run
## Summary
- Replace per-file `setupFiles` + manual
`appBuild`/`appDeploy`/`appInstall` with vitest `globalSetup` that runs
`appDevOnce` once for the entire suite and `appUninstall` in teardown
- Add `fileParallelism: false` to prevent shared-state collisions
between test files
- Replace `app-install.integration-test.ts` with
`schema.integration-test.ts` that verifies app installation, custom
object schema (fields/relations), and CRUD
- Add reusable test helpers (`client.ts`, `metadata.ts`, `mutations.ts`)
- Applied to both `create-twenty-app` template and `postcard` example
# Introduction
Refactoring the upgrade engine to handle cross version upgrade,
completely getting rid of the semver `version` at db and runtime level
It remains a visual a listing indicator for or CD process but also
during devenv in order to prepare next release
Will write a release process runbook documentation on how to handle
upgrade step patch, command insertion etc as it needs to be cascaded
across all the involved supported version
**The upgrade sequence model:**
The sequence is a flat, ordered array of upgrade steps
(`UpgradeStep[]`), built from the registry by chaining all versions in
order, each version contributing its fast-instance → slow-instance →
workspace commands sorted by timestamp. Version is metadata for logging,
not used in the algorithm.
**Segments:**
The sequence naturally splits into alternating segments of contiguous
instance steps and contiguous workspace steps. The runner processes
segments in order:
- **Instance segment:** Run sequentially from the instance cursor. Each
step runs once globally.
- **Workspace segment:** Each workspace independently walks from its own
cursor through the end of the segment. Workspaces are independent within
a segment — they can be at different positions.
- **Synchronization (workspace → instance):** The runner blocks before
entering an instance segment. All active/suspended workspaces must have
completed the last workspace step of the preceding workspace segment. If
any workspace failed, abort. This is the only explicit synchronization
point.
- Instance → workspace ordering is implicit — the runner processes
segments sequentially, so the instance segment naturally completes
before the workspace segment begins.
full docs
https://gist.github.com/prastoin/e62106d455fd72d6b6ebada8351e5492
## Version constants & type-level deprecation
Version management is split into three atomic constants:
`TWENTY_PREVIOUS_VERSIONS`, `TWENTY_CURRENT_VERSION`, and
`TWENTY_NEXT_VERSIONS`. Two derived constants compose them:
`CROSS_UPGRADE_SUPPORTED_VERSIONS` (previous + current — what the engine
runs) and `ALL_TWENTY_VERSIONS` (the full ordered tuple including next).
The registry service validates at module init that no version is
duplicated across constants and that at least one previous version
exists.
A `DeprecatedSinceVersion<RemoveAtVersion, T>` type utility resolves to
`T` while `TWENTY_CURRENT_VERSION` is below `RemoveAtVersion`, and to
`never` once it reaches it — turning deprecation into a compile-time
guarantee via `IndexOf` and `IsGreaterOrEqual` generics in
`twenty-shared`.
### `workspace.version` column deprecation
The column is replaced by cursor-based state inference from
`UpgradeMigration` records, but cannot be dropped in 1.22: workspaces
activated during 1.21 predate the cursor system and need their initial
cursor backfilled first (`backfillWorkspaceCreatedIn1_21_0Cursors`).
This backfill itself depends on a new `isInitial` column on
`UpgradeMigration`, bootstrapped via a targeted TypeORM migration before
the upgrade sequence runs.
Both functions and the entity field are typed with
`DeprecatedSinceVersion<'1.23.0', ...>`. When `TWENTY_CURRENT_VERSION`
reaches `1.23.0`, compile errors force their removal — and the
pre-declared `DropWorkspaceVersionColumnFastInstanceCommand` takes over
to drop the column.
## What's next
- ci cross version upgrade ( wip )
- banner asking to contact twenty administrator if workspace is outdated
- upgrade healthcheck cli
## New unit/integ test pattern
Create a dedicated `createNestApp` that consumes a real database in
order not to have to mack any database interaction to the
`upgradeMigrations` allowing full coverage of the whole
`upgradeRunnerService.run` core logic
## Introduction
View standard backfill seed has been introduced during a partial patch,
a window exists where some workspaces got created without the view and
wasn't included in the upgrade process
The fix isn't covering all the view that has been missed during that
time, only the one required for the command to pass
## Overview
Adds comprehensive admin panel functionality for viewing workspace
details and AI chat threads.
## Changes
### Frontend
- **New Routes**: Added `AdminPanelWorkspaceDetail` and
`AdminPanelWorkspaceChatThread` pages with lazy loading
- **New Queries**:
- `getAdminWorkspaceChatThreads` - fetch chat threads for a workspace
- `getAdminChatThreadMessages` - fetch messages for a specific thread
- `workspaceLookupAdminPanel` - lookup workspace info and users
- **New Components**:
- `SettingsAdminWorkspaceDetail` - displays workspace info and chat
sessions tabs
- `SettingsAdminWorkspaceChatThread` - renders chat conversation with
message bubbles
- **Navigation**: Updated AI admin panel to link to workspace detail
pages
- **Settings Paths**: Added `AdminPanelWorkspaceDetail` and
`AdminPanelWorkspaceChatThread` paths
### Backend
- **New DTOs**:
- `AdminWorkspaceChatThreadDTO` - workspace chat thread data
- `AdminChatThreadMessagesDTO` - thread with messages
- `AdminChatMessageDTO` - individual message with parts
- **New Resolvers**: Added three queries to `AdminPanelResolver`
- **New Service Methods**:
- `workspaceLookup()` - fetch workspace info
- `getWorkspaceChatThreads()` - list chat threads
- `getChatThreadMessages()` - fetch thread messages with validation
- **Module Updates**: Added entity imports for workspace, user, AI chat,
and feature flag data
### Security
- Added `allowImpersonation` check before accessing chat data
- Validates workspace ownership and access permissions
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
fixes: #19571
## What's happening
BlockNote's `.bn-panel` (the Upload/Embed popup for images, videos,
audio) has a hardcoded `width: 500px` in the upstream
`blocknoteStyles.css`. When you open it inside the side panel (~350px
wide), it overflows past the container's `overflow: hidden` and gets
clipped.
## Why percentage-based fixes don't work here
I initially tried `max-width: 100%` on `.bn-panel` but it doesn't work
the panel sits inside a Floating UI popover that's absolutely positioned
with no explicit width. Its width comes from its content (the 500px
panel), so `max-width: 100%` just resolves to that same 500px. It's a
circular reference.
## The fix
```css
& .bn-mantine {
container-type: inline-size;
}
& .bn-mantine .bn-panel {
width: min(500px, 100cqi);
box-sizing: border-box;
}
```
container-type: inline-size on .bn-mantine lets us reference the
editor's actual width using cqi units. min(500px, 100cqi) keeps the
panel at 500px in wide containers (main note view) and shrinks it to fit
in narrow ones (side panel).
I scoped container-type to .bn-mantine (BlockNote's own root wrapper)
rather than the outer StyledEditor div, so the containment context stays
within BlockNote's own tree.
#Before
<img width="373" height="530" alt="image"
src="https://github.com/user-attachments/assets/63fb407c-79d1-4faa-afb4-15a98fe4ba22"
/>
#After
<img width="332" height="597" alt="image"
src="https://github.com/user-attachments/assets/f7e91c28-ad35-44a5-9450-0b6d52714977"
/>
<img width="599" height="579" alt="image"
src="https://github.com/user-attachments/assets/20fd457b-2d06-4799-9164-f993ef37d833"
/>
Tested
Side panel: panel fits correctly, matches "Add image" button width
Full-width editor: panel stays at 500px
Slash menu, formatting toolbar, drag handles: all unaffected
---------
Co-authored-by: Félix Malfait <felix.malfait@gmail.com>
- Defer backend view creation for FIELDS widgets to prevent orphan views
when users cancel or leave before saving. Views are now created
optimistically in Jotai state (client-side UUID) and only persisted to
the database when the page layout is saved.
- Align the frontend's default field groups fallback with the backend
logic: split into General/Other groups, hide relation fields by default,
and filter invisible fields consistently across all code paths.
## Summary
- Follows up on #19610 which exported view/page-layout types but missed
field settings types
- Adds re-exports for `DateDisplayFormat`, `NumberDataType`, and
`FieldMetadataSettingsOnClickAction` from the SDK public API
- These enums are referenced in the return type of `defineObject` (via
field settings types like `FieldMetadataSettingsDate`,
`FieldMetadataSettingsNumber`, `FieldMetadataSettingsMultiItem`) but
were not publicly exported
- This causes TS4082 ("private name") errors for SDK consumers that have
`declaration: true` in their tsconfig, specifically when using
`defineObject` with fields that have settings (e.g. SELECT, DATE, NUMBER
fields)
## Summary
- Fixes TS4082 errors for SDK consumers with `declaration: true` in
their tsconfig
- The `define*` functions (e.g. `defineView`, `definePageLayout`,
`defineLogicFunction`) return `ValidationResult<T>` where `T` references
types from `twenty-shared` that were not re-exported from the SDK's
public barrel
- TypeScript cannot generate `.d.ts` files when the return type
references "private names" — types that exist in the package but aren't
publicly exported
### Added re-exports
**Enums** (used in `ViewManifest`, `HttpRouteTriggerSettings`):
- `ViewType`, `ViewFilterOperand`, `ViewFilterGroupLogicalOperator`,
`ViewOpenRecordIn`, `ViewVisibility`
- `HTTPMethod`
**Types** (used in `PageLayoutWidgetManifest`, `LogicFunctionConfig`):
- `GridPosition`, `PageLayoutWidgetConditionalDisplay`
- `InputJsonSchema`
## Summary
- Adds `isUnique?: boolean` to `RegularFieldManifest` in
`twenty-shared`, allowing SDK applications to declare unique constraints
on fields
- Updates the manifest-to-flat-field converter to read `isUnique` from
the manifest instead of hardcoding `false`
- Generates corresponding unique index metadata in
`computeApplicationManifestAllUniversalFlatEntityMaps` when a field has
`isUnique: true`, matching the behavior of the `CreateFieldInput` path
- Adds SDK-side validation rejecting `isUnique` on RELATION,
MORPH_RELATION, and FILES field types
- Adds integration test verifying manifest sync creates a unique index
for `isUnique` fields
- Adds SDK unit tests for `isUnique` validation on unsupported field
types
## Test plan
- [x] SDK unit tests: `defineField` accepts `isUnique: true` on TEXT,
rejects on RELATION and FILES
- [ ] Integration test: manifest sync with `isUnique: true` creates the
unique index in DB
- [ ] Verify `isUnique` defaults to `false` when not specified (backward
compatible)
- [ ] Verify standalone manifest fields (not nested in objects) also
generate unique indexes correctly
## Summary
- refine `/halftone` controls for material selection and slower 3D
rotation tuning
- switch the default halftone material to solid for new sessions
- include related halftone rendering, export, and glass material updates
across the website app
## Testing
- `yarn nx run twenty-website-new:typecheck`
- `yarn prettier
packages/twenty-website-new/src/app/halftone/_components/controls/AnimationsTab.tsx
--check`
- `yarn prettier
packages/twenty-website-new/src/app/halftone/_components/controls/controls-ui.tsx
packages/twenty-website-new/src/app/halftone/_components/controls/DesignTab.tsx
--check`
- `yarn prettier
packages/twenty-website-new/src/app/halftone/_lib/state.ts --check`
## Summary
- update pricing card styling to use a 4px border radius
- make pricing card CTAs fill the available card width
- reduce spacing below the pricing cards section
- fix pricing copy casing and punctuation, including the tailor-made
plan banner
- make engagement band headings support a configurable size and default
them to the smaller variant
## Testing
- Not run (not requested)
## Summary
- After a successful `uninstall`, clear all app registration config
(`appRegistrationId`, `appRegistrationClientId`, `appAccessToken`,
`appRefreshToken`) so the next `dev` run doesn't hit "app not found"
errors from leftover credentials
- On app re-registration (`ensureAppRegistration`), clear stale
`appAccessToken`/`appRefreshToken` that belonged to the previous
registration
- On OAuth token refresh failure, only clear config when the server
returns `invalid_client` (registration was deleted), not on transient
failures like network issues
- Extract a shared `parse-server-error` utility for consistently parsing
both GraphQL error codes (`extensions.code`/`subCode`) and OAuth error
responses (`error`/`error_description`)
## Summary
- `twenty-shared` is private and never published to npm, so SDK
consumers couldn't resolve type imports like `from
'twenty-shared/types'` in the generated `.d.ts` files
- Replace `vite.config.sdk.ts` with `rollup-plugin-dts` which compiles
`src/sdk/index.ts` directly into a self-contained `dist/sdk/index.d.ts`
with all `twenty-shared` types inlined
- The JS output from `vite.config.sdk.ts` (`dist/sdk/*.js`) was unused —
the main export already maps to `dist/index.mjs` from the node Vite
config
## Changes
- **Deleted** `vite.config.sdk.ts` — its preserved-module JS output
wasn't referenced by any `package.json` export
- **Added** `rollup.config.dts.mjs` — uses `rollup-plugin-dts` to
compile SDK types from source with `twenty-shared` inlined (~850ms)
- **Updated** `project.json` — build/dev/build:sdk targets now use
rollup instead of the removed vite config
- **Updated** `tsconfig.json` — removed `vite.config.sdk.ts` from
include
## Test plan
- [ ] Run `npx nx build twenty-sdk` and verify `dist/sdk/index.d.ts`
contains no `twenty-shared` references
- [ ] Verify `dist/index.mjs` and `dist/index.cjs` are still produced
correctly
- [ ] Verify CLI (`dist/cli.cjs`) still works
- [ ] Verify `npx nx build:sdk twenty-sdk` works standalone
Made with [Cursor](https://cursor.com)
## Summary
- Fix `syncApplication` crashing with `ENTITY_NOT_FOUND` when a
navigation menu item child (with `folderUniversalIdentifier`) appears
before its folder in the manifest's `navigationMenuItems` array
- Add a generic topological sort in
`WorkspaceEntityMigrationBuilderService.validateAndBuild()` that detects
self-referential FKs from `ALL_MANY_TO_ONE_METADATA_RELATIONS` and
ensures parents are created before children
- This also covers other self-referential entities:
`viewFilterGroup.parentViewFilterGroup`,
`fieldMetadata.relationTargetFieldMetadata`, and
`rowLevelPermissionPredicateGroup.parentRowLevelPermissionPredicateGroup`
## Root cause
The builder iterated `createdFlatEntityMaps.byUniversalIdentifier` in
insertion order (from the manifest). Validation passed because it
checked both optimistic maps and remaining-to-create maps. But the
runner processed create actions sequentially, so
`resolveUniversalRelationIdentifiersToIds` threw when the folder hadn't
been created yet.
## Summary
- Enable application log console output in the `twenty-app-dev` Docker
image by adding `APPLICATION_LOG_DRIVER=CONSOLE` to the Dockerfile ENV
block
- Rename `APPLICATION_LOG_DRIVER_TYPE` to `APPLICATION_LOG_DRIVER` and
`ApplicationLogDriverType` to `ApplicationLogDriver` for consistency
with all other driver config variables (`EMAIL_DRIVER`, `LOGGER_DRIVER`,
`CAPTCHA_DRIVER`, etc.)
- Add `APPLICATION_LOG_DRIVER` to `.env.example`
## Summary
Refactored AI chat state management to use component family states
instead of global atoms. This change enables proper state isolation per
chat thread, preventing state leakage between different chat instances.
## Key Changes
- **Converted global atoms to component family states:**
- `agentChatErrorState` → `agentChatErrorComponentFamilyState`
- `agentChatIsStreamingState` →
`agentChatIsStreamingComponentFamilyState`
- `agentChatFirstLiveSeqState` →
`agentChatFirstLiveSeqComponentFamilyState`
- `agentChatHandleEventCallbackState` →
`agentChatHandleEventCallbackComponentFamilyState`
- `agentChatUsageState` → `agentChatUsageComponentFamilyState`
- `currentAIChatThreadTitleState` →
`currentAIChatThreadTitleComponentFamilyState`
- **Updated state access patterns:**
- Introduced `useAtomComponentFamilyStateCallbackState` hook to get
family state callbacks
- Changed from direct atom access to family key-based access using `{
threadId }` as the family key
- Updated all components and hooks to use the new family state patterns
- **Removed instance ID dependency:**
- Eliminated `AGENT_CHAT_INSTANCE_ID` constant usage
- Simplified state access by using thread ID directly as the family key
- **Updated affected components and hooks:**
- `useAgentChatSubscription`: Now creates family state atoms per thread
- `AgentChatMessagesFetchEffect`: Uses family state callbacks for event
handling
- `AgentChatThreadInitializationEffect`: Sets family state values per
thread
- `useAIChatThreadClick`: Updates family states when switching threads
- `AIChatErrorUnderMessageList`, `AIChatLastMessageWithStreamingState`,
`AIChatEmptyState`, `AIChatStandaloneError`, `SendMessageButton`,
`AIChatContextUsageButton`, `SidePanelAskAIInfo`: Updated to use family
state values with thread ID
## Implementation Details
- All new component family states use
`AgentChatComponentInstanceContext` for proper component-level isolation
- Family key structure: `{ threadId: string | null }`
- Removed `disposed` flag logic as it's no longer needed with proper
state isolation
- Updated dependency arrays in hooks to include family callback
functions
https://claude.ai/code/session_01D8syiJtCXu5bEHSr5KYkCt
---------
Co-authored-by: Claude <noreply@anthropic.com>
## Summary
This PR enhances the handling of sensitive configuration variables by
improving masking logic and user experience when editing them. It adds
metadata-aware masking for dynamically marked sensitive variables and
clears sensitive values when entering edit mode.
## Key Changes
- **Backend masking improvements**: Updated `maskSensitiveValue()` to
accept metadata parameter, enabling masking of variables marked as
sensitive via metadata (not just predefined masking config). Sensitive
non-string values are masked as `********`.
- **Edit mode UX**: When editing a sensitive variable, the value field
is now cleared on entering edit mode to prevent exposing masked values
and ensure users intentionally provide new secret values.
- **Form state tracking**: Enhanced `useConfigVariableForm()` hook to
accept an `isEditing` parameter and properly track value changes for
sensitive variables during edit operations.
- **Input placeholder**: Updated placeholder text for sensitive variable
inputs to show `Enter a new secret value` instead of the generic
database storage message, providing clearer intent to users.
## Implementation Details
- The `maskSensitiveValue()` method now checks both predefined masking
configurations and runtime metadata to determine if a value should be
masked
- Sensitive string values use LAST_N_CHARS strategy (4 characters),
while non-string values are masked uniformly
- The form's `hasValueChanged` flag is set to true when editing
sensitive variables to ensure proper validation and submission handling
https://claude.ai/code/session_01JiTckmuVJMQWpJ7TUGwsqb
---------
Co-authored-by: Claude <noreply@anthropic.com>
## Summary
- Bump `twenty-sdk`, `twenty-client-sdk`, and `create-twenty-app` from
`1.22.0-canary.2` to `1.22.0-canary.3` for publishing.
## Test plan
- Version-only change, no code modifications.
Made with [Cursor](https://cursor.com)
## Summary
- The `sdk-e2e-test` CI job has been failing since at least April 9th
because the nx dependency graph runs `database:reset` and
`start:ci-if-needed` in **parallel** — neither depends on the other. The
server starts before the DB tables are created, crashes with `relation
"core.keyValuePair" does not exist`, and `wait-on` times out after 10
minutes.
- Replace the `nx-affected` orchestration for e2e with explicit
**sequential** CI steps: build → create DB → reset DB → start server →
wait for health → run tests.
- Add server log dump on failure for easier future debugging.
## Root cause
In `packages/twenty-sdk/project.json`, the `test:e2e` target has:
```json
"dependsOn": [
"build",
{ "target": "database:reset", "projects": "twenty-server" },
{ "target": "start:ci-if-needed", "projects": "twenty-server" }
]
```
Since `database:reset` and `start:ci-if-needed` don't depend on each
other, nx can (and does) run them concurrently. `start:ci-if-needed`
fires `nohup nest start &` and immediately completes. The server process
tries to query `core.keyValuePair` before `database:reset` creates it →
crash → wait-on timeout → job failure.
## Summary
Enhanced the documentation and user guidance for null/empty value
filters across all field types by updating the `NullCheckEnum`
descriptions in the field filter Zod schemas. The descriptions now
provide clear, actionable guidance on how to use the "NULL" and
"NOT_NULL" filter options.
## Changes
- Updated 20+ field type filter descriptions to replace generic "Is null
or not null" text with more descriptive guidance
- New descriptions follow the pattern: "Check for missing or empty
[field type]. Use "NULL" to find records with no [value], "NOT_NULL" for
records with a [value]"
- Applied consistent messaging across all field types including:
- Basic types: UUID, text, number, boolean, date
- Complex types: select, multi-select, rating
- Composite types: amount, currency, first name, last name, street,
city, country, email, phone, link URL
- Relationship and JSON fields
- Specialized descriptions for composite fields (e.g., "Check for
missing or empty amount" for currency amount field)
## Benefits
- Improved API documentation clarity for developers using field filters
- Reduced ambiguity about NULL vs NOT_NULL filter behavior
- Better discoverability of filtering capabilities through schema
descriptions
- Consistent messaging across all field types for improved user
experience
https://claude.ai/code/session_0139Nb5bZykacNE69GQsFSrr
Co-authored-by: Claude <noreply@anthropic.com>
## Summary
- **Fix `createApplicationRegistration` flow**: The server's
`createApplicationRegistration` mutation returns a `clientSecret`, not
`accessToken`/`refreshToken` directly. The SDK now correctly requests
`clientSecret` and immediately performs an OAuth `client_credentials`
exchange to obtain `appAccessToken` and `appRefreshToken`, then stores
them in config.
