## Summary
- New workflow `ci-visual-regression.yaml` that runs on PRs touching
`twenty-ui` or `twenty-shared`
- Builds `twenty-ui` storybook and uploads the tarball as a GitHub
Actions artifact
- Dispatches to `twentyhq/ci-privileged` which handles the pixel-diff
comparison and posts a PR comment
### Flow
```
twenty CI (this PR) ci-privileged pixel-perfect
───────────────── ────────────── ──────────────
Build storybook
Upload artifact
Dispatch ──────────────► Download artifact
Upload to S3 (OIDC)
POST /import-from-storage ────► Import build
POST /diffs/run ──────────────► Screenshots + diff
◄────────────────────────────── Diff report JSON
Post PR comment
```
Companion PRs:
- https://github.com/twentyhq/ci-privileged/pull/1 (ci-privileged
workflow)
- https://github.com/twentyhq/twenty-infra/pull/497 (OIDC trust + Helm
cleanup)
- https://github.com/twentyhq/pixel-perfect/pull/5 (API simplification)
## Test plan
- [ ] Merge companion PRs first and configure secrets/environments
- [ ] Open a test PR touching twenty-ui, verify storybook builds and
dispatch fires
- [ ] Verify visual regression comment appears on the PR
## Summary
- Fixes a bug where removing an option from a MULTI_SELECT field fails
with `invalid input value for enum` when existing records contain the
removed value alongside surviving values.
- The root cause was the `ELSE` branch in `updateArrayEnum` which tried
to cast removed enum values (e.g. `DISTRIBUTOR`) to the new enum type
that no longer includes them.
- The fix replaces the `ELSE` cast with a NULL-producing implicit CASE
default and uses `array_agg(...) FILTER (WHERE mapped_value IS NOT
NULL)` to silently strip removed values from existing arrays.
### Before (bug)
```sql
-- ELSE branch tries to cast removed value to new enum → crash
CASE unnest_value::text
WHEN 'IMPL' THEN 'IMPL'::new_enum
WHEN 'APP' THEN 'APP'::new_enum
ELSE unnest_value::text::new_enum -- 'DISTRIBUTOR' fails here
END
```
### After (fix)
```sql
-- No ELSE: removed values produce NULL, filtered out by array_agg
SELECT array_agg(mapped_value) FILTER (WHERE mapped_value IS NOT NULL)
FROM (
SELECT CASE unnest_value::text
WHEN 'IMPL' THEN 'IMPL'::new_enum
WHEN 'APP' THEN 'APP'::new_enum
END AS mapped_value
FROM unnest(old_column) AS unnest_value
) enum_mapping
```
Three feature flags changed the UI:
IS_NAVIGATION_MENU_ITEM_ENABLED — Sidebar no longer has a default
"People" link → navigate via URL instead
IS_COMMAND_MENU_ITEM_ENABLED — Button label is now "Create new Person"
(interpolated from object name) instead of static "Create new record"
IS_JUNCTION_RELATIONS_ENABLED — Company is now a junction relation
("Previous Companies") displayed inline, no longer a boxed
dynamic-relation-widget on the record page
Updated upgrade command to also backfill command menu items for
workflows.
This has been done in the same command because we need to enable the
feature flag once both operations are complete: the creation of the
standard command menu items and the creation of workflow command menu
items.
## Summary
- File serving endpoints (`GET file/:fileFolder/:id` and `GET
public-assets/...`) were piping S3/local file streams directly to the
response without any HTTP headers, allowing a stored XSS attack via
uploaded HTML files rendered inline on the CRM origin.
- Adds `Content-Type`, `Content-Disposition`, and
`X-Content-Type-Options: nosniff` headers to all file serving responses.
Only known-safe MIME types (images, PDF, plain text, audio, video) are
served inline; everything else (HTML, SVG, XML, etc.) forces
`Content-Disposition: attachment` to