- **New `exchangeCredentialsForTokens` helper**: Shared by both `dev`
and `dev --once` flows. Takes `clientId` + `clientSecret`, calls
`/oauth/token` with `client_credentials` grant, and persists the
resulting tokens.
- **Bump `twenty-sdk`, `twenty-client-sdk`, `create-twenty-app` to
`1.22.0-canary.2`**
## Context
The `1.22.0-canary.1` SDK release expected
`createApplicationRegistration` to return `accessToken`/`refreshToken`
directly, but the `v1.22.0` server returns `clientSecret`. This caused
`yarn twenty dev` and `yarn twenty dev --once` to fail with "No
registration found" errors.
## Summary
- Add Playwright inspection scripts for problem canvas export, chain
logging, and metrics capture
- Save updated markdown snapshots and screenshot artifacts for homepage
and problem section debugging
- Include generated browser profile data used during the investigation
## Testing
- Not run (not requested)
## Summary
- Bumps `twenty-sdk`, `twenty-client-sdk`, and `create-twenty-app`
package versions from `0.9.0` to `1.22.0-canary.1` for the 1.22 canary
release.
## Test plan
- [ ] Verify packages build successfully (`npx nx build twenty-sdk`,
`npx nx build twenty-client-sdk`, `npx nx build create-twenty-app`)
- [ ] Verify `create-twenty-app` scaffolds new apps with the correct SDK
version
Made with [Cursor](https://cursor.com)
## Summary
- **SDK (`dev` & `dev --once`)**: After app registration, the CLI now
obtains an `APPLICATION_ACCESS` token via `client_credentials` grant
using the app's own `clientId`/`clientSecret`, and uses that token for
CoreApiClient schema introspection — instead of the user's
`config.accessToken` which returns the full unscoped schema.
- **Config**: `oauthClientSecret` is now persisted alongside
`oauthClientId` in `~/.twenty/config.json` when creating a new app
registration, so subsequent `dev`/`dev --once` runs can obtain fresh app
tokens without re-registration.
- **CI action**: `spawn-twenty-app-dev-test` now outputs a proper
`API_KEY` JWT (signed with the seeded dev workspace secret) instead of
the previous hardcoded `ACCESS` token — giving consumers a real API key
rather than a user session token.
## Motivation
When developing Twenty apps, `yarn twenty dev` was using the CLI user's
OAuth token for GraphQL schema introspection during CoreApiClient
generation. This token (type `ACCESS`) has no `applicationId` claim, so
the server returns the **full workspace schema** — including all objects
— rather than the scoped schema the app should see at runtime (filtered
by `applicationId`).
This caused a discrepancy: the generated CoreApiClient contained fields
the app couldn't actually query at runtime with its `APPLICATION_ACCESS`
token.
By switching to `client_credentials` grant, the SDK now introspects with
the same token type the app will use in production, ensuring the
generated client accurately reflects the app's runtime capabilities.
## Summary
This PR removes the `IS_USAGE_ANALYTICS_ENABLED` feature flag and makes
usage analytics features universally available. The feature flag guard
has been removed from the usage analytics resolver and all conditional
rendering based on this flag has been eliminated.
## Key Changes
- **Removed feature flag dependency**: Deleted
`IS_USAGE_ANALYTICS_ENABLED` from the `FeatureFlagKey` enum in
`twenty-shared`
- **Updated AI Usage tab**: Simplified `SettingsAIUsageTab` to remove
enterprise access checks and feature flag conditionals, now only checks
if ClickHouse is configured
- **Updated Usage Analytics section**: Removed feature flag guard from
`SettingsUsageAnalyticsSection` and added loading/empty state handling
- **Updated AI settings navigation**: Made the Usage tab always visible
in the AI settings tabs, removing conditional rendering based on feature
flag
- **Updated Billing Credits section**: Removed feature flag check before
showing the "View usage" button
- **Updated Settings routes**: Removed `SettingsProtectedRouteWrapper`
with feature flag requirement from usage routes
- **Updated GraphQL resolver**: Removed `@RequireFeatureFlag` decorator
and `FeatureFlagGuard` from the `getUsageAnalytics` query
- **Updated dev seeder**: Removed the feature flag seed entry for
`IS_USAGE_ANALYTICS_ENABLED`
## Implementation Details
- Usage analytics now gracefully handles loading states with
`UsageSectionSkeleton`
- Empty state messaging is shown when no usage data is available yet
- ClickHouse configuration remains the only requirement for usage
analytics functionality
- All enterprise-specific gating for AI usage analytics has been removed
in favor of ClickHouse availability checks
https://claude.ai/code/session_01MRFVXtquL3wS7qmQkDU3AT
---------
Co-authored-by: Claude <noreply@anthropic.com>
## Summary
Fixes the side panel close animation cleanup
(`sidePanelCloseAnimationCompleteCleanup`) never running during the
close transition.
This eventually fixes the issue where in navbar edit mode, clicking on a
nav item sometimes doesn't get selected, the side panel opens but shows
`Select a navigation item to edit` instead of the selected item's
settings. The root cause was stale `isSidePanelClosing` state from a
previous close that wasn't cleaned up, causing the next open to run
cleanup synchronously (without emitting the close event), which
interfered with the navigation item selection flow.
### Root cause
The CSS `transition` on `StyledSidePanelWrapper` used
`${themeCssVariables.animation.duration.normal}s`, which Linaria
converts to a CSS custom property value like `width
var(--t-animation-duration-normal)s`. After CSS variable substitution,
`0.3` and `s` become two separate tokens, not a valid `<time>` dimension
(`0.3s`). The browser rejects the entire transition declaration, falls
back to `all 0s`, and the width change happens instantly with no
`transitionend` event.
### Fix
- Use `calc(var(--t-animation-duration-normal) * 1s)` to properly
construct the `<time>` value (consistent with ~30 other usages in the
codebase).
- Filter `handleTransitionEnd` to only process `width` transitions on
the wrapper element itself, ignoring bubbled child `background-color`
transitions.
## Before
https://github.com/user-attachments/assets/496ccbff-dc96-487d-a994-451dbf2a5165
## After
https://github.com/user-attachments/assets/78255240-1d7d-42cd-9b56-25627613883a
## Summary
This PR refactors the AI chat thread state management to use `null`
instead of a sentinel string value (`AGENT_CHAT_UNKNOWN_THREAD_ID`) to
represent an uninitialized or new chat thread. This improves type safety
and makes the code more idiomatic by using `null` to represent the
absence of a value.
## Key Changes
- **Removed sentinel constant**: Deleted `AGENT_CHAT_UNKNOWN_THREAD_ID`
constant and replaced all usages with `null`
- **Updated state types**: Changed `currentAIChatThreadState`,
`agentChatLastDiffSyncedThreadState`, and
`agentChatDisplayedThreadState` to use `string | null` type with `null`
as default value
- **Updated component family states**: Modified message-related state
families to accept `threadId: string | null` instead of `threadId:
string`
- **Refined null checks**: Added explicit `null` checks in:
- `AgentChatMessagesFetchEffect`: Updated `isNewThread` logic to check
for `null` first
- `useAIChatThreadClick` and `useSwitchToNewAIChat`: Added guards to
only save drafts when `currentAIChatThread !== null`
- `AgentChatThreadInitializationEffect`: Added null check before UUID
validation
- `useEnsureAgentChatThreadIdForSend`: Added null check before comparing
with draft key
- **Updated fallback logic**: Used nullish coalescing operator (`??`) in
`AIChatTab` and `useAIChatEditor` to default to
`AGENT_CHAT_NEW_THREAD_DRAFT_KEY` when thread is null
- **Enhanced refetch safety**: Added early return in
`handleRefetchMessages` to prevent refetching when in new thread state
## Implementation Details
- The change maintains backward compatibility by treating `null` the
same way the code previously treated `AGENT_CHAT_UNKNOWN_THREAD_ID`
- All draft saving operations now safely check for null before
attempting to store drafts
- The nullish coalescing pattern (`currentAIChatThread ??
AGENT_CHAT_NEW_THREAD_DRAFT_KEY`) ensures proper fallback behavior when
accessing draft storage
https://claude.ai/code/session_01Pz8KCygSNgBPYsndbMq8f7
---------
Co-authored-by: Claude <noreply@anthropic.com>
Fixes
https://twenty-v7.sentry.io/issues/7398810663/?project=4507072563183616
## Bug description
The mergeMultipleRecords command's conditionalAvailabilityExpression
used numberOfSelectedRecords >= 2, which evaluates correctly in both
selection and exclusion (Select All) modes. However, when the command
executes, buildHeadlessCommandContextApi returns an empty
selectedRecords array in exclusion mode because it can't synchronously
resolve record IDs from an "all minus excluded" set. This caused
MergeMultipleRecordsCommand to throw on the empty array guard.
## Fix
Prepended not isSelectAll and to the merge command's conditional
expression, preventing it from appearing when records are selected via
Select All.
When creating a db from scrath we still need to run the slow instance
command so the associated queries are still applied to the empty db
Please note that by default if there's no workspace the instance slow
runner will skip the runDataMigration part
# Introduction
Avoid having a time window where the logic function is unavailable while
being built, by implementing an update flow instead of delete and create
everytime
fixes https://twenty-v7.sentry.io/issues/7371110591/
**Note: executor code propagation (pre-existing limitation)**
The Lambda executor shim
(`logic-function-drivers/constants/executor/index.mjs`) is only deployed
when a Lambda is first created. If this file is edited, the change won't
propagate to existing Lambdas — neither before nor after this PR. A
follow-up could compare the deployed `CodeSha256` against a local
checksum to detect drift and trigger a code update via
`UpdateFunctionCodeCommand`.
Should implem as code versioning or checksum diffing
## Summary
- Added `flex: 1` to `StyledColumnContainer` in `RecordBoardColumns.tsx`
- The column wrapper wasn't stretching vertically, so the droppable area
only covered the top of each column where cards existed
- Now the entire column height is a valid drop target
## Test plan
- [ ] Open a board/kanban view
- [ ] Drag a card from one column
- [ ] Drop it in the lower/empty area of another column
- [ ] Verify the drop registers correctly
Fixes#18842
## Summary
Improved error handling in the StreamAgentChatJob to properly propagate
and reject errors that occur during the agent chat stream processing.
## Key Changes
- Added error re-throw in the catch block to ensure errors are not
silently swallowed
- Introduced `streamError` variable to capture errors from the stream's
`onError` callback
- Modified the promise resolution logic to reject with the captured
stream error instead of silently resolving on success, ensuring proper
error propagation to callers
## Implementation Details
The changes ensure that errors occurring during stream processing are
properly captured and propagated:
1. Stream errors are now captured in the `onError` callback and stored
in the `streamError` variable
2. After the stream completes and the message is persisted to the
database, the promise is rejected if a stream error occurred
3. The outer catch block now re-throws errors to prevent silent failures
This fix prevents scenarios where stream processing errors would be
masked by a successful database persistence operation.
https://claude.ai/code/session_016DQodL32HYv6CR1cMASNHZ
---------
Co-authored-by: Claude <noreply@anthropic.com>
## Summary
- Removes all workspace schema definitions for `Favorite` and
`FavoriteFolder` entities, which have been fully migrated to
`NavigationMenuItems`
- Deletes 26 standalone files including workspace entities, NestJS
modules, services, listeners, jobs, standard application builders (field
metadata, views, view fields, view field groups, indexes, page layouts),
mocks, and integration tests
- Cleans up ~40 modified files: removes `favorites` relation from 10
workspace entities and their field metadata utils, removes entries from
all builder maps, shared constants (`STANDARD_OBJECTS`,
`CoreObjectNameSingular`, `DEFAULT_RELATIONS_OBJECTS_STANDARD_IDS`), SDK
default relations, AI tool filtering, and standard object icons
## Summary
Aligns **object metadata** icons with the **tinted tile** look
everywhere we show a workspace object, and **retires** the
navigation-only `NavigationMenuItemStyleIcon` wrapper in favor of
**shared** UI primitives under `@/ui/display` and `@/object-metadata`.
## What changed
### Global tinted icon building blocks (`@/ui/display`)
- **`TintedIconTile`** / **`StyledTintedIconTileContainer`** support
optional **`size`** and **`stroke`**, and grow the tile when **`size`**
is set so layouts match previous `theme.icon` usage.
- Shared helpers and constants for theme color parsing and tinted
backgrounds/borders/icon color (e.g.
**`getTintedIconTileStyleFromColor`**, **`parseThemeColor`**,
**`getColorFromTheme`**, related constants).
### Object metadata icon (`@/object-metadata`)
- **`ObjectMetadataIcon`** composes **`TintedIconTile`** with
**`getObjectColorWithFallback`**, forwards optional **`size`** /
**`stroke`** for **visual parity** with old `getIcon` + explicit sizing.
- **`getSelectOptionIconFromObjectMetadataItem`** returns an
**`IconComponent`** for selects/menus that expect a component, not a
React node.
### Navigation module cleanup
- **Removed** **`NavigationMenuItemStyleIcon`**; call sites use
**`ObjectMetadataIcon`**, **`TintedIconTile`**, and/or the shared
**`getTintedIconTileStyleFromColor`** pipeline so the same treatment is
**not** tied to the navigation package.
- **`NavigationMenuItemIcon`**, view/link overlays, DnD handle, and
sidebar editor flows updated to use the shared pattern where they render
object (or tinted) icons.
### Product surfaces updated (non-exhaustive)
- **Settings:** role object picker/rows, data model
tables/graph/overview, object preview summary, webhooks entity list,
morph relation multiselect.
- **Workflows:** create/update/delete/upsert/find records, triggers,
filters, variables dropdowns, AI agent object rows, object dropdowns.
- **Shell:** side panel object filter / data sources / folder chrome
where object icons appear.
- **Records:** index header icon, show breadcrumb styling.
- **Activity:** timeline event icon when linked object metadata applies.
- **`NavigationDrawerItem`:** tinted branch aligned with shared
**`TintedIconTile`** behavior.
---------
Co-authored-by: Etienne <45695613+etiennejouan@users.noreply.github.com>
## Context
Emails and calendars can only be associated with specific objects
(companies, people...) and we were adding those tabs for all custom
objects. I'm removing these from the default config
## Summary
- refresh the website halftone studio with new rendering controls, UI
updates, and export/runtime plumbing
- improve halftone output quality by opening highlights, grouping
shadows, and reducing visible row artifacts
- update home/problem illustration assets, 3D model references, and
related illustration components to use the new visuals
- adjust footer and layout-related website wiring to support the updated
illustration experience
## Testing
- Not run (not requested)
---------
Co-authored-by: Abdullah <125115953+mabdullahabaid@users.noreply.github.com>
- Removes the intermediate `CommandMenuItemConfig` /
`CommandConfigContext` / `CommandMenuItemDisplay` abstraction layers,
replacing them with a single `CommandMenuItemRenderer` that renders
directly from the command menu items from the backend
- Eliminates the server-items/ subdirectory by moving its contents
(hooks/, contexts/, states/, display/, edit/) up into the parent
command-menu-item/ module, removing an unnecessary nesting level.
## Changes
### Email Attachments
- Added `EmailAttachmentsField` component for uploading and managing
email attachments
- New `useUploadEmailAttachment` hook for handling file uploads with
size validation
- New `UPLOAD_EMAIL_ATTACHMENT_FILE` mutation for backend file
persistence
- Integrated attachments into email composer with file validation
- Added `EmailRecipientLimits` constant to enforce max recipients (100)
on frontend
### Open in App Click Action
- New `useOpenEmailInAppOrFallback` hook to open emails in the in-app
composer
- Email fields now default to "Open in app" action instead of "Open as
link"
- New `SettingsDataModelFieldOnClickActionForm` support for
`OPEN_IN_APP` action
- Email secondary table cell button now offers in-app composer as
alternative action
- `AttachmentChip` component moved from advanced-text-editor to file
module for reuse
### Refactoring & New Utilities
- Extracted `useComposeEmailForTargetRecord` hook for consistent email
composer opening
- New `useResolveDefaultEmailRecipient` hook to resolve recipient based
on record type
- New `getPrimaryEmailFromRecord` utility for safe email field access
- New `EmptyInboxPlaceholder` component with CTA button
- Simplified `ComposeEmailButton` using new hooks
- Enhanced `ComposeEmailCommand` to support bulk Person selections
- Updated `useSendEmail` to accept and forward attachments
- Recipient count validation with warning in composer footer
### Backend
- New `FileEmailAttachmentModule` with resolver and service
- New `file-email-attachment.command` for record selection menu items
- Updated `SendEmailInput` GraphQL type to include `files` field
- Email tool constants and exceptions updated
### Type Updates
- Added `SendEmailAttachmentInput` GraphQL type
- Added `FileFolder.EMAIL_ATTACHMENT` to file folder interface
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com>
## Summary
- Add SSE (`text/event-stream`) as an alternative response format on
`POST /mcp` per the MCP streamable-http spec
- When clients send `Accept: text/event-stream`, the server responds
with SSE wire format; otherwise returns JSON as before (fully backwards
compatible)
- Emit a `notifications/progress` SSE event before tool execution to
signal long-running operations
- No sessions, no GET SSE, no protocol version bump — this is
transport-level only (Phase 2)
## Changes
- **New**: `write-sse-event.util.ts` — writes correctly formatted SSE
events to Express Response
- **New**: `mcp-progress-notification.const.ts` — constants for progress
notification method and token prefix
- **Modified**: `mcp-core.controller.ts` — checks `Accept` header,
branches into SSE vs JSON response path
- **Modified**: `mcp-protocol.service.ts` — passes optional `sseWriter`
callback to tool executor
- **Modified**: `mcp-tool-executor.service.ts` — emits progress
notification via `sseWriter` before tool execution
- **Tests**: Unit tests for SSE utility, controller SSE/JSON paths, tool
executor progress notifications, and integration tests for SSE streaming
## Test plan
- [x] Unit tests: `writeSseEvent` utility produces correct SSE wire
format
- [x] Unit tests: Controller returns SSE headers and writes events when
`Accept: text/event-stream`
- [x] Unit tests: Controller returns JSON when `Accept:
application/json` only
- [x] Unit tests: Notifications (no `id`) return 202 regardless of
Accept header
- [x] Unit tests: Tool executor emits progress notification via
sseWriter
- [x] Unit tests: Tool executor works without sseWriter (backwards
compatible)
- [x] Integration tests: SSE response for ping, JSON fallback, progress
notification before tool call
- [x] All 503 test suites pass, typecheck clean
https://claude.ai/code/session_01QrqjBUXePJkPMd6gBAoWaR
---------
Co-authored-by: Claude <noreply@anthropic.com>
## Summary
- **Drop the `objectMetadata.dataSourceId` foreign key and index** via a
1-22 fast instance command — column kept nullable for data preservation
- **Delete `DataSourceService`, `DataSourceModule`, and
`DataSourceException`** — all code now uses `workspace.databaseSchema`
directly
- **Remove `IS_DATASOURCE_MIGRATED` feature flag** from default flags
and all branching logic
- **Simplify workspace/object creation pipelines** —
`WorkspaceManagerService`, `DevSeederService`, and the object creation
action handler no longer route through `DataSourceService`
- **Keep `DataSourceEntity` and the `dataSource` table** for historical
data — entity stripped of all ORM relations
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.
Co-authored-by: FelixMalfait <6399865+FelixMalfait@users.noreply.github.com>
## Summary
Fix MCP `/mcp` transport handling for clients using `streamable-http`.
## Changes
- add explicit `GET /mcp` and `DELETE /mcp` handlers
- return `405 Method Not Allowed` with `Allow: POST`
- keep `POST /mcp` protected by MCP auth guards
- mark `GET` and `DELETE` as intentionally public with
`PublicEndpointGuard` + `NoPermissionGuard`
- update the advertised MCP protocol version to `2025-03-26`
- add unit and integration coverage for the new behavior
## Why
The frontend advertises the MCP server as `streamable-http`, but the
backend only effectively handled `POST /mcp`. Some MCP clients probe
`GET /mcp` during connection setup, so unsupported methods need explicit
method-level responses instead of falling through or being blocked
before the handler.
## Validation
- verified locally:
- `GET /mcp` -> `405`
- `DELETE /mcp` -> `405`
- unauthenticated `POST /mcp` -> `401`
- passed controller unit tests
- added integration assertions for `GET` and `DELETE`
---------
Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: Félix Malfait <felix.malfait@gmail.com>
## Summary
This PR adds support for attaching files to agent chat messages. Users
can now upload files when sending messages to the AI agent, and these
files are properly processed, stored, and made available to the agent
with signed URLs.
## Key Changes
- **File attachment input**: Added `fileIds` parameter to the
`sendChatMessage` GraphQL mutation to accept file IDs from the client
- **File processing**: Implemented `buildFilePartsFromIds()` method to
convert file IDs into file UI parts with signed URLs
- **Message composition**: Updated user messages to include both text
and file parts when files are attached
- **File URL signing**: Integrated `FileUrlService` to generate signed
URLs for files in the AgentChat folder, ensuring secure access
- **Message persistence**: Files are now included in the message parts
stored in the database and retrieved when loading conversation history
- **File metadata mapping**: Enhanced `mapDBPartToUIMessagePart()` to
properly extract MIME types from file entities and include file IDs
## Implementation Details
- Files are fetched from the database using the provided file IDs and
workspace context
- Each file is converted to an `ExtendedFileUIPart` with proper metadata
(filename, MIME type, signed URL, and file ID)
- When loading messages from the database, file parts are enhanced with
signed URLs to ensure they remain accessible
- The `loadMessagesFromDB()` method now requires the workspace ID to
properly sign file URLs
- File attachments are seamlessly integrated into the existing message
part system alongside text content
https://claude.ai/code/session_01TAdN1gBzeiYELX4XDrrYY1
---------
Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com>
- removes pre-install function
- execute **asyncrhonously** post-install function at application
installation
- add optional `shouldRunOnVersionUpgrade` boolean value on post-install
function definition default false
- update PostInstallPayload to
```
export type PostInstallPayload = {
previousVersion?: string;
newVersion: string;
};
```
---------
Co-authored-by: Charles Bochet <charles@twenty.com>
## Summary
Refactored the `AgentChatMessageUIToolCallPart` output structure to
flatten the nested result object, simplifying the data model and
improving code clarity.
## Key Changes
- **Flattened output structure**: Moved `success`, `message`, and
`result` fields from `output.result` directly to `output`, eliminating
unnecessary nesting
- **Removed `toolName` field**: Deleted the unused `toolName` property
from the output object
- **Made fields optional**: Changed `error` and `result` to optional
fields to better represent cases where they may not be present
- **Updated hook logic**: Modified `useProcessUIToolCallMessage` to
access the flattened structure:
- Changed `toolExecutionPart.output.result.success` to
`toolExecutionPart.output.success`
- Changed `toolExecutionPart.output.result.result` to
`toolExecutionPart.output.result`
- Added null check for `navigateAppOutput` to handle cases where result
is undefined
## Implementation Details
The refactoring maintains backward compatibility in functionality while
simplifying the data structure. The optional `result` field now properly
reflects that navigation output may not always be present, with an
explicit guard clause added to handle this case gracefully.
https://claude.ai/code/session_01EnYKwVaCJyhgNPsi7p3YSq
Co-authored-by: Claude <noreply@anthropic.com>
## Summary
This PR adds a database migration command to backfill standard skills
for existing workspaces and improves the skill loading tool to provide
dynamic, workspace-specific skill availability information instead of
hardcoded skill names.
## Key Changes
- **New Migration Command**: Added `BackfillStandardSkillsCommand`
(v1.22.0) that:
- Identifies missing standard skills in existing workspaces
- Compares workspace skills against the standard skill definitions
- Creates missing skills using the workspace migration service
- Supports dry-run mode for safe testing
- Properly logs all operations and handles failures
- **Enhanced Skill Loading Tool**: Updated `createLoadSkillTool` to:
- Accept a new `listAvailableSkillNames` function parameter
- Dynamically fetch available skills from the workspace instead of using
hardcoded skill names
- Provide accurate, context-aware error messages when skills are not
found
- Gracefully handle workspaces with no available skills
- **Service Updates**: Modified skill tool implementations in:
- `McpProtocolService`: Integrated `findAllFlatSkills` to list available
skills
- `ChatExecutionService`: Integrated `findAllFlatSkills` to list
available skills
- **Module Registration**: Added `BackfillStandardSkillsCommand` to the
v1.22 upgrade module
- **Test Updates**: Updated `McpProtocolService` tests to mock the new
`findAllFlatSkills` method
## Implementation Details
The backfill command uses the existing workspace migration
infrastructure to safely create skills, ensuring consistency with other
metadata operations. The skill availability messaging now reflects the
actual skills present in each workspace, improving user experience when
skills are not found.
https://claude.ai/code/session_012fXeP3bysaEgWsbkyu4ism
Co-authored-by: Claude <noreply@anthropic.com>
## Context
- Extend the FIELD widget to support RICH_TEXT fields alongside existing
RELATION/MORPH_RELATION fields
- Add EDITOR display mode that renders a full rich text editor, and
FIELD display mode that shows a compact single-line preview
- Enforce mutual exclusivity: EDITOR is only available for RICH_TEXT,
CARD only for RELATION, FIELD for both
- Refactor widget configuration into a centralized FIELD_WIDGET_CONFIG
constant and shared useFieldWidgetEligibleFields hook for better
scalability. Later we might use discriminative union from graphql
scehma)
<details>
<summary>
EDITOR mode
</summary>
<img width="1095" height="731" alt="Screenshot 2026-04-09 at 17 31 41"
src="https://github.com/user-attachments/assets/cebffd0e-07ea-4f74-a3dc-ef987daa17ea"
/>
</details>
<details>
<summary>
FIELD mode
</summary>
<img width="986" height="378" alt="Screenshot 2026-04-09 at 17 35 00"
src="https://github.com/user-attachments/assets/c90a8046-fdd0-4321-8ba6-f47d89e9d42a"
/>
</details>
<details>
<summary>
Open FIELD mode
</summary>
<img width="758" height="480" alt="Screenshot 2026-04-09 at 17 35 04"
src="https://github.com/user-attachments/assets/c53cc120-0b6f-47a0-808c-27b26f9f53ec"
/>
</details>
## Summary
- System objects with valid views were incorrectly hidden in the View >
System objects picker
- The system picker page used a different filter (`view.key !==
ViewKey.INDEX`) than the parent page
(`isViewDisplayableInNavigationMenu`), causing a mismatch where the
"System objects" link was visible but clicking it showed an empty list
- Aligned filtering in `SidePanelNewSidebarItemViewSystemPickerSubPage`
to use `isViewDisplayableInNavigationMenu` and exclude views already in
the workspace, consistent with the non-system picker page
## Summary
- **Split the `rolesPermissions` cache query into parallel queries** to
eliminate a 5-way LEFT JOIN cartesian product. A workspace with a role
having 5 objectPermissions × 4 permissionFlags × 124 fieldPermissions ×
5 RLP predicates × 5 RLP groups was returning 62,000 rows from just ~162
distinct records, causing 10s+ query timeouts on view updates.
- **Added missing `roleId` index on `permissionFlag` table** — the only
permission table without one, which was causing sequential scans on the
JOIN.
<img width="1511" height="825" alt="Capture d’écran 2026-04-09 à 14 19
52"
src="https://github.com/user-attachments/assets/cd6c533e-abc9-45f9-b5c1-5cb7341d5dfe"
/>
- Move GetChatThreads query from the generic metadata pipeline
(useLoadStaleMetadataEntities) into AgentChatThreadInitializationEffect,
gated by useHasPermissionFlag(AI_SETTINGS) — prevents "Entity does not
have permissions" errors for users whose role lacks AI access
- Fix initialization bug where isDefined(currentAIChatThread) always
returned true for the 'unknown-thread' sentinel value, preventing thread
initialization from ever running — replaced with isValidUuid check
## Summary
- Introduces a new `application-logs` core module with a driver pattern
(disabled/console/clickhouse) to capture and persist logic function
execution logs
- Adds a ClickHouse `applicationLog` table with per-line log storage,
30-day TTL, and `ORDER BY (workspaceId, timestamp, applicationId,
logicFunctionId)`
- Surfaces application logs in the existing frontend audit logs table as
a new "Application Logs" source with dedicated columns (Function,
Timestamp, Level, Message, Execution ID)
## Details
**Write path**: `LogicFunctionExecutorService.handleExecutionResult()`
parses the multi-line log string from driver output into individual `{
timestamp, level, message }` entries, generates an execution UUID, and
passes them to `ApplicationLogsService.writeLogs()` which delegates to
the configured driver.
**Driver pattern**: Follows the exception-handler module style (Symbol
injection token + `forRootAsync` dynamic module). Three drivers:
- `DISABLED` (default) — no-op, prevents information leaking
- `CONSOLE` — structured stdout logging with level-based `console.*`
calls
- `CLICKHOUSE` — inserts rows into the `applicationLog` ClickHouse table
**Read path**: Extends the existing event-logs module by adding
`APPLICATION_LOG` to the `EventLogTable` enum, table name mapping, and
normalization logic.
**Config**: New `APPLICATION_LOG_DRIVER_TYPE` environment variable
(default: `DISABLED`).
As per title. Replaced .json file with .lottie file - it is 10x smaller.
Illustrations folder went from 150mb to 5.9mb with compressed variants
of 3D models. The DracoLoader, GLTFLoader and some other logic logic is
repetitive across files, but Thomas is working on the same files at the
moment, so not touching too much to avoid conflicts with his PR. Once
he's done styling, we can extract shared logic into helpers.
Migrating some 1.21 migrations to the new pattern so it writes in the
history
When release in prod we will ahve to manually inject them as they have
already been run
It's mainly for self host so when releasing the 1.22 we will be able to
rely on the 1.21 history
<details>
<summary>
Reorganizing standard page layouts (see screenshot below)
</summary>
<img width="355" height="617" alt="Screenshot 2026-04-09 at 10 44 02"
src="https://github.com/user-attachments/assets/0b71d607-1d9d-48b6-8612-3de98d12d1fa"
/>
</details>
<details>
<summary>
Note: All standard objects now have a last system section for
createdBy/createdAt however since all new custom fields are added to the
last section they are set there (see 2nd screenshot below) until people
move them to dedicated section which can be counterintuitive. There is
an upcoming feature that allows user to set a default section and
visibility for newly added fields that should solve that
</summary>
<img width="360" height="849" alt="Screenshot 2026-04-09 at 10 43 44"
src="https://github.com/user-attachments/assets/127990d0-66b7-4b1d-ab00-8251e9707a04"
/>
</details>
Another note: Section are not translated at the moment
## Bug description
Headless command menu items were interpreted as non headless command
menu items and opened in the side panel.
https://github.com/user-attachments/assets/44d8b4d3-9ee5-4f12-9ff8-b3ed51e5b3b6
## Fix
https://github.com/user-attachments/assets/9d6eb900-9817-4ceb-87aa-547ad046a5a1
- Command menu items received via SSE lack the nested `frontComponent`
relation (only `frontComponentId` is present), causing
`item.frontComponent?.isHeadless` to always be undefined.
- Add `frontComponents` to the metadata store's stale entity loading
- Rewrite `commandMenuItemsSelector` to join `commandMenuItems` with
`frontComponents` by `frontComponentId`
## Summary
Added a Redis cache flush step in the breaking changes CI workflow to
prevent stale data contamination between consecutive server runs.
## Key Changes
- Added a new workflow step that executes `redis-cli FLUSHALL` between
the current branch server run and the main branch server run
- Includes error handling to log a warning if the Redis flush fails,
without blocking the workflow
- Added explanatory comments documenting why this step is necessary
## Implementation Details
The Redis flush is necessary because both the current branch and main
branch servers share the same Redis instance during CI testing. The
`CoreEntityCacheService` and `WorkspaceCacheService` persist cached
entities across process restarts, which can cause stale data from the
current branch server to contaminate the main branch server comparison.
This step ensures a clean cache state before switching branches.
https://claude.ai/code/session_01BVMacfAXDMNx5WAFtP7GgW
Co-authored-by: Claude <noreply@anthropic.com>
- disable publish with existing version
- disable installation of app version already installed
---------
Co-authored-by: cubic-dev-ai[bot] <191113872+cubic-dev-ai[bot]@users.noreply.github.com>
When an object is created or enabled we create a navigation command menu
item to that object. And when the object is disabled or deleted, we
remove this command menu item.
## Summary
This PR refactors the EventRow component structure by extracting shared
types and styled components into a dedicated base file, improving code
organization and reducing duplication.
## Key Changes
- Created new `EventRowBase.tsx` file to centralize shared EventRow
utilities
- Moved `EventRowDynamicComponentProps` interface from
`EventRowDynamicComponent.tsx` to `EventRowBase.tsx`
- Moved styled components `StyledEventRowItemColumn` and
`StyledEventRowItemAction` from `EventRowDynamicComponent.tsx` to
`EventRowBase.tsx`
- Updated all imports across 6 files to reference the new `EventRowBase`
module instead of `EventRowDynamicComponent`
- Simplified `EventRowDynamicComponent.tsx` to re-export types and
styles from `EventRowBase` for backward compatibility
## Files Updated
- `EventRowDynamicComponent.tsx` - Simplified to import and re-export
from EventRowBase
- `EventRowActivity.tsx` - Updated import path
- `EventRowCalendarEvent.tsx` - Updated import path
- `EventRowMainObject.tsx` - Updated import path
- `EventRowMainObjectUpdated.tsx` - Updated import path
- `EventRowMessage.tsx` - Updated import path
## Benefits
- Better separation of concerns with shared base utilities in a
dedicated module
- Reduced circular dependency risks
- Clearer module structure for future EventRow-related components
- Maintains backward compatibility through re-exports
https://claude.ai/code/session_011EyhBJ56RGuZHxBVWwzEQy
---------
Co-authored-by: Claude <noreply@anthropic.com>
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.
Co-authored-by: FelixMalfait <6399865+FelixMalfait@users.noreply.github.com>
## Summary
- `buildBedrockProvider` ignored `authType: "role"` and only honored
static `accessKeyId`/`secretAccessKey` pairs, so deployments relying on
IRSA (e.g. EKS pods with `AWS_WEB_IDENTITY_TOKEN_FILE` + `AWS_ROLE_ARN`)
had no way to authenticate — `@ai-sdk/amazon-bedrock` does not walk the
AWS default credential chain on its own.
- When `authType === 'role'`, wire `fromNodeProviderChain()` from
`@aws-sdk/credential-providers` (already a server dependency, used by
the S3 and logic-function drivers) into `createAmazonBedrock` so IRSA
and other ambient AWS credentials resolve correctly.
- The static-credentials path is unchanged for `authType !== 'role'`.
## Test plan
- [ ] Deploy a server pod with IRSA + an `ai-models-config` entry using
`"authType": "role"` and verify Bedrock calls succeed (no `Could not
load credentials from any providers` error).
- [ ] Verify static-credentials providers (`accessKeyId` +
`secretAccessKey`) still work as before.
- [ ] `npx nx lint:diff-with-main twenty-server` passes.
- [ ] `npx nx typecheck twenty-server` passes.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
## Summary
Fixes#19422.
Self-hosted users hitting "No AI models are available" after configuring
API keys via the admin panel were victims of a stale
`AiModelRegistryService` cache. Only the four `addAiProvider` /
`removeAiProvider` / `addModelToProvider` / `removeModelFromProvider`
mutations called `refreshRegistry()` — setting an API key through
`set/update/deleteDatabaseConfigVariable` left the registry pointing at
the pre-mutation provider state.
Rather than patch each mutation site (and re-introduce the same class of
bug on the next one), the registry now invalidates lazily based on the
LLM config-group hash, mirroring the pattern `WebSearchDriverFactory`
already uses via `DriverFactoryBase`. Any mutation to an LLM-tagged
config variable is picked up automatically on the next read — callers
never have to remember to refresh.
- Extracted `getConfigGroupHash` into a shared util reused by both
`DriverFactoryBase` and `AiModelRegistryService` (and switched the hash
to `JSON.stringify` so object-typed config vars like `AI_PROVIDERS`
actually contribute meaningfully).
- `AiModelRegistryService` gates all internal `Map` access behind
private getters that call `ensureFresh()`, so future read paths can't
accidentally observe stale state. Build path uses underscored backing
fields directly to avoid recursing.
- Dropped the now-redundant `refreshRegistry()` public method and its
four call sites in `admin-panel.resolver.ts`.
## Test plan
- [x] `nx typecheck twenty-server` passes
- [x] Existing admin-panel + ai-models specs pass (79 tests)
- [ ] Manual: on a self-hosted instance, set `OPENAI_API_KEY` (or
another provider key) via the admin panel `setDatabaseConfigVariable`
mutation and confirm AI features work without a server restart
- [ ] Manual: existing flows (`addAiProvider`, `addModelToProvider`,
etc.) still pick up new providers on the next read
🤖 Generated with [Claude Code](https://claude.com/claude-code)
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
## Context
- Add "Reset to default" for page layout tabs and widgets — backend
mutation resets overrides, reactivates deactivated children, and deletes
custom children
- Fix override write bug where mutating an overridable property (e.g.
widget title) on a standard-app entity incorrectly overwrote the base
column instead of writing to the overrides JSONB —
PageLayoutUpdateService now uses resolveFlatEntityOverridableProperties
for accurate diffing and routes properties through
sanitizeOverridableEntityInput
- Deprecate isOverridden - We need more time to think about this
feature. Currently this adds too much complexity for a very small
benefit
https://github.com/user-attachments/assets/a84546c8-1e15-4d9e-a489-0825cf8b8ed2
## Summary
- When `WEB_SEARCH_DRIVER` is set to a non-native driver (e.g. `EXA`),
the `web_search` action tool was only listed in the tool catalog and had
to be discovered via `learn_tools` before use. Worse,
`system-prompt-builder` explicitly instructed the model **not** to call
`web_search` whenever it wasn't in the preloaded set, so even setting
`EXA_API_KEY` correctly resulted in the model never calling Exa.
- Fix: preload `web_search` from the registry alongside the other action
tools when `shouldUseNativeSearch()` is false, mirroring how native
provider search is already injected into `directTools`. The system
prompt builder picks this up automatically and advertises `web_search`
as a ready-to-use tool.
- Unrelated: pass `isSystemBuild: true` to the messageThread upgrade
command's migration build.
## Test plan
- [ ] With `WEB_SEARCH_DRIVER=EXA` and `EXA_API_KEY` set, ask the chat
agent for current/news information and verify it calls `web_search`
directly (not via `learn_tools` / `execute_tool`) and that Exa is hit.
- [ ] With `WEB_SEARCH_PREFER_NATIVE=true`, verify native provider
search is still used and the action tool is not preloaded.
- [ ] With `WEB_SEARCH_DRIVER=DISABLED`, verify behavior is unchanged
(`shouldUseNativeSearch()` returns true → no Exa preload).
- [ ] Run the 1-21 fix-message-thread-view-and-label-identifier upgrade
command on a workspace and confirm the migration builds/runs as a system
build.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
# Introduction
Persist in the db the registry computed name that contains
version_classname_timestamp to be consistent with the instance command
pattern too
Fix record table not refreshing after merging records
The record table's virtualization state persists across navigations, so
when the table remounts after a merge redirect, none of the existing
reload conditions were satisfied and the table rendered stale data.
Add an isInitializedOnMount local state to
RecordTableVirtualizedInitialDataLoadEffect that guarantees a fresh
triggerInitialRecordTableDataLoad on every mount, acting as a fallback
when no other reload condition matches.
## Summary
- Adds `deploy-twenty-app` and `install-twenty-app` composite actions to
`.github/actions/` so app repos can reference them remotely — same
pattern as `spawn-twenty-app-dev-test` for CI
- Updates `cd.yml` in template, hello-world, and postcard to use
`twentyhq/twenty/.github/actions/deploy-twenty-app@main` /
`install-twenty-app@main` instead of local `./.github/actions/` copies
- Removes the 6 local action files that were duplicated across template
and example apps
**Before** (each app repo carried its own action copies):
```yaml
uses: ./.github/actions/deploy
```
**After** (centralized, like CI):
```yaml
uses: twentyhq/twenty/.github/actions/deploy-twenty-app@main
```
Made with [Cursor](https://cursor.com)
2026-04-08 15:25:51 +02:00
2498 changed files with 141923 additions and 87810 deletions
This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
## Project Overview
Twenty is an open-source CRM built with modern technologies in a monorepo structure. The codebase is organized as an Nx workspace with multiple packages.
## Key Commands
### Development
```bash
# Start development environment (frontend + backend + worker)
yarn start
# Individual package development
npx nx start twenty-front # Start frontend dev server
npx nx start twenty-server # Start backend server
npx nx run twenty-server:worker # Start background worker
```
### Testing
```bash
# Preferred: run a single test file (fast)
npx jest path/to/test.test.ts --config=packages/PROJECT/jest.config.mjs
# Run all tests for a package
npx nx test twenty-front # Frontend unit tests
npx nx test twenty-server # Backend unit tests
npx nx run twenty-server:test:integration:with-db-reset # Integration tests with DB reset
# To run an indivual test or a pattern of tests, use the following command:
cd packages/{workspace}&& npx jest "pattern or filename"
# Storybook
npx nx storybook:build twenty-front
npx nx storybook:test twenty-front
# When testing the UI end to end, click on "Continue with Email" and use the prefilled credentials.
```
### Code Quality
```bash
# Linting (diff with main - fastest, always prefer this)
IMPORTANT: Use Context7 for code generation, setup or configuration steps, or library/API documentation. Automatically use the Context7 MCP tools to resolve library IDs and get library docs without waiting for explicit requests.
### Before Making Changes
1. Always run linting (`lint:diff-with-main`) and type checking after code changes
2. Test changes with relevant test suites (prefer single-file test runs)
3. Ensure instance commands are generated for entity changes (`database:migrate:generate`)
4. Check that GraphQL schema changes are backward compatible
5. Run `graphql:generate` after any GraphQL schema changes
### Code Style Notes
- Use **Linaria** for styling with zero-runtime CSS-in-JS (styled-components pattern)
- Follow **Nx** workspace conventions for imports
- Use **Lingui** for internationalization
- Apply security first, then formatting (sanitize before format)
### Testing Strategy
- **Test behavior, not implementation** — focus on user perspective
- **Test pyramid**: 70% unit, 20% integration, 10% E2E
- Query by user-visible elements (text, roles, labels) over test IDs
- Use `@testing-library/user-event` for realistic interactions
- Descriptive test names: "should [behavior] when [condition]"
- Clear mocks between tests with `jest.clearAllMocks()`
## Dev Environment Setup
All dev environments (Claude Code web, Cursor, local) use one script:
```bash
bash packages/twenty-utils/setup-dev-env.sh
```
This handles everything: starts Postgres + Redis (auto-detects local services vs Docker), creates databases, and copies `.env` files. Idempotent — safe to run multiple times.
-`--docker` — force Docker mode (uses `packages/twenty-docker/docker-compose.dev.yml`)
-`--down` — stop services
-`--reset` — wipe data and restart fresh
- **Skip the setup script** for tasks that only read code — architecture questions, code review, documentation, etc.
**Note:** CI workflows (GitHub Actions) manage services via Actions service containers and run setup steps individually — they don't use this script.
## Important Files
-`nx.json` - Nx workspace configuration with task definitions
-`tsconfig.base.json` - Base TypeScript configuration
-`package.json` - Root package with workspace definitions
-`.cursor/rules/` - Detailed development guidelines and best practices
- **nx-rules.mdc** - Nx workspace guidelines and best practices (Auto-attached to Nx files)
- **server-migrations.mdc** - Backend migration and TypeORM guidelines for `twenty-server` (Auto-attached to server entities and migration files)
- **server-migrations.mdc** - Upgrade command guidelines (instance commands and workspace commands) for `twenty-server` (Auto-attached to server entities and upgrade command files)
- **creating-syncable-entity.mdc** - Comprehensive guide for creating new syncable entities (with universalIdentifier and applicationId) in the workspace migration system (Agent-requested for metadata-modules and workspace-migration files)
### Code Quality
@@ -81,10 +81,8 @@ npx nx run twenty-server:typecheck # Type checking
npx nx run twenty-server:test # Run unit tests
npx nx run twenty-server:test:integration:with-db-reset # Run integration tests
# Migrations
npx nx run twenty-server:typeorm migration:generate src/database/typeorm/core/migrations/[name] -d src/database/typeorm/core/core.datasource.ts
# Workspace
# Upgrade commands (instance + workspace)
npx nx run twenty-server:database:migrate:generate --name <name> --type <fast|slow>
- **When changing an entity, always generate a migration**
- If you modify a `*.entity.ts` file in `packages/twenty-server/src`, you **must** generate a corresponding TypeORM migration instead of manually editing the database schema.
- Use the Nx + TypeORM command from the project root:
The upgrade system uses two types of commands instead of raw TypeORM migrations:
- **Instance commands** — schema and data migrations that run once at the instance level.
- **Workspace commands** — commands that iterate over all active/suspended workspaces.
See `packages/twenty-server/docs/UPGRADE_COMMANDS.md` for full documentation.
### Instance Commands
- **When changing a `*.entity.ts` file**, generate an instance command:
```bash
npx nx run twenty-server:typeorm migration:generate src/database/typeorm/core/migrations/common/[name] -d src/database/typeorm/core/core.datasource.ts
npx nx run twenty-server:database:migrate:generate --name <name> --type <fast|slow>
```
- Replace `[name]` with a descriptive, kebab-case migration name that reflects the change (for example, `add-agent-turn-evaluation`).
- **Fast commands** (`--type fast`, default) are for schema-only changes that must run immediately. They implement `FastInstanceCommand` with `up`/`down` methods and use the `@RegisteredInstanceCommand` decorator.
- **Prefer generated migrations over manual edits**
- Let TypeORM infer schema changes from the updated entities; only adjust the generated migration file manually if absolutely necessary (for example, for data backfills or complex constraints).
- Keep schema changes (DDL) in these generated migrations and avoid mixing in heavy data migrations unless there is a strong reason and clear comments.
- **Slow commands** (`--type slow`) add a `runDataMigration` method for potentially long-running data backfills that execute before `up`. They only run when `--include-slow` is passed. Use the decorator with `{ type: 'slow' }`.
- **Keep migrations consistent and reversible**
- Ensure the generated migration includes both `up` and `down` logic that correctly applies and reverts the entity change when possible.
- Do not delete or rewrite existing, committed migrations unless you are explicitly working on a pre-release branch where history rewrites are allowed by team conventions.
- The generator auto-registers the command in `instance-commands.constant.ts` — do not edit that file manually.
- **Keep commands consistent and reversible**: include both `up` and `down` logic. Do not delete or rewrite existing, committed commands unless on a pre-release branch.
### Workspace Commands
- Use the `@RegisteredWorkspaceCommand` decorator alongside nest-commander's `@Command` decorator.
- Extend `ActiveOrSuspendedWorkspaceCommandRunner` and implement `runOnWorkspace`.
- The base class provides `--dry-run`, `--verbose`, and workspace filter options automatically.
### Execution Order
Within a given version, commands run in this order (timestamp-sorted within each group):
1. Instance fast commands
2. Instance slow commands (only with `--include-slow`)
echo "::error::Unexpected migration files were generated. Please run 'npx nx database:migrate:generate twenty-server -- --name <migration-name>' and commit the result."
echo "$CORE_MIGRATION_OUTPUT"
echo ""
echo "The following migration changes were detected:"
- Creating an object without an index view associated. Unless this is a technical object, user will need to visualize it.
- Creating a view without a navigationMenuItem associated. This will make the view available on the left sidebar.
- Creating a front-end component that has a scroll instead of being responsive to its fixed widget height and width, unless it is specifically meant to be used in a canvas tab.
- Creating an object without an index view associated. Unless this is a technical object, user will need to visualize it.
- Creating a view without a navigationMenuItem associated. This will make the view available on the left sidebar.
- Creating a front-end component that has a scroll instead of being responsive to its fixed widget height and width, unless it is specifically meant to be used in a canvas tab.
- Creating an object without an index view associated. Unless this is a technical object, user will need to visualize it.
- Creating a view without a navigationMenuItem associated. This will make the view available on the left sidebar.
- Creating a front-end component that has a scroll instead of being responsive to its fixed widget height and width, unless it is specifically meant to be used in a canvas tab.
**Write a good `description`.** AI agents rely on the function's `description` field to decide when to use the tool. Be specific about what the tool does and when it should be called.
</Note>
</Accordion>
<Accordion title="definePreInstallLogicFunction" description="Define a pre-install logic function (one per app)">
A pre-install function is a logic function that runs automatically before your app is installed on a workspace. This is useful for validation tasks, prerequisite checks, or preparing workspace state before the main installation proceeds.
```ts src/logic-functions/pre-install.ts
import { definePreInstallLogicFunction, type InstallLogicFunctionPayload } from 'twenty-sdk';
description: 'Runs before installation to prepare the application.',
timeoutSeconds: 300,
handler,
});
```
You can also manually execute the pre-install function at any time using the CLI:
```bash filename="Terminal"
yarn twenty exec --preInstall
```
Key points:
- Pre-install functions use `definePreInstallLogicFunction()` — a specialized variant that omits trigger settings (`cronTriggerSettings`, `databaseEventTriggerSettings`, `httpRouteTriggerSettings`, `isTool`).
- The handler receives an `InstallLogicFunctionPayload` with `{ previousVersion: string }` — the version of the app that was previously installed (or an empty string for fresh installs).
- Only one pre-install function is allowed per application. The manifest build will error if more than one is detected.
- The function's `universalIdentifier` is automatically set as `preInstallLogicFunctionUniversalIdentifier` on the application manifest during the build — you do not need to reference it in `defineApplication()`.
- The default timeout is set to 300 seconds (5 minutes) to allow for longer preparation tasks.
</Accordion>
<Accordion title="definePostInstallLogicFunction" description="Define a post-install logic function (one per app)">
A post-install function is a logic function that runs automatically after your app is installed on a workspace. This is useful for one-time setup tasks such as seeding default data, creating initial records, or configuring workspace settings.
A post-install function is a logic function that runs automatically once your app has finished installing on a workspace. The server executes it **after** the app's metadata has been synchronized and the SDK client has been generated, so the workspace is fully ready to use and the new schema is in place. Typical use cases include seeding default data, creating initial records, configuring workspace settings, or provisioning resources on third-party services.
```ts src/logic-functions/post-install.ts
import { definePostInstallLogicFunction, type InstallLogicFunctionPayload } from 'twenty-sdk';
import { definePostInstallLogicFunction, type InstallPayload } from 'twenty-sdk';
- Post-install functions use `definePostInstallLogicFunction()` — a specialized variant that omits trigger settings (`cronTriggerSettings`, `databaseEventTriggerSettings`, `httpRouteTriggerSettings`, `isTool`).
- The handler receives an `InstallLogicFunctionPayload` with `{ previousVersion: string }` — the version of the app that was previously installed (or an empty string for fresh installs).
- The handler receives an `InstallPayload` with `{ previousVersion?: string; newVersion: string }` — `newVersion` is the version being installed, and `previousVersion` is the version that was previously installed (or `undefined` on a fresh install). Use these values to distinguish fresh installs from upgrades and to run version-specific migration logic.
- **When the hook runs**: on fresh installs only, by default. Pass `shouldRunOnVersionUpgrade: true` if you also want it to run when the app is upgraded from a previous version. When omitted, the flag defaults to `false` and upgrades skip the hook.
- **Execution model — async by default, sync opt-in**: the `shouldRunSynchronously` flag controls *how* post-install is executed.
- `shouldRunSynchronously: false` *(default)* — the hook is **enqueued on the message queue** with `retryLimit: 3` and runs asynchronously in a worker. The install response returns as soon as the job is enqueued, so a slow or failing handler does not block the caller. The worker will retry up to three times. **Use this for long-running jobs** — seeding large datasets, calling slow third-party APIs, provisioning external resources, anything that might exceed a reasonable HTTP response window.
- `shouldRunSynchronously: true` — the hook is executed **inline during the install flow** (same executor as pre-install). The install request blocks until the handler finishes, and if it throws, the install caller receives a `POST_INSTALL_ERROR`. No automatic retries. **Use this for fast, must-complete-before-response work** — for example, emitting a validation error to the user, or quick setup that the client will rely on immediately after the install call returns. Keep in mind the metadata migration has already been applied by the time post-install runs, so a sync-mode failure does **not** roll back the schema changes — it only surfaces the error.
- Make sure your handler is idempotent. In async mode the queue may retry up to three times; in either mode the hook may run again on upgrades when `shouldRunOnVersionUpgrade: true`.
- The environment variables `APPLICATION_ID`, `APP_ACCESS_TOKEN`, and `API_URL` are available inside the handler (same as any other logic function), so you can call the Twenty API with an application access token scoped to your app.
- Only one post-install function is allowed per application. The manifest build will error if more than one is detected.
- The function's `universalIdentifier` is automatically set as `postInstallLogicFunctionUniversalIdentifier` on the application manifest during the build — you do not need to reference it in `defineApplication()`.
- The function's `universalIdentifier`, `shouldRunOnVersionUpgrade`, and `shouldRunSynchronously` are automatically attached to the application manifest under the `postInstallLogicFunction` field during the build — you do not need to reference them in `defineApplication()`.
- The default timeout is set to 300 seconds (5 minutes) to allow for longer setup tasks like data seeding.
- **Not executed in dev mode**: when an app is registered locally (via `yarn twenty dev`), the server skips the install flow entirely and syncs files directly through the CLI watcher — so post-install never runs in dev mode, regardless of `shouldRunSynchronously`. Use `yarn twenty exec --postInstall` to trigger it manually against a running workspace.
</Accordion>
<Accordion title="definePreInstallLogicFunction" description="Define a pre-install logic function (one per app)">
A pre-install function is a logic function that runs automatically during installation, **before the workspace metadata migration is applied**. It shares the same payload shape as post-install (`InstallPayload`), but it is positioned earlier in the install flow so it can prepare state that the upcoming migration depends on — typical uses include backing up data, validating compatibility with the new schema, or archiving records that are about to be restructured or dropped.
```ts src/logic-functions/pre-install.ts
import { definePreInstallLogicFunction, type InstallPayload } from 'twenty-sdk';
description: 'Runs before installation to prepare the application.',
timeoutSeconds: 300,
shouldRunOnVersionUpgrade: true,
handler,
});
```
You can also manually execute the pre-install function at any time using the CLI:
```bash filename="Terminal"
yarn twenty exec --preInstall
```
Key points:
- Pre-install functions use `definePreInstallLogicFunction()` — same specialized config as post-install, just attached to a different lifecycle slot.
- Both pre- and post-install handlers receive the same `InstallPayload` type: `{ previousVersion?: string; newVersion: string }`. Import it once and reuse it for both hooks.
- **When the hook runs**: positioned just before the workspace metadata migration (`synchronizeFromManifest`). Before executing, the server runs a purely additive "pared-down sync" that registers the **new** version's pre-install function in the workspace metadata — nothing else is touched — and then executes it. Because this sync is additive-only, the previous version's objects, fields, and data are still intact when your handler runs: you can safely read and back up pre-migration state.
- **Execution model**: pre-install is executed **synchronously** and **blocks the install**. If the handler throws, the install is aborted before any schema changes are applied — the workspace stays on the previous version in a consistent state. This is intentional: pre-install is your last chance to refuse a risky upgrade.
- As with post-install, only one pre-install function is allowed per application. It is attached to the application manifest under `preInstallLogicFunction` automatically during the build.
- **Not executed in dev mode**: same as post-install — the install flow is skipped entirely for locally-registered apps, so pre-install never runs under `yarn twenty dev`. Use `yarn twenty exec --preInstall` to trigger it manually.
</Accordion>
<Accordion title="Pre-install vs post-install: when to use which" description="Choosing the right install hook">
Both hooks are part of the same install flow and receive the same `InstallPayload`. The difference is **when** they run relative to the workspace metadata migration, and that changes what data they can safely touch.
Pre-install is always **synchronous** (it blocks the install and can abort it). Post-install is **asynchronous by default** — enqueued on a worker with automatic retries — but can opt into synchronous execution with `shouldRunSynchronously: true`. See the `definePostInstallLogicFunction` accordion above for when to use each mode.
**Use `post-install` for anything that needs the new schema to exist.** This is the common case:
- Seeding default data (creating initial records, default views, demo content) against newly-added objects and fields.
- Registering webhooks with third-party services now that the app has its credentials.
- Calling your own API to finish setup that depends on the synchronized metadata.
- Idempotent "ensure this exists" logic that should reconcile state on every upgrade — combine with `shouldRunOnVersionUpgrade: true`.
Example — seed a default `PostCard` record after install:
```ts src/logic-functions/post-install.ts
import { definePostInstallLogicFunction, type InstallPayload } from 'twenty-sdk';
import { createClient } from './generated/client';
description: 'Seeds a welcome post card after install.',
timeoutSeconds: 300,
shouldRunOnVersionUpgrade: false,
handler,
});
```
**Use `pre-install` when a migration would otherwise destroy or corrupt existing data.** Because pre-install runs against the *previous* schema and its failure rolls back the upgrade, it is the right place for anything risky:
- **Backing up data that is about to be dropped or restructured** — e.g. you are removing a field in v2 and need to copy its values into another field or export them to storage before the migration runs.
- **Archiving records that a new constraint would invalidate** — e.g. a field is becoming `NOT NULL` and you need to delete or fix rows with null values first.
- **Validating compatibility and refusing the upgrade if the current data cannot be migrated cleanly** — throw from the handler and the install aborts with no changes applied. This is safer than discovering the incompatibility mid-migration.
- **Renaming or rekeying data** ahead of a schema change that would lose the association.
Example — archive records before a destructive migration:
```ts src/logic-functions/pre-install.ts
import { definePreInstallLogicFunction, type InstallPayload } from 'twenty-sdk';
import { createClient } from './generated/client';
| Run long-running seeding or third-party calls that shouldn't block the install response | `post-install` (default — `shouldRunSynchronously: false`, with worker retries) |
| Run fast setup that the caller will rely on immediately after the install call returns | `post-install` with `shouldRunSynchronously: true` |
| Read or back up data that the upcoming migration would lose | `pre-install` |
| Reject an upgrade that would corrupt existing data | `pre-install` (throw from the handler) |
| Run reconciliation on every upgrade | `post-install` with `shouldRunOnVersionUpgrade: true` |
| Do one-off setup on the first install only | `post-install` with `shouldRunOnVersionUpgrade: false` (default) |
<Note>
If in doubt, default to **post-install**. Only reach for pre-install when the migration itself is destructive and you need to intercept the previous state before it is gone.
</Note>
</Accordion>
<Accordion title="defineFrontComponent" description="Define front components for custom UI" >
@@ -220,6 +220,7 @@ The scaffolder already started a local Twenty server for you. To manage it later
|---------|-------------|
| `yarn twenty server start` | Start the local server (pulls image if needed) |
| `yarn twenty server start --port 3030` | Start on a custom port |
| `yarn twenty server start --test` | Start a separate test instance on port 2021 |
| `yarn twenty server stop` | Stop the server (preserves data) |
| `yarn twenty server status` | Show server status, URL, and credentials |
| `yarn twenty server logs` | Stream server logs |
@@ -228,6 +229,20 @@ The scaffolder already started a local Twenty server for you. To manage it later
Data is persisted across restarts in two Docker volumes (`twenty-app-dev-data` for PostgreSQL, `twenty-app-dev-storage` for files). Use `reset` to wipe everything and start fresh.
### Running a test instance
Pass `--test` to any `server` command to manage a second, fully isolated instance — useful for running integration tests or experimenting without touching your main dev data.
| Command | Description |
|---------|-------------|
| `yarn twenty server start --test` | Start the test instance (defaults to port 2021) |
| `yarn twenty server stop --test` | Stop the test instance |
| `yarn twenty server status --test` | Show test instance status, URL, and credentials |
| `yarn twenty server logs --test` | Stream test instance logs |
| `yarn twenty server reset --test` | Wipe test data and start fresh |
The test instance runs in its own Docker container (`twenty-app-dev-test`) with dedicated volumes (`twenty-app-dev-test-data`, `twenty-app-dev-test-storage`) and config, so it can run in parallel with your main instance without conflicts. Combine `--test` with `--port` to override the default 2021.
<Note>
The server requires **Docker** to be running. If you see a "Docker not running" error, make sure Docker Desktop (or the Docker daemon) is started.
@@ -66,14 +66,71 @@ The share link uses the server's base URL (without any workspace subdomain) so i
### Version management
When updating an already deployed tarball app, the server requires the `version` in `package.json` to be **strictly higher** (per [semver](https://semver.org) ordering) than the currently deployed version. Re-deploying the same version, or pushing a lower one, is rejected before the tarball is stored — you'll see a `VERSION_ALREADY_EXISTS` error from the CLI.
To release an update:
1. Bump the `version` field in your `package.json`
1. Bump the `version` field in your `package.json` (e.g. `1.2.3` → `1.2.4`, `1.3.0`, or `2.0.0`)
3. Workspaces that have the app installed will see the upgrade available in their settings
<Note>
Pre-release tags work as expected: bumping `1.0.0-rc.1` → `1.0.0-rc.2` is allowed, and a final release like `1.0.0` is correctly recognized as higher than `1.0.0-rc.5`. The version in `package.json` must itself be a valid semver string.
</Note>
{/* TODO: add screenshot of the Upgrade button */}
## Automated CI/CD (scaffolded workflows)
Apps generated with `create-twenty-app` ship with two GitHub Actions workflows out of the box, under `.github/workflows/`. They are ready to run as soon as you push the repo to GitHub — no extra setup is needed for CI, and CD only requires a single secret.
### CI — `ci.yml`
Runs integration tests on every push to `main` and every pull request.
**What it does:**
1. Checks out your app's source.
2. Spawns an isolated Twenty test instance using the `twentyhq/twenty/.github/actions/spawn-twenty-app-dev-test@main` composite action (the CI equivalent of `yarn twenty server start --test`).
3. Enables Corepack, sets up Node.js from your `.nvmrc`, and installs dependencies with `yarn install --immutable`.
4. Runs `yarn test`, passing `TWENTY_API_URL` and `TWENTY_API_KEY` from the spawned instance so your tests can talk to a real server.
**Config knobs:**
- `TWENTY_VERSION` (env, defaults to `latest`) — pin the Twenty server version used in CI by editing this in `ci.yml`.
- Concurrency is grouped by `github.ref` and cancels in-progress runs on new pushes.
No secrets are required — the test instance is ephemeral and lives only for the duration of the job.
### CD — `cd.yml`
Deploys your app to a configured Twenty server on every push to `main`, and optionally from a pull request when the `deploy` label is applied.
**What it does:**
1. Checks out the PR head (for labeled PRs) or the pushed commit.
2. Runs `twentyhq/twenty/.github/actions/deploy-twenty-app@main` — the CI equivalent of `yarn twenty deploy`.
3. Runs `twentyhq/twenty/.github/actions/install-twenty-app@main` so the newly deployed version is installed into the target workspace.
**Required configuration:**
| Setting | Where | Purpose |
|---------|-------|---------|
| `TWENTY_DEPLOY_URL` | `env` in `cd.yml` (defaults to `http://localhost:3000`) | The Twenty server to deploy to. Change this to your real server URL before first use. |
| `TWENTY_DEPLOY_API_KEY` | GitHub repo **Settings → Secrets and variables → Actions** | API key with deploy permission on the target server. |
<Note>
The default `TWENTY_DEPLOY_URL` of `http://localhost:3000` is a placeholder — it will not reach anything from a GitHub-hosted runner. Update it to your server's public URL (or use a self-hosted runner with network access) before enabling CD.
</Note>
**Triggering a preview deploy from a PR:**
Add the `deploy` label to a pull request. The `if:` guard in `cd.yml` will run the job for that PR using the PR's head commit, letting you validate a change on the target server before merging.
### Pinning the reusable actions
Both workflows reference reusable actions at `@main`, so action updates in the `twentyhq/twenty` repo are picked up automatically. If you want deterministic builds, replace `@main` with a commit SHA or release tag on each `uses:` line.
## Publishing to npm
Publishing to npm makes your app discoverable in the Twenty marketplace. Any Twenty workspace can browse, install, and upgrade marketplace apps directly from the UI.
@@ -81,7 +138,7 @@ Publishing to npm makes your app discoverable in the Twenty marketplace. Any Twe
### Requirements
- An [npm](https://www.npmjs.com) account
- The `twenty-app` keyword in your `package.json` `keywords` array (already included when you scaffold with `create-twenty-app`)
- The `twenty-app` keyword in your `package.json` `keywords` array (add it manually — it is not included by default in the `create-twenty-app` template)
```json filename="package.json"
{
@@ -189,3 +246,12 @@ You can also install apps from the command line:
```bash filename="Terminal"
yarn twenty install
```
<Note>
The server enforces semver versioning on install, mirroring the rules on deploy:
- Installing the same version that is already installed in your workspace is rejected with an `APP_ALREADY_INSTALLED` error.
- Installing a lower version than the one currently installed is rejected with a `CANNOT_DOWNGRADE_APPLICATION` error.
To install a newer version, deploy or publish it first, then re-run `yarn twenty install`.
**اكتب `description` جيدًا.** يعتمد وكلاء الذكاء الاصطناعي على حقل `description` الخاص بالدالة لتحديد وقت استخدام الأداة. كن محددًا بشأن ما تفعله الأداة ومتى ينبغي استدعاؤها.
</Note>
</Accordion>
<Accordion title="definePreInstallLogicFunction" description="تعريف دالة منطقية لما قبل التثبيت (واحدة لكل تطبيق)">
دالة ما قبل التثبيت هي دالة منطقية تعمل تلقائيًا قبل تثبيت تطبيقك على مساحة عمل. يفيد ذلك في مهام التحقق، وفحص المتطلبات المسبقة، أو تجهيز حالة مساحة العمل قبل متابعة التثبيت الرئيسي.
```ts src/logic-functions/pre-install.ts
import { definePreInstallLogicFunction, type InstallLogicFunctionPayload } from 'twenty-sdk';
description: 'Runs before installation to prepare the application.',
timeoutSeconds: 300,
handler,
});
```
يمكنك أيضًا تنفيذ دالة ما قبل التثبيت يدويًا في أي وقت باستخدام CLI:
```bash filename="Terminal"
yarn twenty exec --preInstall
```
النقاط الرئيسية:
* تستخدم دوال ما قبل التثبيت `definePreInstallLogicFunction()` — وهو إصدار متخصص يستبعد إعدادات المُشغِّل (`cronTriggerSettings` و`databaseEventTriggerSettings` و`httpRouteTriggerSettings` و`isTool`).
* يتلقى المُعالج `InstallLogicFunctionPayload` يحوي `{ previousVersion: string }` — إصدار التطبيق الذي كان مُثبّتًا سابقًا (أو سلسلة فارغة للتثبيتات الجديدة).
* يُسمح بدالة ما قبل التثبيت واحدة فقط لكل تطبيق. سيُنتج إنشاء ملف البيان خطأً إذا تم اكتشاف أكثر من واحدة.
* يتم تعيين `universalIdentifier` للدالة تلقائيًا كـ `preInstallLogicFunctionUniversalIdentifier` في بيان التطبيق أثناء الإنشاء — لست بحاجة إلى الإشارة إليه في `defineApplication()`.
* تم ضبط المهلة الافتراضية على 300 ثانية (5 دقائق) للسماح بمهام التحضير الأطول.
</Accordion>
<Accordion title="definePostInstallLogicFunction" description="تعريف دالة منطقية لما بعد التثبيت (واحدة لكل تطبيق)">
دالة ما بعد التثبيت هي دالة منطقية تعمل تلقائيًا بعد تثبيت تطبيقك على مساحة عمل. هذا مفيد لمهام الإعداد لمرة واحدة مثل تهيئة البيانات الافتراضية، وإنشاء السجلات الأولية، أو تكوين إعدادات مساحة العمل.
دالة ما بعد التثبيت هي دالة منطقية تعمل تلقائيًا بعد تثبيت تطبيقك على مساحة عمل. ينفّذه الخادم **بعد** مزامنة البيانات الوصفية للتطبيق وإنشاء عميل SDK، بحيث تكون مساحة العمل جاهزة تمامًا للاستخدام ويكون المخطط الجديد مطبَّقًا. تشمل حالات الاستخدام النموذجية تهيئة البيانات الافتراضية، وإنشاء السجلات الأولية، وتكوين إعدادات مساحة العمل، أو توفير الموارد على خدمات جهات خارجية.
```ts src/logic-functions/post-install.ts
import { definePostInstallLogicFunction, type InstallLogicFunctionPayload } from 'twenty-sdk';
import { definePostInstallLogicFunction, type InstallPayload } from 'twenty-sdk';
* تستخدم دوال ما بعد التثبيت `definePostInstallLogicFunction()` — وهو إصدار متخصص يستبعد إعدادات المُشغِّل (`cronTriggerSettings` و`databaseEventTriggerSettings` و`httpRouteTriggerSettings` و`isTool`).
* يتلقى المُعالج `InstallLogicFunctionPayload` يحوي `{ previousVersion: string }` — إصدار التطبيق الذي كان مُثبّتًا سابقًا (أو سلسلة فارغة للتثبيتات الجديدة).
* يتلقى المعالج `InstallPayload` يحتوي على `{ previousVersion?: string; newVersion: string }` — حيث إن `newVersion` هو الإصدار الجاري تثبيته، و`previousVersion` هو الإصدار الذي كان مُثبّتًا سابقًا (أو `undefined` عند التثبيت الأولي). استخدم هذه القيم للتمييز بين عمليات التثبيت الجديدة والترقيات ولتشغيل منطق الترحيل الخاص بالإصدار.
* **موعد تشغيل الخطاف**: في عمليات التثبيت الجديدة فقط، افتراضيًا. مرّر `shouldRunOnVersionUpgrade: true` إذا كنت تريد تشغيله أيضًا عند ترقية التطبيق من إصدار سابق. عند إغفاله، تكون القيمة الافتراضية للعلم `false`، وتتجاوز الترقيات هذا الخطاف.
* **نموذج التنفيذ — غير متزامن افتراضيًا، والتزامني اختياري**: يتحكّم العلم `shouldRunSynchronously` في كيفية تنفيذ ما بعد التثبيت.
* `shouldRunSynchronously: false` *(الإعداد الافتراضي)* — يتم **إدراج الخطاف في قائمة الرسائل** مع `retryLimit: 3` ويعمل بشكل غير متزامن داخل عامل عمل. يعود ردّ التثبيت بمجرد وضع المهمة في الطابور، لذا فإن معالجًا بطيئًا أو متعطلًا لا يحجب المستدعي. سيُجرِّب العامل إعادة المحاولة حتى ثلاث مرات. **استخدم هذا للمهام طويلة التشغيل** — بَذر مجموعات بيانات كبيرة، استدعاء واجهات برمجة تطبيقات خارجية بطيئة، تهيئة موارد خارجية، أو أي شيء قد يتجاوز نافذة استجابة HTTP المعقولة.
* `shouldRunSynchronously: true` — يُنفّذ الخطاف **ضمن تدفّق التثبيت مباشرةً** (نفس المنفِّذ كما قبل التثبيت). يَحجُب طلب التثبيت حتى ينتهي المعالج، وإذا رمى استثناءً، سيتلقى مستدعي التثبيت `POST_INSTALL_ERROR`. لا توجد محاولات إعادة تلقائية. **استخدم هذا للمهام السريعة التي يجب إكمالها قبل الاستجابة** — مثل إظهار خطأ تحقق للمستخدم، أو إعداد سريع سيعتمد عليه العميل مباشرةً بعد عودة نداء التثبيت. ضع في اعتبارك أن ترحيل البيانات الوصفية يكون قد طُبِّق بالفعل عند تشغيل ما بعد التثبيت، لذلك فإن فشل الوضع المتزامن **لا** يعيد التغييرات على المخطط إلى الوراء — بل يكتفي بإبراز الخطأ.
* تأكّد من أن معالجك قابل للتنفيذ المتكرر دون آثار جانبية. في الوضع غير المتزامن قد تُعيد قائمة الانتظار المحاولة حتى ثلاث مرات؛ وفي أي من الوضعين قد يعمل الخطاف مجددًا أثناء الترقيات عند ضبط `shouldRunOnVersionUpgrade: true`.
* متغيرات البيئة `APPLICATION_ID` و`APP_ACCESS_TOKEN` و`API_URL` متاحة داخل المعالج (كما في أي دالة منطق أخرى)، لذا يمكنك استدعاء واجهة Twenty API باستخدام رمز وصول للتطبيق مقيّد بنطاق تطبيقك.
* يُسمح بدالة ما بعد التثبيت واحدة فقط لكل تطبيق. سيُنتج إنشاء ملف البيان خطأً إذا تم اكتشاف أكثر من واحدة.
* يتم تعيين `universalIdentifier` للدالة تلقائيًا كـ `postInstallLogicFunctionUniversalIdentifier` في بيان التطبيق أثناء الإنشاء — لست بحاجة إلى الإشارة إليه في `defineApplication()`.
* تُرفَق خصائص الدالة `universalIdentifier` و`shouldRunOnVersionUpgrade` و`shouldRunSynchronously` تلقائيًا ببيان التطبيق ضمن الحقل `postInstallLogicFunction` أثناء عملية البناء — ولا تحتاج إلى الإشارة إليها في `defineApplication()`.
* تم تعيين مهلة افتراضية إلى 300 ثانية (5 دقائق) للسماح بمهام الإعداد الأطول مثل تهيئة البيانات.
* **لا يُنفَّذ في وضع التطوير**: عند تسجيل تطبيق محليًا (عبر `yarn twenty dev`)، يتجاوز الخادم تدفّق التثبيت بالكامل ويُزامن الملفات مباشرةً عبر مراقِب CLI — لذا لن يعمل ما بعد التثبيت في وضع التطوير مطلقًا، بغضّ النظر عن `shouldRunSynchronously`. استخدم `yarn twenty exec --postInstall` لتشغيله يدويًا على مساحة عمل قيد التشغيل.
</Accordion>
<Accordion title="definePreInstallLogicFunction" description="تعريف دالة منطقية لما قبل التثبيت (واحدة لكل تطبيق)">
دالة ما قبل التثبيت هي دالة منطقية تعمل تلقائيًا أثناء التثبيت، **قبل تطبيق ترحيل البيانات الوصفية لمساحة العمل**. تتشارك نفس بنية الحمولة مع ما بعد التثبيت (`InstallPayload`)، لكنها موضوعة أبكر في تدفّق التثبيت كي تجهّز حالة يعتمد عليها الترحيل القادم — ومن الاستخدامات الشائعة: نسخ البيانات احتياطيًا، التحقق من التوافق مع المخطط الجديد، أو أرشفة السجلات التي ستُعاد هيكلتها أو ستُحذف.
```ts src/logic-functions/pre-install.ts
import { definePreInstallLogicFunction, type InstallPayload } from 'twenty-sdk';
description: 'Runs before installation to prepare the application.',
timeoutSeconds: 300,
shouldRunOnVersionUpgrade: true,
handler,
});
```
يمكنك أيضًا تنفيذ دالة ما قبل التثبيت يدويًا في أي وقت باستخدام CLI:
```bash filename="Terminal"
yarn twenty exec --preInstall
```
النقاط الرئيسية:
* تستخدم دوال ما قبل التثبيت `definePreInstallLogicFunction()` — نفس الإعدادات المتخصصة كما في ما بعد التثبيت، لكنها مرتبطة بموضع مختلف ضمن دورة الحياة.
* يتلقّى كلٌّ من معالجي ما قبل التثبيت وما بعد التثبيت النوع نفسه `InstallPayload`: `{ previousVersion?: string; newVersion: string }`. استورده مرة واحدة وأعد استخدامه لكلا الخطافين.
* **موعد تشغيل الخطاف**: موضوع مباشرةً قبل ترحيل البيانات الوصفية لمساحة العمل (`synchronizeFromManifest`). قبل التنفيذ، يُشغِّل الخادم مزامنة "pared-down sync" ذات طابع إضافي فقط تقوم بتسجيل دالة ما قبل التثبيت للإصدار **الجديد** في البيانات الوصفية لمساحة العمل — دون لمس أي شيء آخر — ثم يُنفّذها. لأن هذه المزامنة «إضافية فقط»، تبقى كائنات وحقول وبيانات الإصدار السابق سليمة عند تشغيل معالجك: يمكنك قراءة حالة ما قبل الترحيل ونسخها احتياطيًا بأمان.
* **نموذج التنفيذ**: يُنفَّذ ما قبل التثبيت **بشكل متزامن** و**يحجب عملية التثبيت**. إذا رمى المعالج استثناءً، تُلغى عملية التثبيت قبل تطبيق أي تغييرات على المخطط — وتبقى مساحة العمل على الإصدار السابق بحالة متّسقة. هذا مقصود: ما قبل التثبيت هو فرصتك الأخيرة لرفض ترقية تنطوي على مخاطر.
* كما هو الحال مع ما بعد التثبيت، يُسمح بدالة ما قبل التثبيت واحدة فقط لكل تطبيق. تُربَط تلقائيًا ببيان التطبيق تحت `preInstallLogicFunction` أثناء عملية البناء.
* **لا يُنفَّذ في وضع التطوير**: كما في ما بعد التثبيت — يتم تجاوز تدفّق التثبيت بالكامل للتطبيقات المسجّلة محليًا، لذا لن يعمل ما قبل التثبيت مطلقًا عند `yarn twenty dev`. استخدم `yarn twenty exec --preInstall` لتشغيله يدويًا.
</Accordion>
<Accordion title="ما قبل التثبيت مقابل ما بعد التثبيت: متى تستخدم أيّهما" description="اختيار خطاف التثبيت المناسب">
كلا الخطافين جزء من تدفّق التثبيت نفسه ويتلقّيان نفس `InstallPayload`. الاختلاف يكمن في **موعد** تشغيلهما نسبةً إلى ترحيل البيانات الوصفية لمساحة العمل، وهذا يغيّر البيانات التي يمكنهما التعامل معها بأمان.
ما قبل التثبيت دائمًا **متزامن** (يحجب التثبيت ويمكنه إحباطه). ما بعد التثبيت **غير متزامن افتراضيًا** — يُدرج على عامل مع محاولات إعادة تلقائية — لكن يمكن التبديل إلى تنفيذ متزامن عبر `shouldRunSynchronously: true`. راجع الأكورديون `definePostInstallLogicFunction` أعلاه لمعرفة متى تستخدم كل وضع.
**استخدم `post-install` لأي شيء يتطلّب وجود المخطط الجديد.** وهذا هو السيناريو الشائع:
description: 'Seeds a welcome post card after install.',
timeoutSeconds: 300,
shouldRunOnVersionUpgrade: false,
handler,
});
```
**استخدم `pre-install` عندما قد يُتلف الترحيل أو يدمّر البيانات الحالية.** لأن ما قبل التثبيت يعمل مقابل المخطط *السابق* وفشله يُرجِع الترقية إلى الوراء، فهو المكان المناسب لأي شيء محفوف بالمخاطر:
* **نسخ البيانات احتياطيًا قبل حذفها أو إعادة هيكلتها** — مثل إزالة حقل في v2 وتحتاج إلى نسخ قيمه إلى حقل آخر أو تصديرها إلى التخزين قبل تشغيل الترحيل.
* **أرشفة السجلات التي سيبطلها قيد جديد** — مثل أن يصبح حقل ما `NOT NULL` وتحتاج أولًا إلى حذف الصفوف ذات القيم الفارغة أو إصلاحها.
* **التحقق من التوافق ورفض الترقية إذا تعذّر ترحيل البيانات الحالية بسلاسة** — ارمِ من داخل المعالج وسيُلغى التثبيت دون تطبيق أي تغييرات. هذا أكثر أمانًا من اكتشاف عدم التوافق في منتصف الترحيل.
* **إعادة تسمية البيانات أو إعادة تعيين مفاتيحها** قبل تغيير في المخطط قد يؤدي إلى فقدان الارتباط.
مثال — أرشف السجلات قبل ترحيل هدّام:
```ts src/logic-functions/pre-install.ts
import { definePreInstallLogicFunction, type InstallPayload } from 'twenty-sdk';
import { createClient } from './generated/client';
| بذر بيانات افتراضية، تهيئة مساحة العمل، تسجيل موارد خارجية | `post-install` |
| تشغيل بذر طويل الأمد أو استدعاءات أطراف ثالثة لا ينبغي أن تحجب استجابة التثبيت | `post-install` (الإعداد الافتراضي — `shouldRunSynchronously: false`، مع محاولات إعادة من العامل) |
| تشغيل إعداد سريع سيعتمد عليه المستدعي مباشرةً بعد عودة نداء التثبيت | `post-install` مع `shouldRunSynchronously: true` |
| قراءة البيانات أو نسخها احتياطيًا والتي قد يفقدها الترحيل القادم | `pre-install` |
| رفض ترقية قد تُفسد البيانات الحالية | `pre-install` (ارمِ من المعالج) |
| تنفيذ مواءمة في كل ترقية | `post-install` مع `shouldRunOnVersionUpgrade: true` |
| تنفيذ إعداد لمرة واحدة في التثبيت الأول فقط | `post-install` مع `shouldRunOnVersionUpgrade: false` (الإعداد الافتراضي) |
<Note>
إذا ساورك الشك، فاجعل الافتراضي هو **post-install**. الجأ إلى ما قبل التثبيت فقط عندما يكون الترحيل نفسه هدّامًا وتحتاج إلى التقاط الحالة السابقة قبل أن تزول.
يتم الاحتفاظ بالبيانات عبر عمليات إعادة التشغيل في وحدتي تخزين Docker (`twenty-app-dev-data` لـ PostgreSQL، و`twenty-app-dev-storage` للملفات). استخدم `reset` لمسح كل شيء والبدء من جديد.
### تشغيل مثيل الاختبار
مرر `--test` إلى أي أمر `server` لإدارة مثيل ثانٍ معزول تمامًا — مفيد لتشغيل اختبارات التكامل أو للتجربة من دون لمس بيانات التطوير الرئيسية لديك.
| `yarn twenty server reset --test` | محو بيانات الاختبار والبدء من جديد |
يعمل مثيل الاختبار في حاوية Docker خاصة به (`twenty-app-dev-test`) مع وحدات تخزين مخصصة (`twenty-app-dev-test-data`, `twenty-app-dev-test-storage`) وتهيئة مستقلة، بحيث يمكنه العمل بالتوازي مع مثيلك الرئيسي من دون تعارضات. اجمع `--test` مع `--port` لتجاوز القيمة الافتراضية 2021.
<Note>
يتطلّب الخادم أن يكون **Docker** قيد التشغيل. إذا ظهرت لك رسالة خطأ "Docker not running"، فتأكّد من تشغيل Docker Desktop (أو خادوم Docker).
عند تحديث تطبيق tarball منشور مسبقًا، يشترط الخادم أن تكون قيمة `version` في `package.json` **أعلى قطعًا** (وفق ترتيب [الإصدار الدلالي](https://semver.org)) من الإصدار المنشور حاليًا. إعادة نشر الإصدار نفسه، أو دفع إصدار أدنى، يُرفَض قبل تخزين ملف tarball — سترى خطأ `VERSION_ALREADY_EXISTS` من CLI.
لطرح تحديث:
1. ارفع قيمة الحقل `version` في ملف `package.json`
1. قم بزيادة الحقل `version` في ملف `package.json` (مثلًا: `1.2.3` → `1.2.4`، `1.3.0`، أو `2.0.0`)
3. سترى مساحات العمل التي ثبّتت التطبيق الترقية متاحة في إعداداتها
<Note>
علامات ما قبل الإصدار تعمل كما هو متوقع: زيادة `1.0.0-rc.1` → `1.0.0-rc.2` مسموح بها، ويُعترَف بالإصدار النهائي مثل `1.0.0` على أنه أعلى من `1.0.0-rc.5`. يجب أن يكون الإصدار في `package.json` بنفسه سلسلة semver صالحة.
</Note>
{/* TODO: add screenshot of the Upgrade button */}
## CI/CD المؤتمتة (مهام سير عمل مُولَّدة بالقوالب)
التطبيقات المُولَّدة باستخدام `create-twenty-app` تأتي افتراضيًا مع مهمَّتي سير عمل من GitHub Actions ضمن `.github/workflows/`. هي جاهزة للتشغيل بمجرد دفع المستودع إلى GitHub — لا حاجة لأي إعداد إضافي لـ CI، وCD يتطلّب سرًّا واحدًا فقط.
### CI — `ci.yml`
يشغّل اختبارات التكامل عند كل دفع إلى `main` وعند كل طلب سحب.
**ماذا يفعل:**
1. يجلب مصدر تطبيقك.
2. ينشئ مثيلاً اختبارياً معزولاً من Twenty باستخدام الإجراء المركّب `twentyhq/twenty/.github/actions/spawn-twenty-app-dev-test@main` (المكافئ في CI للأمر `yarn twenty server start --test`).
3. يُفعِّل Corepack، ويُعدّ Node.js من ملف `.nvmrc` لديك، ويثبّت التبعيات بواسطة `yarn install --immutable`.
4. يشغّل `yarn test`، ويمرّر `TWENTY_API_URL` و`TWENTY_API_KEY` من المثيل الذي تم إنشاؤه بحيث تتمكّن اختباراتك من التواصل مع خادم حقيقي.
**خيارات التكوين:**
* `TWENTY_VERSION` (متغيّر بيئة، القيمة الافتراضية `latest`) — ثبّت نسخة خادم Twenty المستخدمة في CI عبر تعديل هذا في `ci.yml`.
* يتم تجميع التشغيل المتزامن حسب `github.ref` ويلغي التشغيلات قيد التقدّم عند أي دفع جديد.
لا تتطلّب أي أسرار — مثيل الاختبار مؤقّت ويستمر فقط طوال مدّة المهمّة.
### CD — `cd.yml`
ينشر تطبيقك إلى خادم Twenty مُهيّأ عند كل دفع إلى `main`، وبشكل اختياري من طلب سحب عند تطبيق الوسم `deploy`.
**ماذا يفعل:**
1. يجلب رأس طلب السحب (للطلبات الموسومة) أو الالتزام المدفوع.
2. يشغّل `twentyhq/twenty/.github/actions/deploy-twenty-app@main` — وهو المكافئ في CI للأمر `yarn twenty deploy`.
3. يشغّل `twentyhq/twenty/.github/actions/install-twenty-app@main` بحيث تُثبَّت النسخة المُنشَرة حديثًا في مساحة العمل المستهدفة.
| `TWENTY_DEPLOY_URL` | `env` في `cd.yml` (القيمة الافتراضية `http://localhost:3000`) | خادم Twenty الذي سيتم النشر إليه. غيّر هذا إلى عنوان URL لخادمك الحقيقي قبل أول استخدام. |
| `TWENTY_DEPLOY_API_KEY` | في مستودع GitHub **Settings → Secrets and variables → Actions** | مفتاح API يمتلك إذن النشر على الخادم المستهدف. |
<Note>
القيمة الافتراضية لـ `TWENTY_DEPLOY_URL` وهي `http://localhost:3000` مجرد عنصر نائب — لن تصل إلى أي شيء من مُشغِّل مستضاف لدى GitHub. حدّثها إلى عنوان URL العام لخادمك (أو استخدم مُشغِّلًا مستضافًا ذاتيًا مع وصول شبكي) قبل تمكين CD.
</Note>
**تشغيل نشر معاينة من طلب سحب:**
أضِف الوسم `deploy` إلى طلب سحب. الشرط `if:` في `cd.yml` سيشغّل المهمّة لذلك الطلب مستخدمًا التزام رأس الطلب، مما يتيح لك التحقّق من التغيير على الخادم المستهدف قبل الدمج.
### تثبيت الإجراءات القابلة لإعادة الاستخدام
يشير كلا سيرَي العمل إلى إجراءات قابلة لإعادة الاستخدام عند `@main`، لذا تُلتقط تحديثات الإجراءات في مستودع `twentyhq/twenty` تلقائيًا. إذا كنت تريد بناءات حتمية، فاستبدِل `@main` بقيمة SHA لالتزام أو بوسم إصدار في كل سطر `uses:`.
## النشر على npm
يُتيح النشر على npm إمكانية العثور على تطبيقك في سوق Twenty. يمكن لأي مساحة عمل في Twenty استعراض تطبيقات السوق وتثبيتها وترقيتها مباشرةً من واجهة المستخدم.
@@ -81,7 +138,7 @@ yarn twenty deploy
### المتطلبات
* حساب على [npm](https://www.npmjs.com)
* الكلمة المفتاحية `twenty-app` في مصفوفة `keywords` في `package.json` (موجودة مسبقًا عند تهيئة المشروع باستخدام `create-twenty-app`)
* الكلمة المفتاحية `twenty-app` في مصفوفة `keywords` في `package.json` (أضفها يدويًا — فهي غير مضمنة افتراضيًا في قالب `create-twenty-app`)
```json filename="package.json"
{
@@ -189,3 +246,12 @@ jobs:
```bash filename="Terminal"
yarn twenty install
```
<Note>
يفرض الخادم اعتماد إصدارات semver عند التثبيت، بما يعكس القواعد المطبّقة عند النشر:
* تثبيت الإصدار نفسه المثبّت بالفعل في مساحة عملك يُرفَض بخطأ `APP_ALREADY_INSTALLED`.
* تثبيت إصدار أدنى من الإصدار المثبّت حاليًا يُرفَض بخطأ `CANNOT_DOWNGRADE_APPLICATION`.
لتثبيت إصدار أحدث، انشره (deploy) أو انشره إلى السجل (publish) أولًا، ثم أعد تشغيل `yarn twenty install`.
**Napište kvalitní `description`.** Agenti AI se spoléhají na pole funkce `description` při rozhodování, kdy nástroj použít. Buďte konkrétní ohledně toho, co nástroj dělá a kdy se má volat.
</Note>
</Accordion>
<Accordion title="definePreInstallLogicFunction" description="Definujte předinstalační logickou funkci (jedna na aplikaci)">
Předinstalační funkce je logická funkce, která se automaticky spouští před instalací vaší aplikace v pracovním prostoru. To je užitečné pro validační úlohy, kontrolu předpokladů nebo přípravu stavu pracovního prostoru před zahájením hlavní instalace.
```ts src/logic-functions/pre-install.ts
import { definePreInstallLogicFunction, type InstallLogicFunctionPayload } from 'twenty-sdk';
description: 'Runs before installation to prepare the application.',
timeoutSeconds: 300,
handler,
});
```
Předinstalační funkci můžete také kdykoli spustit ručně pomocí CLI:
```bash filename="Terminal"
yarn twenty exec --preInstall
```
Hlavní body:
* Předinstalační funkce používají `definePreInstallLogicFunction()` — specializovanou variantu, která vynechává nastavení spouštěčů (`cronTriggerSettings`, `databaseEventTriggerSettings`, `httpRouteTriggerSettings`, `isTool`).
* Obslužná funkce (handler) obdrží `InstallLogicFunctionPayload` s `{ previousVersion: string }` — verzi aplikace, která byla dříve nainstalována (nebo prázdný řetězec při čisté instalaci).
* Na jednu aplikaci je povolena pouze jedna předinstalační funkce. Sestavení manifestu skončí chybou, pokud je zjištěna více než jedna.
* Identifikátor `universalIdentifier` funkce se během sestavení automaticky nastaví v manifestu aplikace jako `preInstallLogicFunctionUniversalIdentifier` — není potřeba jej uvádět v `defineApplication()`.
* Výchozí časový limit je nastaven na 300 sekund (5 minut), aby umožnil delší přípravné úlohy.
</Accordion>
<Accordion title="definePostInstallLogicFunction" description="Definujte postinstalační logickou funkci (jedna na aplikaci)">
Postinstalační funkce je logická funkce, která se automaticky spouští po instalaci vaší aplikace do pracovního prostoru. To je užitečné pro jednorázové úlohy nastavení, jako je naplnění výchozími daty, vytvoření počátečních záznamů nebo konfigurace nastavení pracovního prostoru.
Postinstalační funkce je logická funkce, která se spustí automaticky, jakmile je instalace vaší aplikace v pracovním prostoru dokončena. Server ji provede **poté**, co byla synchronizována metadata aplikace a vygenerován klient SDK, takže je pracovní prostor plně připraven k použití a nové schéma je zavedeno. Mezi typické případy použití patří naplnění výchozími daty, vytvoření počátečních záznamů, konfigurace nastavení pracovního prostoru nebo zřizování prostředků ve službách třetích stran.
```ts src/logic-functions/post-install.ts
import { definePostInstallLogicFunction, type InstallLogicFunctionPayload } from 'twenty-sdk';
import { definePostInstallLogicFunction, type InstallPayload } from 'twenty-sdk';
* Postinstalační funkce používají `definePostInstallLogicFunction()` — specializovanou variantu, která vynechává nastavení spouštěčů (`cronTriggerSettings`, `databaseEventTriggerSettings`, `httpRouteTriggerSettings`, `isTool`).
* Obslužná funkce (handler) obdrží `InstallLogicFunctionPayload` s `{ previousVersion: string }` — verzi aplikace, která byla dříve nainstalována (nebo prázdný řetězec při čisté instalaci).
* Obslužná funkce obdrží `InstallPayload` s `{ previousVersion?: string; newVersion: string }` — `newVersion` je verze, která se instaluje, a `previousVersion` je verze, která byla nainstalována dříve (nebo `undefined` při čisté instalaci). Tyto hodnoty použijte k rozlišení čistých instalací od aktualizací a ke spuštění migrační logiky specifické pro verzi.
* **Kdy se hook spouští**: ve výchozím nastavení pouze při čistých instalacích. Předejte `shouldRunOnVersionUpgrade: true`, pokud chcete, aby se spouštěl i při aktualizaci aplikace z předchozí verze. Pokud je vynechán, příznak má výchozí hodnotu `false` a při aktualizacích se hook přeskočí.
* **Model provádění — ve výchozím nastavení asynchronní, synchronní volitelně**: příznak `shouldRunSynchronously` určuje *jak* se spouští post-install.
* `shouldRunSynchronously: false` *(výchozí)* — hook je **zařazen do fronty zpráv** s `retryLimit: 3` a běží asynchronně ve workeru. Odezva instalace se vrátí hned po zařazení úlohy do fronty, takže pomalá nebo chybující obslužná funkce neblokuje volajícího. Worker se pokusí o opakování až třikrát. **Použijte pro dlouho běžící úlohy** — plnění velkých datových sad, volání pomalých externích API, zřizování externích prostředků, cokoli, co by mohlo přesáhnout rozumné časové okno HTTP odezvy.
* `shouldRunSynchronously: true` — hook se provádí **inline během instalačního procesu** (stejný vykonavatel jako pre-install). Instalační požadavek blokuje, dokud obslužná funkce nedokončí, a pokud vyvolá výjimku, volající instalace obdrží `POST_INSTALL_ERROR`. Žádné automatické opakování. **Použijte pro rychlé úlohy, které se musí dokončit před odpovědí** — například vrácení validační chyby uživateli nebo rychlé nastavení, na kterém bude klient záviset ihned po návratu volání instalace. Mějte na paměti, že v době, kdy se spustí post-install, už byla migrace metadat aplikována, takže selhání v synchronním režimu změny schématu **ne**vrací zpět — pouze odhalí chybu.
* Ujistěte se, že vaše obslužná funkce je idempotentní. V asynchronním režimu se může fronta pokusit až třikrát; v obou režimech se může hook znovu spustit při aktualizacích, pokud je `shouldRunOnVersionUpgrade: true`.
* Proměnné prostředí `APPLICATION_ID`, `APP_ACCESS_TOKEN` a `API_URL` jsou dostupné uvnitř obslužné funkce (stejně jako u jakékoli jiné logické funkce), takže můžete volat Twenty API s aplikačním přístupovým tokenem omezeným na vaši aplikaci.
* Na jednu aplikaci je povolena pouze jedna postinstalační funkce. Sestavení manifestu skončí chybou, pokud je zjištěna více než jedna.
* Identifikátor `universalIdentifier` funkce se během sestavení automaticky nastaví v manifestu aplikace jako `postInstallLogicFunctionUniversalIdentifier` — není potřeba jej uvádět v `defineApplication()`.
* Atributy funkce `universalIdentifier`, `shouldRunOnVersionUpgrade` a `shouldRunSynchronously` jsou během buildu automaticky připojeny k manifestu aplikace do pole `postInstallLogicFunction` — není potřeba je uvádět v `defineApplication()`.
* Výchozí časový limit je nastaven na 300 sekund (5 minut), aby umožnil delší úlohy nastavení, jako je naplnění daty.
* **Nespouští se v režimu dev**: když je aplikace registrována lokálně (pomocí `yarn twenty dev`), server zcela přeskočí instalační tok a synchronizuje soubory přímo prostřednictvím sledovače CLI — takže se post-install v režimu dev nikdy nespustí bez ohledu na `shouldRunSynchronously`. Použijte `yarn twenty exec --postInstall` k ručnímu spuštění nad běžícím pracovním prostorem.
</Accordion>
<Accordion title="definePreInstallLogicFunction" description="Definujte předinstalační logickou funkci (jedna na aplikaci)">
Funkce pre-install je logická funkce, která se během instalace spouští automaticky, **před aplikováním migrace metadat pracovního prostoru**. Má stejný tvar payloadu jako post-install (`InstallPayload`), ale je zařazena dříve v instalačním toku, aby mohla připravit stav, na němž nadcházející migrace závisí — typické použití zahrnuje zálohování dat, ověření kompatibility s novým schématem nebo archivaci záznamů, které se chystají přeuspořádat nebo odstranit.
```ts src/logic-functions/pre-install.ts
import { definePreInstallLogicFunction, type InstallPayload } from 'twenty-sdk';
description: 'Runs before installation to prepare the application.',
timeoutSeconds: 300,
shouldRunOnVersionUpgrade: true,
handler,
});
```
Předinstalační funkci můžete také kdykoli spustit ručně pomocí CLI:
```bash filename="Terminal"
yarn twenty exec --preInstall
```
Hlavní body:
* Funkce pre-install používají `definePreInstallLogicFunction()` — stejné specializované nastavení jako u post-install, pouze připojené k jiné fázi životního cyklu.
* Obě obslužné funkce pre- i post-install přijímají stejný typ `InstallPayload`: `{ previousVersion?: string; newVersion: string }`. Importujte jej jednou a znovu použijte pro oba hooky.
* **Kdy se hook spouští**: umístěn těsně před migrací metadat pracovního prostoru (`synchronizeFromManifest`). Před spuštěním server provede čistě aditivní "zjednodušenou synchronizaci", která v metadatech pracovního prostoru zaregistruje pre-install funkci **nové** verze — ničeho dalšího se nedotkne — a poté ji spustí. Protože tato synchronizace je pouze aditivní, objekty, pole a data předchozí verze zůstávají při spuštění vaší obslužné funkce zachována: můžete bezpečně číst a zálohovat stav před migrací.
* **Model provádění**: pre-install se provádí **synchronně** a **blokuje instalaci**. Pokud obslužná funkce vyvolá výjimku, instalace se přeruší ještě před aplikováním jakýchkoli změn schématu — pracovní prostor zůstane na předchozí verzi v konzistentním stavu. Je to záměrné: pre-install je vaše poslední šance odmítnout rizikovou aktualizaci.
* Stejně jako u post-install je na jednu aplikaci povolena pouze jedna funkce pre-install. Během buildu je automaticky připojena k manifestu aplikace pod `preInstallLogicFunction`.
* **Nespouští se v režimu dev**: stejně jako u post-install — u lokálně registrovaných aplikací je instalační tok zcela přeskočen, takže se pre-install pod `yarn twenty dev` nikdy nespustí. Použijte `yarn twenty exec --preInstall` k ručnímu spuštění.
</Accordion>
<Accordion title="Pre-install vs post-install: kdy použít který" description="Výběr správného instalačního hooku">
Oba hooky jsou součástí téhož instalačního toku a přijímají stejný `InstallPayload`. Rozdíl je v tom, **kdy** se spouštějí vzhledem k migraci metadat pracovního prostoru, a to určuje, jakých dat se mohou bezpečně dotýkat.
Pre-install je vždy **synchronní** (blokuje instalaci a může ji přerušit). Post-install je **ve výchozím nastavení asynchronní** — zařazen do workeru s automatickými pokusy o opakování — ale může přejít na synchronní provádění pomocí `shouldRunSynchronously: true`. Viz accordion `definePostInstallLogicFunction` výše, kdy použít jednotlivé režimy.
**Použijte `post-install` pro cokoli, co vyžaduje existenci nového schématu.** To je běžný případ:
* Plnění výchozími daty (vytváření počátečních záznamů, výchozích pohledů, demo obsahu) vůči nově přidaným objektům a polím.
* Registrace webhooků u služeb třetích stran poté, co má aplikace své přihlašovací údaje.
* Volání vlastního API k dokončení nastavení, které závisí na synchronizovaných metadatech.
* Idempotentní logika "zajisti, že to existuje", která má při každé aktualizaci uvést stav do souladu — kombinujte s `shouldRunOnVersionUpgrade: true`.
Příklad — po instalaci naplňte výchozí záznam `PostCard`:
```ts src/logic-functions/post-install.ts
import { definePostInstallLogicFunction, type InstallPayload } from 'twenty-sdk';
import { createClient } from './generated/client';
description: 'Seeds a welcome post card after install.',
timeoutSeconds: 300,
shouldRunOnVersionUpgrade: false,
handler,
});
```
**Použijte `pre-install`, pokud by migrace jinak zničila nebo poškodila existující data.** Protože pre-install běží proti *předchozímu* schématu a jeho selhání vrací aktualizaci zpět, je to správné místo pro cokoli rizikového:
* **Zálohování dat, která se chystají odstranit nebo přeuspořádat** — např. odstraňujete pole ve verzi v2 a potřebujete jeho hodnoty zkopírovat do jiného pole nebo je před spuštěním migrace exportovat do úložiště.
* **Archivace záznamů, které by nové omezení zneplatnilo** — např. pole se stává `NOT NULL` a je třeba nejprve smazat nebo opravit řádky s hodnotami null.
* **Ověření kompatibility a odmítnutí aktualizace, pokud nelze aktuální data čistě migrovat** — vyhoďte výjimku z obslužné funkce a instalace se ukončí bez provedených změn. Je to bezpečnější, než zjistit nekompatibilitu uprostřed migrace.
* **Přejmenování nebo změna klíčů dat** před změnou schématu, která by ztratila vazby.
Příklad — archivujte záznamy před destruktivní migrací:
```ts src/logic-functions/pre-install.ts
import { definePreInstallLogicFunction, type InstallPayload } from 'twenty-sdk';
import { createClient } from './generated/client';
| Naplňte výchozí data, nakonfigurujte pracovní prostor, zaregistrujte externí prostředky | `post-install` |
| Spusťte dlouho běžící plnění nebo volání třetích stran, která by neměla blokovat odezvu instalace | `post-install` (výchozí — `shouldRunSynchronously: false`, s opakovanými pokusy workeru) |
| Spusťte rychlé nastavení, na které bude volající spoléhat ihned po návratu volání instalace | `post-install` s `shouldRunSynchronously: true` |
| Čtěte nebo zálohujte data, která by nadcházející migrace ztratila | `pre-install` |
| Odmítněte aktualizaci, která by poškodila existující data | `pre-install` (vyhoďte výjimku z obslužné funkce) |
| Spouštějte srovnání stavu při každé aktualizaci | `post-install` s `shouldRunOnVersionUpgrade: true` |
| Proveďte jednorázové nastavení pouze při první instalaci | `post-install` s `shouldRunOnVersionUpgrade: false` (výchozí) |
<Note>
Pokud si nejste jisti, výchozí volbou je **post-install**. Po pre-install sáhněte pouze tehdy, když je samotná migrace destruktivní a potřebujete zachytit předchozí stav, než zmizí.
</Note>
</Accordion>
<Accordion title="defineFrontComponent" description="Definujte frontendové komponenty pro vlastní uživatelské rozhraní">
| `yarn twenty server start` | Spustí lokální server (v případě potřeby stáhne image) |
| `yarn twenty server start --port 3030` | Spustí na vlastním portu |
| `yarn twenty server start --test` | Spusťte samostatnou testovací instanci na portu 2021 |
| `yarn twenty server stop` | Zastaví server (zachová data) |
| `yarn twenty server status` | Zobrazí stav serveru, URL a přihlašovací údaje |
| `yarn twenty server logs` | Streamuje protokoly serveru |
@@ -228,6 +229,20 @@ Nástroj pro vytvoření kostry vám již spustil lokální server Twenty. Pro j
Data přetrvávají při restartech ve dvou svazcích Dockeru (`twenty-app-dev-data` pro PostgreSQL, `twenty-app-dev-storage` pro soubory). Pomocí `reset` vymažte vše a začněte znovu.
### Spuštění testovací instance
Předejte volbu `--test` libovolnému příkazu `server` pro správu druhé, plně izolované instance — užitečné pro spouštění integračních testů nebo experimentování, aniž byste se dotkli svých hlavních vývojových dat.
| `yarn twenty server reset --test` | Vymaže testovací data a začne znovu |
Testovací instance běží ve vlastním kontejneru Docker (`twenty-app-dev-test`) s vyhrazenými svazky (`twenty-app-dev-test-data`, `twenty-app-dev-test-storage`) a konfigurací, takže může běžet paralelně s vaší hlavní instancí bez konfliktů. Zkombinujte `--test` s `--port` pro změnu výchozího portu 2021.
<Note>
Server vyžaduje, aby **Docker** běžel. Pokud vidíte chybu "Docker not running", ujistěte se, že je spuštěný Docker Desktop (nebo démon Dockeru).
@@ -66,14 +66,71 @@ Odkaz ke sdílení používá základní adresu URL serveru (bez jakékoli subdo
### Správa verzí
Při aktualizaci již nasazené tarballové aplikace server vyžaduje, aby hodnota `version` v `package.json` byla **přísně vyšší** (podle řazení [semver](https://semver.org)) než aktuálně nasazená verze. Opětovné nasazení stejné verze nebo odeslání nižší verze je odmítnuto ještě před uložením tarballu — v CLI uvidíte chybu `VERSION_ALREADY_EXISTS`.
Chcete-li vydat aktualizaci:
1. Zvyšte hodnotu pole `version` v souboru `package.json`
1. Zvyšte hodnotu pole `version` v souboru `package.json` (např. `1.2.3` → `1.2.4`, `1.3.0` nebo `2.0.0`)
3. Pracovní prostory, které mají aplikaci nainstalovanou, uvidí dostupnou aktualizaci ve svém nastavení
<Note>
Předběžné tagy fungují podle očekávání: zvýšení z `1.0.0-rc.1` → `1.0.0-rc.2` je povoleno a finální vydání jako `1.0.0` je správně rozpoznáno jako vyšší než `1.0.0-rc.5`. Verze v `package.json` musí být platným řetězcem semver.
</Note>
{/* TODO: add screenshot of the Upgrade button */}
Aplikace vygenerované pomocí `create-twenty-app` jsou hned připravené se dvěma workflowy GitHub Actions ve složce `.github/workflows/`. Jsou připravené ke spuštění hned, jakmile repozitář pushnete na GitHub — pro CI není potřeba žádné další nastavení a CD vyžaduje pouze jeden secret.
### CI — `ci.yml`
Automaticky spouští integrační testy při každém pushi do `main` a u pull requestů.
2. Spustí izolovanou testovací instanci Twenty pomocí složené akce `twentyhq/twenty/.github/actions/spawn-twenty-app-dev-test@main` (ekvivalent v CI k `yarn twenty server start --test`).
3. Povolí Corepack, nastaví Node.js podle vašeho `.nvmrc` a nainstaluje závislosti pomocí `yarn install --immutable`.
4. Spustí `yarn test` a předá `TWENTY_API_URL` a `TWENTY_API_KEY` ze spuštěné instance, aby vaše testy mohly komunikovat se skutečným serverem.
**Konfigurační volby:**
* `TWENTY_VERSION` (env, výchozí hodnota `latest`) — uzamkněte v CI používanou verzi serveru Twenty úpravou této hodnoty v `ci.yml`.
* Souběžné běhy jsou seskupeny podle `github.ref` a při nových pushích ruší právě probíhající běhy.
Nejsou potřeba žádné secrety — testovací instance je efemérní a existuje pouze po dobu běhu úlohy.
### CD — `cd.yml`
Nasazuje vaši aplikaci na nakonfigurovaný server Twenty při každém pushi do `main` a volitelně také z pull requestu, pokud je přidán štítek `deploy`.
**K čemu slouží:**
1. Provede checkout headu PR (u označených PR) nebo pushnutého commitu.
2. Spustí `twentyhq/twenty/.github/actions/deploy-twenty-app@main` — ekvivalent v CI k `yarn twenty deploy`.
3. Spustí `twentyhq/twenty/.github/actions/install-twenty-app@main`, aby se nově nasazená verze nainstalovala do cílového workspace.
| `TWENTY_DEPLOY_URL` | `env` v `cd.yml` (výchozí `http://localhost:3000`) | Server Twenty, na který se nasazuje. Před prvním použitím to změňte na skutečnou URL vašeho serveru. |
| `TWENTY_DEPLOY_API_KEY` | GitHub repozitář **Settings → Secrets and variables → Actions** | API klíč s oprávněním k nasazení na cílovém serveru. |
<Note>
Výchozí `TWENTY_DEPLOY_URL` `http://localhost:3000` je pouze zástupná hodnota — z runneru hostovaného GitHubem tato adresa nebude dosažitelná. Před povolením CD ji aktualizujte na veřejnou URL vašeho serveru (nebo použijte self-hosted runner s přístupem do sítě).
</Note>
**Spuštění náhledového nasazení z PR:**
Přidejte k pull requestu štítek `deploy`. Podmínka `if:` v `cd.yml` spustí úlohu pro dané PR s použitím head commitu PR, což vám umožní ověřit změnu na cílovém serveru před sloučením.
### Připnutí verzí znovupoužitelných akcí
Obě workflowy odkazují na znovupoužitelné akce na `@main`, takže aktualizace akcí v repozitáři `twentyhq/twenty` se přeberou automaticky. Pokud chcete deterministická sestavení, nahraďte `@main` v každém řádku `uses:` za commit SHA nebo tag vydání.
## Publikování na npm
Publikování na npm zajistí, že bude vaše aplikace dohledatelná v Marketplace Twenty. Jakýkoli pracovní prostor Twenty může procházet, instalovat a aktualizovat aplikace z Marketplace přímo z UI.
@@ -81,7 +138,7 @@ Publikování na npm zajistí, že bude vaše aplikace dohledatelná v Marketpla
### Požadavky
* Účet na [npm](https://www.npmjs.com)
* Klíčové slovo `twenty-app` ve vašem poli `keywords` v souboru `package.json` (již je zahrnuto, když založíte projekt pomocí `create-twenty-app`)
* Klíčové slovo `twenty-app` ve vašem poli `keywords` v souboru `package.json` (přidejte je ručně — ve výchozím nastavení není zahrnuto v šabloně `create-twenty-app`)
```json filename="package.json"
{
@@ -189,3 +246,12 @@ Aplikace můžete nainstalovat také z příkazového řádku:
```bash filename="Terminal"
yarn twenty install
```
<Note>
Server při instalaci vynucuje verzování semver a zrcadlí pravidla pro nasazení:
* Instalace stejné verze, která je již nainstalována ve vašem pracovním prostoru, je odmítnuta s chybou `APP_ALREADY_INSTALLED`.
* Instalace nižší verze, než je aktuálně nainstalovaná, je odmítnuta s chybou `CANNOT_DOWNGRADE_APPLICATION`.
K instalaci novější verze ji nejprve nasaďte nebo publikujte, poté znovu spusťte `yarn twenty install`.
**Schreiben Sie eine gute `description`.** KI-Agenten verlassen sich auf das `description`-Feld der Funktion, um zu entscheiden, wann das Tool verwendet werden soll. Seien Sie konkret darin, was das Tool tut und wann es aufgerufen werden soll.
</Note>
</Accordion>
<Accordion title="definePreInstallLogicFunction" description="Eine Pre-Installations-Logikfunktion definieren (eine pro App)">
Eine Pre-Installationsfunktion ist eine Logikfunktion, die automatisch ausgeführt wird, bevor Ihre App in einem Arbeitsbereich installiert wird. Dies ist nützlich für Validierungsaufgaben, Überprüfungen von Voraussetzungen oder die Vorbereitung des Status des Arbeitsbereichs, bevor die Hauptinstallation fortgesetzt wird.
```ts src/logic-functions/pre-install.ts
import { definePreInstallLogicFunction, type InstallLogicFunctionPayload } from 'twenty-sdk';
description: 'Runs before installation to prepare the application.',
timeoutSeconds: 300,
handler,
});
```
Sie können die Pre-Installationsfunktion auch jederzeit manuell über die CLI ausführen:
```bash filename="Terminal"
yarn twenty exec --preInstall
```
Hauptpunkte:
* Pre-Installationsfunktionen verwenden `definePreInstallLogicFunction()` — eine spezialisierte Variante, die Trigger-Einstellungen (`cronTriggerSettings`, `databaseEventTriggerSettings`, `httpRouteTriggerSettings`, `isTool`) weglässt.
* Der Handler erhält ein `InstallLogicFunctionPayload` mit `{ previousVersion: string }` — die Version der App, die zuvor installiert war (oder eine leere Zeichenkette bei Neuinstallationen).
* Pro Anwendung ist nur eine Pre-Installationsfunktion zulässig. Der Manifest-Build schlägt fehl, wenn mehr als eine erkannt wird.
* Der `universalIdentifier` der Funktion wird während des Builds im Anwendungsmanifest automatisch als `preInstallLogicFunctionUniversalIdentifier` gesetzt — Sie müssen ihn nicht in `defineApplication()` referenzieren.
* Das standardmäßige Timeout ist auf 300 Sekunden (5 Minuten) festgelegt, um längere Vorbereitungsvorgänge zu ermöglichen.
</Accordion>
<Accordion title="definePostInstallLogicFunction" description="Eine Post-Installations-Logikfunktion definieren (eine pro App)">
Eine Post-Installationsfunktion ist eine Logikfunktion, die automatisch ausgeführt wird, nachdem Ihre App in einem Arbeitsbereich installiert wurde. Dies ist nützlich für einmalige Einrichtungsvorgänge wie das Befüllen mit Standarddaten, das Erstellen erster Datensätze oder das Konfigurieren von Arbeitsbereichseinstellungen.
Eine Post-Installationsfunktion ist eine Logikfunktion, die automatisch ausgeführt wird, nachdem Ihre App in einem Arbeitsbereich installiert wurde. Der Server führt sie **nach** der Synchronisierung der Metadaten der App und der Generierung des SDK-Clients aus, sodass der Arbeitsbereich vollständig einsatzbereit ist und das neue Schema bereitsteht. Typische Anwendungsfälle umfassen das Befüllen von Standarddaten, das Erstellen anfänglicher Datensätze, das Konfigurieren von Arbeitsbereichseinstellungen oder das Bereitstellen von Ressourcen bei Diensten von Drittanbietern.
```ts src/logic-functions/post-install.ts
import { definePostInstallLogicFunction, type InstallLogicFunctionPayload } from 'twenty-sdk';
import { definePostInstallLogicFunction, type InstallPayload } from 'twenty-sdk';
* Post-Installationsfunktionen verwenden `definePostInstallLogicFunction()` — eine spezialisierte Variante, die Trigger-Einstellungen (`cronTriggerSettings`, `databaseEventTriggerSettings`, `httpRouteTriggerSettings`, `isTool`) weglässt.
* Der Handler erhält ein `InstallLogicFunctionPayload` mit `{ previousVersion: string }` — die Version der App, die zuvor installiert war (oder eine leere Zeichenkette bei Neuinstallationen).
* Der Handler erhält ein `InstallPayload` mit `{ previousVersion?: string; newVersion: string }` — `newVersion` ist die zu installierende Version, und `previousVersion` ist die zuvor installierte Version (oder `undefined` bei einer Neuinstallation). Verwenden Sie diese Werte, um Neuinstallationen von Upgrades zu unterscheiden und versionsspezifische Migrationslogik auszuführen.
* **Wann der Hook ausgeführt wird**: standardmäßig nur bei Neuinstallationen. Übergeben Sie `shouldRunOnVersionUpgrade: true`, wenn er auch beim Upgrade der App von einer vorherigen Version ausgeführt werden soll. Wenn weggelassen, ist das Flag standardmäßig `false` und Upgrades überspringen den Hook.
* **Ausführungsmodell — standardmäßig asynchron, synchron optional**: Das Flag `shouldRunSynchronously` steuert, *wie* Post-Install ausgeführt wird.
* `shouldRunSynchronously: false` *(Standard)* — der Hook wird **in die Nachrichtenwarteschlange eingereiht** mit `retryLimit: 3` und läuft asynchron in einem Worker. Die Installationsantwort kommt zurück, sobald der Job eingereiht ist, sodass ein langsamer oder fehlschlagender Handler den Aufrufer nicht blockiert. Der Worker versucht es bis zu dreimal erneut. **Verwenden Sie dies für lang laufende Jobs** — das Befüllen großer Datensätze, Aufrufe langsamer Drittanbieter-APIs, Bereitstellung externer Ressourcen, alles, was ein vernünftiges HTTP-Antwortfenster überschreiten könnte.
* `shouldRunSynchronously: true` — der Hook wird **inline während des Installationsablaufs** ausgeführt (gleicher Executor wie bei Pre-Install). Die Installationsanforderung blockiert, bis der Handler fertig ist, und wenn er einen Fehler wirft, erhält der Installationsaufrufer einen `POST_INSTALL_ERROR`. Keine automatischen Wiederholungen. **Verwenden Sie dies für schnelle Aufgaben, die vor der Antwort abgeschlossen sein müssen** — z. B. um dem Benutzer einen Validierungsfehler auszugeben oder für eine schnelle Einrichtung, auf die der Client unmittelbar nach der Rückkehr des Installationsaufrufs angewiesen ist. Beachten Sie, dass die Metadatenmigration bereits angewendet wurde, wenn Post-Install läuft, sodass ein Fehler im Synchronmodus die Schemaänderungen **nicht** rückgängig macht — er zeigt lediglich den Fehler an.
* Stellen Sie sicher, dass Ihr Handler idempotent ist. Im asynchronen Modus kann die Warteschlange bis zu dreimal erneut versuchen; in beiden Modi kann der Hook bei Upgrades erneut laufen, wenn `shouldRunOnVersionUpgrade: true`.
* Die Umgebungsvariablen `APPLICATION_ID`, `APP_ACCESS_TOKEN` und `API_URL` sind im Handler verfügbar (wie bei jeder anderen Logikfunktion), sodass Sie die Twenty API mit einem auf Ihre App beschränkten Anwendungszugriffstoken aufrufen können.
* Pro Anwendung ist nur eine Post-Installationsfunktion zulässig. Der Manifest-Build schlägt fehl, wenn mehr als eine erkannt wird.
* Der `universalIdentifier` der Funktion wird während des Builds im Anwendungsmanifest automatisch als `postInstallLogicFunctionUniversalIdentifier` gesetzt — Sie müssen ihn nicht in `defineApplication()` referenzieren.
* Die `universalIdentifier`, `shouldRunOnVersionUpgrade` und `shouldRunSynchronously` der Funktion werden während des Builds automatisch dem Anwendungsmanifest unter dem Feld `postInstallLogicFunction` hinzugefügt — Sie müssen sie in `defineApplication()` nicht referenzieren.
* Das standardmäßige Timeout ist auf 300 Sekunden (5 Minuten) festgelegt, um längere Einrichtungsvorgänge wie Daten-Seeding zu ermöglichen.
* **Nicht im Dev-Modus ausgeführt**: Wenn eine App lokal registriert ist (über `yarn twenty dev`), überspringt der Server den Installationsablauf vollständig und synchronisiert Dateien direkt über den CLI-Watcher — daher läuft Post-Install im Dev-Modus nie, unabhängig von `shouldRunSynchronously`. Verwenden Sie `yarn twenty exec --postInstall`, um es manuell gegen einen laufenden Workspace auszulösen.
</Accordion>
<Accordion title="definePreInstallLogicFunction" description="Eine Pre-Installations-Logikfunktion definieren (eine pro App)">
Eine Pre-Install-Funktion ist eine Logikfunktion, die automatisch während der Installation ausgeführt wird, **bevor die Metadatenmigration des Workspaces angewendet wird**. Sie hat die gleiche Payload-Struktur wie Post-Install (`InstallPayload`), ist aber früher im Installationsablauf positioniert, sodass sie Zustände vorbereiten kann, von denen die bevorstehende Migration abhängt — typische Anwendungsfälle sind das Sichern von Daten, die Validierung der Kompatibilität mit dem neuen Schema oder das Archivieren von Datensätzen, die umstrukturiert oder entfernt werden sollen.
```ts src/logic-functions/pre-install.ts
import { definePreInstallLogicFunction, type InstallPayload } from 'twenty-sdk';
description: 'Runs before installation to prepare the application.',
timeoutSeconds: 300,
shouldRunOnVersionUpgrade: true,
handler,
});
```
Sie können die Pre-Installationsfunktion auch jederzeit manuell über die CLI ausführen:
```bash filename="Terminal"
yarn twenty exec --preInstall
```
Hauptpunkte:
* Pre-Install-Funktionen verwenden `definePreInstallLogicFunction()` — dieselbe spezialisierte Konfiguration wie bei Post-Install, nur an einen anderen Lifecycle-Slot gebunden.
* Sowohl Pre- als auch Post-Install-Handler erhalten denselben `InstallPayload`-Typ: `{ previousVersion?: string; newVersion: string }`. Importieren Sie ihn einmal und verwenden Sie ihn für beide Hooks wieder.
* **Wann der Hook ausgeführt wird**: positioniert direkt vor der Metadatenmigration des Workspaces (`synchronizeFromManifest`). Vor der Ausführung führt der Server einen rein additiven "pared-down sync" durch, der die Pre-Install-Funktion der **neuen** Version in den Workspace-Metadaten registriert — sonst wird nichts angefasst — und führt sie dann aus. Da dieser Sync nur additiv ist, sind die Objekte, Felder und Daten der vorherigen Version noch intakt, wenn Ihr Handler läuft: Sie können den Zustand vor der Migration gefahrlos lesen und sichern.
* **Ausführungsmodell**: Pre-Install wird **synchron** ausgeführt und **blockiert die Installation**. Wenn der Handler einen Fehler wirft, wird die Installation abgebrochen, bevor Schemaänderungen angewendet werden — der Workspace verbleibt in der vorherigen Version in einem konsistenten Zustand. Das ist beabsichtigt: Pre-Install ist Ihre letzte Chance, ein riskantes Upgrade abzulehnen.
* Wie bei Post-Install ist pro Anwendung nur eine Pre-Installationsfunktion zulässig. Sie wird während des Builds automatisch dem Anwendungsmanifest unter `preInstallLogicFunction` hinzugefügt.
* **Nicht im Dev-Modus ausgeführt**: wie bei Post-Install — der Installationsablauf wird für lokal registrierte Apps vollständig übersprungen, daher läuft Pre-Install unter `yarn twenty dev` nie. Verwenden Sie `yarn twenty exec --preInstall`, um es manuell auszulösen.
</Accordion>
<Accordion title="Pre-Install vs. Post-Install: wann was verwenden" description="Den richtigen Installations-Hook wählen">
Beide Hooks sind Teil desselben Installationsablaufs und erhalten dasselbe `InstallPayload`. Der Unterschied besteht darin, **wann** sie relativ zur Metadatenmigration des Workspaces ausgeführt werden, und das ändert, auf welche Daten sie gefahrlos zugreifen können.
Pre-Install ist immer **synchron** (blockiert die Installation und kann sie abbrechen). Post-Install ist **standardmäßig asynchron** — in einen Worker eingereiht mit automatischen Wiederholungen — kann aber per `shouldRunSynchronously: true` in die synchrone Ausführung wechseln. Siehe das Akkordeon zu `definePostInstallLogicFunction` oben, wann welcher Modus zu verwenden ist.
**Verwenden Sie `post-install` für alles, wofür das neue Schema existieren muss.** Dies ist der Regelfall:
* Standarddaten befüllen (Anlegen anfänglicher Datensätze, Standardansichten, Demo-Inhalte) für neu hinzugefügte Objekte und Felder.
* Registrieren von Webhooks bei Drittanbieter-Diensten, jetzt, da die App ihre Anmeldedaten hat.
* Aufrufen Ihrer eigenen API, um eine Einrichtung abzuschließen, die von den synchronisierten Metadaten abhängt.
* Idempotente "Stelle sicher, dass dies existiert"-Logik, die bei jedem Upgrade den Zustand abgleichen soll — kombinieren Sie dies mit `shouldRunOnVersionUpgrade: true`.
Beispiel — nach der Installation einen Standard-`PostCard`-Datensatz anlegen:
```ts src/logic-functions/post-install.ts
import { definePostInstallLogicFunction, type InstallPayload } from 'twenty-sdk';
import { createClient } from './generated/client';
description: 'Seeds a welcome post card after install.',
timeoutSeconds: 300,
shouldRunOnVersionUpgrade: false,
handler,
});
```
**Verwenden Sie `pre-install`, wenn eine Migration ansonsten vorhandene Daten löschen oder beschädigen würde.** Da Pre-Install gegen das vorherige Schema läuft und ein Fehlschlag das Upgrade zurückrollt, ist es der richtige Ort für alles Riskante:
* **Sichern von Daten, die gleich gelöscht oder umstrukturiert werden** — z. B. Sie entfernen in v2 ein Feld und müssen dessen Werte vor der Migration in ein anderes Feld kopieren oder in einen Speicher exportieren.
* **Archivieren von Datensätzen, die eine neue Einschränkung ungültig machen würde** — z. B. ein Feld wird `NOT NULL` und Sie müssen zuerst Zeilen mit Null-Werten löschen oder korrigieren.
* **Kompatibilität validieren und das Upgrade ablehnen, wenn die aktuellen Daten nicht sauber migriert werden können** — werfen Sie im Handler einen Fehler, und die Installation wird ohne Änderungen abgebrochen. Das ist sicherer, als die Inkompatibilität mitten in der Migration zu entdecken.
* **Daten umbenennen oder Schlüssel neu zuweisen** vor einer Schemaänderung, bei der sonst die Zuordnung verloren ginge.
Beispiel — Datensätze vor einer destruktiven Migration archivieren:
```ts src/logic-functions/pre-install.ts
import { definePreInstallLogicFunction, type InstallPayload } from 'twenty-sdk';
import { createClient } from './generated/client';
| Lang laufendes Seeding oder Drittanbieteraufrufe ausführen, die die Installationsantwort nicht blockieren sollten | `post-install` (Standard — `shouldRunSynchronously: false`, mit Worker-Wiederholungen) |
| Schnelle Einrichtung ausführen, auf die sich der Aufrufer unmittelbar nach der Rückkehr des Installationsaufrufs verlassen wird | `post-install` mit `shouldRunSynchronously: true` |
| Daten lesen oder sichern, die bei der bevorstehenden Migration verloren gingen | `pre-install` |
| Ein Upgrade ablehnen, das vorhandene Daten beschädigen würde | `pre-install` (`throw` im Handler) |
| Bei jedem Upgrade einen Abgleich ausführen | `post-install` mit `shouldRunOnVersionUpgrade: true` |
| Einmalige Einrichtung nur bei der ersten Installation durchführen | `post-install` mit `shouldRunOnVersionUpgrade: false` (Standard) |
<Note>
Im Zweifel auf **Post-Install** setzen. Greifen Sie nur zu Pre-Install, wenn die Migration selbst destruktiv ist und Sie den vorherigen Zustand abfangen müssen, bevor er verloren geht.
</Note>
</Accordion>
<Accordion title="defineFrontComponent" description="Frontend-Komponenten für benutzerdefinierte UI definieren">
| `yarn twenty server start` | Lokalen Server starten (lädt das Image bei Bedarf herunter) |
| `yarn twenty server start --port 3030` | Auf einem benutzerdefinierten Port starten |
| `yarn twenty server start --test` | Starten Sie eine separate Testinstanz auf Port 2021 |
| `yarn twenty server stop` | Server stoppen (Daten bleiben erhalten) |
| `yarn twenty server status` | Serverstatus, URL und Anmeldedaten anzeigen |
| `yarn twenty server logs` | Serverprotokolle streamen |
@@ -228,6 +229,20 @@ Der Scaffolder hat bereits einen lokalen Twenty-Server für Sie gestartet. Um ih
Daten bleiben über Neustarts hinweg in zwei Docker-Volumes bestehen (`twenty-app-dev-data` für PostgreSQL, `twenty-app-dev-storage` für Dateien). Verwenden Sie `reset`, um alles zu löschen und neu zu beginnen.
### Eine Testinstanz ausführen
Übergeben Sie `--test` an jeden `server`-Befehl, um eine zweite, vollständig isolierte Instanz zu verwalten — nützlich, um Integrationstests auszuführen oder zu experimentieren, ohne Ihre Hauptentwicklungsdaten anzutasten.
| `yarn twenty server start --test` | Die Testinstanz starten (standardmäßig Port 2021) |
| `yarn twenty server stop --test` | Die Testinstanz stoppen |
| `yarn twenty server status --test` | Status, URL und Anmeldedaten der Testinstanz anzeigen |
| `yarn twenty server logs --test` | Protokolle der Testinstanz streamen |
| `yarn twenty server reset --test` | Alle Testdaten löschen und neu starten |
Die Testinstanz läuft in einem eigenen Docker-Container (`twenty-app-dev-test`) mit dedizierten Volumes (`twenty-app-dev-test-data`, `twenty-app-dev-test-storage`) und eigener Konfiguration, sodass sie parallel zu Ihrer Hauptinstanz ohne Konflikte ausgeführt werden kann. Kombinieren Sie `--test` mit `--port`, um den Standardport 2021 zu überschreiben.
<Note>
Der Server erfordert, dass **Docker** läuft. Wenn der Fehler "Docker not running" angezeigt wird, stellen Sie sicher, dass Docker Desktop (oder der Docker-Daemon) gestartet ist.
@@ -66,14 +66,71 @@ Der Freigabelink verwendet die Basis-URL des Servers (ohne Workspace-Subdomain),
### Versionsverwaltung
Beim Aktualisieren einer bereits bereitgestellten Tarball-App verlangt der Server, dass die `version` in `package.json` **strikt höher** (gemäß der [semver](https://semver.org)-Reihenfolge) ist als die derzeit bereitgestellte Version. Das erneute Bereitstellen derselben Version oder das Pushen einer niedrigeren Version wird abgelehnt, bevor das Tarball gespeichert wird — in der CLI wird ein `VERSION_ALREADY_EXISTS`-Fehler angezeigt.
So veröffentlichen Sie ein Update:
1. Erhöhen Sie das Feld `version` in Ihrer `package.json`
1. Erhöhen Sie das Feld `version` in Ihrer `package.json` (z. B. `1.2.3` → `1.2.4`, `1.3.0` oder `2.0.0`).
2. Führen Sie `yarn twenty deploy` aus (oder `yarn twenty deploy --remote production`)
3. Arbeitsbereiche, die die App installiert haben, sehen in ihren Einstellungen, dass ein Upgrade verfügbar ist.
<Note>
Pre-Release-Tags funktionieren wie erwartet: Das Erhöhen von `1.0.0-rc.1` → `1.0.0-rc.2` ist zulässig, und eine finale Version wie `1.0.0` wird korrekt als höher als `1.0.0-rc.5` erkannt. Die Version in `package.json` muss selbst eine gültige semver-Zeichenfolge sein.
</Note>
{/* TODO: add screenshot of the Upgrade button */}
Apps, die mit `create-twenty-app` erzeugt wurden, enthalten von Haus aus zwei GitHub-Actions-Workflows unter `.github/workflows/`. Sie sind einsatzbereit, sobald Sie das Repository zu GitHub pushen — für CI ist keine zusätzliche Einrichtung erforderlich, und für CD ist nur ein einziges Secret nötig.
### CI — `ci.yml`
Führt Ihre Integrationstests bei jedem Push auf `main` und bei Pull Requests aus.
**Was sie macht:**
1. Checkt den Quellcode Ihrer App aus.
2. Startet eine isolierte Twenty-Testinstanz mithilfe der Composite-Action `twentyhq/twenty/.github/actions/spawn-twenty-app-dev-test@main` (das CI-Äquivalent zu `yarn twenty server start --test`).
3. Aktiviert Corepack, richtet Node.js anhand Ihrer `.nvmrc` ein und installiert Abhängigkeiten mit `yarn install --immutable`.
4. Führt `yarn test` aus und übergibt `TWENTY_API_URL` und `TWENTY_API_KEY` aus der gestarteten Instanz, damit Ihre Tests mit einem echten Server kommunizieren können.
**Konfigurationsoptionen:**
* `TWENTY_VERSION` (env, standardmäßig `latest`) — fixieren Sie die in CI verwendete Twenty-Server-Version, indem Sie dies in `ci.yml` anpassen.
* Die Parallelität wird nach `github.ref` gruppiert und bricht laufende Ausführungen bei neuen Pushes ab.
Es sind keine Secrets erforderlich — die Testinstanz ist flüchtig und existiert nur für die Dauer des Jobs.
### CD — `cd.yml`
Stellt Ihre App bei jedem Push auf `main` auf einem konfigurierten Twenty-Server bereit und optional aus einem Pull Request, wenn das Label `deploy` gesetzt ist.
**Was sie macht:**
1. Checkt den PR-Head (bei PRs mit Label) oder den gepushten Commit aus.
2. Führt `twentyhq/twenty/.github/actions/deploy-twenty-app@main` aus — das CI-Äquivalent zu `yarn twenty deploy`.
3. Führt `twentyhq/twenty/.github/actions/install-twenty-app@main` aus, damit die neu bereitgestellte Version in den Ziel-Workspace installiert wird.
| `TWENTY_DEPLOY_URL` | `env` in `cd.yml` (standardmäßig `http://localhost:3000`) | Der Twenty-Server, auf den bereitgestellt werden soll. Ändern Sie dies vor der ersten Verwendung auf die echte Server-URL. |
| `TWENTY_DEPLOY_API_KEY` | GitHub-Repository **Settings → Secrets and variables → Actions** | API-Schlüssel mit Berechtigung zum Bereitstellen auf dem Zielserver. |
<Note>
Der Standardwert von `TWENTY_DEPLOY_URL` (`http://localhost:3000`) ist ein Platzhalter — von einem GitHub-gehosteten Runner ist er nicht erreichbar. Aktualisieren Sie sie auf die öffentliche URL Ihres Servers (oder verwenden Sie einen selbstgehosteten Runner mit Netzwerkzugriff), bevor Sie CD aktivieren.
</Note>
**Eine Vorschau-Bereitstellung aus einem PR auslösen:**
Fügen Sie einem Pull Request das Label `deploy` hinzu. Die `if:`-Bedingung in `cd.yml` führt den Job für diesen PR mit dem Head-Commit des PR aus, sodass Sie eine Änderung auf dem Zielserver vor dem Mergen validieren können.
### Fixieren der wiederverwendbaren Actions
Beide Workflows verweisen auf wiederverwendbare Actions mit `@main`, sodass Aktualisierungen der Actions im Repository `twentyhq/twenty` automatisch übernommen werden. Wenn Sie deterministische Builds möchten, ersetzen Sie `@main` in jeder `uses:`-Zeile durch eine Commit-SHA oder einen Release-Tag.
## Auf npm veröffentlichen
Die Veröffentlichung auf npm macht Ihre App im Twenty-Marktplatz auffindbar. Jeder Twenty-Arbeitsbereich kann Marktplatz-Apps direkt über die Benutzeroberfläche durchsuchen, installieren und aktualisieren.
@@ -81,7 +138,7 @@ Die Veröffentlichung auf npm macht Ihre App im Twenty-Marktplatz auffindbar. Je
### Anforderungen
* Ein [npm](https://www.npmjs.com)-Konto
* Das Schlüsselwort `twenty-app` in Ihrem `package.json`-Array `keywords` (bereits enthalten, wenn Sie mit `create-twenty-app` ein Gerüst erstellen)
* Das Schlüsselwort `twenty-app` in Ihrem `package.json`-Array `keywords` (manuell hinzufügen — es ist in der `create-twenty-app`-Vorlage standardmäßig nicht enthalten)
```json filename="package.json"
{
@@ -189,3 +246,12 @@ Sie können Apps auch über die Befehlszeile installieren:
```bash filename="Terminal"
yarn twenty install
```
<Note>
Der Server erzwingt bei der Installation semver-Versionierung und spiegelt damit die Regeln beim Bereitstellen wider:
* Die Installation derselben Version, die in Ihrem Arbeitsbereich bereits installiert ist, wird mit einem `APP_ALREADY_INSTALLED`-Fehler abgelehnt.
* Die Installation einer niedrigeren Version als die aktuell installierte wird mit einem `CANNOT_DOWNGRADE_APPLICATION`-Fehler abgelehnt.
Um eine neuere Version zu installieren, stellen Sie sie zuerst bereit oder veröffentlichen Sie sie und führen Sie dann `yarn twenty install` erneut aus.
**Scrivi una buona `description`.** Gli agenti IA fanno affidamento sul campo `description` della funzione per decidere quando usare lo strumento. Sii specifico su cosa fa lo strumento e quando dovrebbe essere invocato.
</Note>
</Accordion>
<Accordion title="definePreInstallLogicFunction" description="Definisci una funzione logica di pre-installazione (una per app)">
Una funzione di pre-installazione è una funzione logica che viene eseguita automaticamente prima che la tua app venga installata in uno spazio di lavoro. È utile per attività di convalida, controlli dei prerequisiti o per preparare lo stato dello spazio di lavoro prima che proceda l'installazione principale.
```ts src/logic-functions/pre-install.ts
import { definePreInstallLogicFunction, type InstallLogicFunctionPayload } from 'twenty-sdk';
description: 'Runs before installation to prepare the application.',
timeoutSeconds: 300,
handler,
});
```
Puoi anche eseguire manualmente la funzione di pre-installazione in qualsiasi momento utilizzando la CLI:
```bash filename="Terminal"
yarn twenty exec --preInstall
```
Punti chiave:
* Le funzioni di pre-installazione utilizzano `definePreInstallLogicFunction()` — una variante specializzata che omette le impostazioni dei trigger (`cronTriggerSettings`, `databaseEventTriggerSettings`, `httpRouteTriggerSettings`, `isTool`).
* L'handler riceve un `InstallLogicFunctionPayload` con `{ previousVersion: string }` — la versione dell'app precedentemente installata (oppure una stringa vuota per nuove installazioni).
* È consentita una sola funzione di pre-installazione per applicazione. La build del manifesto genererà un errore se ne viene rilevata più di una.
* L'`universalIdentifier` della funzione viene impostato automaticamente come `preInstallLogicFunctionUniversalIdentifier` nel manifesto dell'applicazione durante la build — non è necessario farvi riferimento in `defineApplication()`.
* Il timeout predefinito è impostato a 300 secondi (5 minuti) per consentire attività di preparazione più lunghe.
</Accordion>
<Accordion title="definePostInstallLogicFunction" description="Definisci una funzione logica di post-installazione (una per app)">
Una funzione post-installazione è una funzione logica che viene eseguita automaticamente dopo che la tua app è stata installata in uno spazio di lavoro. Questo è utile per attività di configurazione una tantum come il popolamento di dati predefiniti, la creazione di record iniziali o la configurazione delle impostazioni dello spazio di lavoro.
Una funzione post-installazione è una funzione logica che viene eseguita automaticamente dopo che la tua app è stata installata in uno spazio di lavoro. Il server la esegue **dopo** che i metadati dell'app sono stati sincronizzati e il client SDK è stato generato, così lo spazio di lavoro è completamente pronto per l'uso e il nuovo schema è attivo. I casi d'uso tipici includono il popolamento di dati predefiniti, la creazione di record iniziali, la configurazione delle impostazioni dello spazio di lavoro o il provisioning di risorse su servizi di terze parti.
```ts src/logic-functions/post-install.ts
import { definePostInstallLogicFunction, type InstallLogicFunctionPayload } from 'twenty-sdk';
import { definePostInstallLogicFunction, type InstallPayload } from 'twenty-sdk';
* Le funzioni di post-installazione utilizzano `definePostInstallLogicFunction()` — una variante specializzata che omette le impostazioni dei trigger (`cronTriggerSettings`, `databaseEventTriggerSettings`, `httpRouteTriggerSettings`, `isTool`).
* L'handler riceve un `InstallLogicFunctionPayload` con `{ previousVersion: string }` — la versione dell'app precedentemente installata (oppure una stringa vuota per nuove installazioni).
* L'handler riceve un `InstallPayload` con `{ previousVersion?: string; newVersion: string }` — `newVersion` è la versione in fase di installazione e `previousVersion` è la versione installata in precedenza (oppure `undefined` in caso di nuova installazione). Usa questi valori per distinguere le nuove installazioni dagli aggiornamenti e per eseguire logiche di migrazione specifiche per versione.
* **Quando viene eseguito l'hook**: solo sulle nuove installazioni, per impostazione predefinita. Passa `shouldRunOnVersionUpgrade: true` se vuoi che venga eseguito anche quando l'app viene aggiornata da una versione precedente. Se omesso, il flag è `false` per impostazione predefinita e gli aggiornamenti saltano l'hook.
* **Modello di esecuzione — asincrono per impostazione predefinita, sincrono su richiesta**: il flag `shouldRunSynchronously` controlla *come* viene eseguito il post-install.
* `shouldRunSynchronously: false` *(default)* — l'hook viene **messo in coda nella coda dei messaggi** con `retryLimit: 3` ed eseguito in modo asincrono in un worker. La risposta di installazione viene restituita non appena il job è messo in coda, quindi un handler lento o in errore non blocca il chiamante. Il worker riproverà fino a tre volte. **Usalo per job di lunga durata** — popolamento di dataset di grandi dimensioni, chiamate a API di terze parti lente, provisioning di risorse esterne, qualsiasi cosa che possa superare una finestra di risposta HTTP ragionevole.
* `shouldRunSynchronously: true` — l'hook viene eseguito **inline durante il flusso di installazione** (stesso executor del pre-install). La richiesta di installazione rimane bloccata finché l'handler non termina e, se genera un'eccezione, il chiamante dell'installazione riceve un `POST_INSTALL_ERROR`. Nessun tentativo automatico. **Usalo per attività rapide che devono completarsi prima della risposta** — ad esempio, emettere un errore di validazione all'utente, oppure un setup rapido di cui il client avrà bisogno immediatamente dopo il ritorno della chiamata di installazione. Tieni presente che la migrazione dei metadati è già stata applicata quando viene eseguito il post-install, quindi un errore in modalità sincrona **non** annulla le modifiche allo schema — si limita a far emergere l'errore.
* Assicurati che il tuo handler sia idempotente. In modalità asincrona la coda può riprovare fino a tre volte; in entrambe le modalità l'hook può essere eseguito di nuovo durante gli aggiornamenti quando `shouldRunOnVersionUpgrade: true`.
* Le variabili d'ambiente `APPLICATION_ID`, `APP_ACCESS_TOKEN` e `API_URL` sono disponibili all'interno dell'handler (come in qualsiasi altra funzione logica), quindi puoi chiamare le API di Twenty con un token di accesso applicativo con ambito sulla tua app.
* È consentita una sola funzione di post-installazione per applicazione. La build del manifesto genererà un errore se ne viene rilevata più di una.
* L'`universalIdentifier` della funzione viene impostato automaticamente come `postInstallLogicFunctionUniversalIdentifier` nel manifesto dell'applicazione durante la build — non è necessario farvi riferimento in `defineApplication()`.
* I campi `universalIdentifier`, `shouldRunOnVersionUpgrade` e `shouldRunSynchronously` della funzione vengono associati automaticamente al manifest dell'applicazione nel campo `postInstallLogicFunction` durante la build — non è necessario referenziarli in `defineApplication()`.
* Il timeout predefinito è impostato a 300 secondi (5 minuti) per consentire attività di configurazione più lunghe, come il popolamento dei dati.
* **Non eseguito in modalità dev**: quando un'app è registrata in locale (tramite `yarn twenty dev`), il server salta completamente il flusso di installazione e sincronizza i file direttamente tramite il watcher della CLI — quindi il post-install non viene mai eseguito in modalità dev, indipendentemente da `shouldRunSynchronously`. Usa `yarn twenty exec --postInstall` per attivarlo manualmente su un workspace in esecuzione.
</Accordion>
<Accordion title="definePreInstallLogicFunction" description="Definisci una funzione logica di pre-installazione (una per app)">
Una funzione di pre-install è una funzione logica che viene eseguita automaticamente durante l'installazione, **prima che venga applicata la migrazione dei metadati del workspace**. Condivide la stessa struttura di payload del post-install (`InstallPayload`), ma è posizionata prima nel flusso di installazione così da poter preparare lo stato da cui dipenderà la migrazione imminente — usi tipici includono il backup dei dati, la validazione della compatibilità con il nuovo schema o l'archiviazione di record che stanno per essere ristrutturati o eliminati.
```ts src/logic-functions/pre-install.ts
import { definePreInstallLogicFunction, type InstallPayload } from 'twenty-sdk';
description: 'Runs before installation to prepare the application.',
timeoutSeconds: 300,
shouldRunOnVersionUpgrade: true,
handler,
});
```
Puoi anche eseguire manualmente la funzione di pre-installazione in qualsiasi momento utilizzando la CLI:
```bash filename="Terminal"
yarn twenty exec --preInstall
```
Punti chiave:
* Le funzioni di pre-install usano `definePreInstallLogicFunction()` — stessa configurazione specialistica del post-install, solo agganciata a uno slot di ciclo di vita diverso.
* Sia gli handler di pre- sia quelli di post-install ricevono lo stesso tipo `InstallPayload`: `{ previousVersion?: string; newVersion: string }`. Importalo una volta e riutilizzalo per entrambi gli hook.
* **Quando viene eseguito l'hook**: posizionato appena prima della migrazione dei metadati del workspace (`synchronizeFromManifest`). Prima dell'esecuzione, il server esegue una "sincronizzazione ridotta" puramente additiva che registra nei metadati del workspace la funzione di pre-install della versione **nuova** — nient'altro viene toccato — e poi la esegue. Poiché questa sincronizzazione è solo additiva, gli oggetti, i campi e i dati della versione precedente restano intatti quando il tuo handler viene eseguito: puoi leggere ed eseguire in sicurezza il backup dello stato pre-migrazione.
* **Modello di esecuzione**: il pre-install è eseguito **in modo sincrono** e **blocca l'installazione**. Se l'handler genera un'eccezione, l'installazione viene interrotta prima che vengano applicate modifiche allo schema — il workspace rimane sulla versione precedente in uno stato coerente. Questo è intenzionale: il pre-install è la tua ultima possibilità per rifiutare un aggiornamento rischioso.
* Come per il post-install, è consentita una sola funzione di pre-installazione per applicazione. Viene collegata automaticamente al manifest dell'applicazione nel campo `preInstallLogicFunction` durante la build.
* **Non eseguito in modalità dev**: come per il post-install — il flusso di installazione viene completamente saltato per le app registrate localmente, quindi il pre-install non viene mai eseguito con `yarn twenty dev`. Usa `yarn twenty exec --preInstall` per attivarlo manualmente.
</Accordion>
<Accordion title="Pre-install vs post-install: quando usare l'uno o l'altro" description="Scegliere l'hook di installazione giusto">
Entrambi gli hook fanno parte dello stesso flusso di installazione e ricevono lo stesso `InstallPayload`. La differenza è **quando** vengono eseguiti rispetto alla migrazione dei metadati del workspace, e questo modifica quali dati possono gestire in sicurezza.
Il pre-install è sempre **sincrono** (blocca l'installazione e può interromperla). Il post-install è **asincrono per impostazione predefinita** — messo in coda su un worker con retry automatici — ma può optare per l'esecuzione sincrona con `shouldRunSynchronously: true`. Vedi l'accordion `definePostInstallLogicFunction` sopra per quando usare ciascuna modalità.
**Usa `post-install` per tutto ciò che richiede l'esistenza del nuovo schema.** Questo è il caso più comune:
* Popolamento di dati predefiniti (creazione di record iniziali, viste predefinite, contenuti demo) su oggetti e campi appena aggiunti.
* Registrazione di webhook con servizi di terze parti ora che l'app ha le proprie credenziali.
* Chiamare la tua API per completare il setup che dipende dai metadati sincronizzati.
* Logica idempotente di "ensure this exists" che dovrebbe riconciliare lo stato a ogni aggiornamento — da combinare con `shouldRunOnVersionUpgrade: true`.
Esempio — eseguire il seeding di un record `PostCard` predefinito dopo l'installazione:
```ts src/logic-functions/post-install.ts
import { definePostInstallLogicFunction, type InstallPayload } from 'twenty-sdk';
import { createClient } from './generated/client';
description: 'Seeds a welcome post card after install.',
timeoutSeconds: 300,
shouldRunOnVersionUpgrade: false,
handler,
});
```
**Usa `pre-install` quando una migrazione altrimenti distruggerebbe o corromperebbe i dati esistenti.** Poiché il pre-install viene eseguito contro lo schema *precedente* e un suo fallimento annulla l'aggiornamento, è il posto giusto per qualsiasi operazione rischiosa:
* **Eseguire il backup dei dati che stanno per essere eliminati o ristrutturati** — ad esempio, stai rimuovendo un campo nella v2 e devi copiarne i valori in un altro campo o esportarli su uno storage prima che venga eseguita la migrazione.
* **Archiviare i record che un nuovo vincolo renderebbe non validi** — ad esempio, un campo sta diventando `NOT NULL` e devi prima eliminare o correggere le righe con valori nulli.
* **Validare la compatibilità e rifiutare l'aggiornamento se i dati attuali non possono essere migrati correttamente** — genera un'eccezione dall'handler e l'installazione si interrompe senza applicare modifiche. Questo è più sicuro che scoprire l'incompatibilità a migrazione in corso.
* **Rinominare o rigenerare le chiavi dei dati** prima di una modifica dello schema che farebbe perdere l'associazione.
Esempio — archiviare i record prima di una migrazione distruttiva:
```ts src/logic-functions/pre-install.ts
import { definePreInstallLogicFunction, type InstallPayload } from 'twenty-sdk';
import { createClient } from './generated/client';
| Popolare dati predefiniti, configurare il workspace, registrare risorse esterne | `post-install` |
| Eseguire seeding di lunga durata o chiamate a terze parti che non dovrebbero bloccare la risposta dell'installazione | `post-install` (predefinito — `shouldRunSynchronously: false`, con retry del worker) |
| Eseguire un setup rapido di cui il chiamante farà affidamento immediatamente dopo il ritorno della chiamata di installazione | `post-install` con `shouldRunSynchronously: true` |
| Leggere o eseguire il backup dei dati che la prossima migrazione perderebbe | `pre-install` |
| Rifiutare un aggiornamento che corromperebbe i dati esistenti | `pre-install` (genera un'eccezione dall'handler) |
| Eseguire la riconciliazione a ogni aggiornamento | `post-install` con `shouldRunOnVersionUpgrade: true` |
| Eseguire un setup una tantum solo alla prima installazione | `post-install` con `shouldRunOnVersionUpgrade: false` (predefinito) |
<Note>
In caso di dubbio, usa **post-install**. Ricorri al pre-install solo quando la migrazione stessa è distruttiva e devi intercettare lo stato precedente prima che vada perso.
</Note>
</Accordion>
<Accordion title="defineFrontComponent" description="Definisci componenti front-end per un'interfaccia utente personalizzata">
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.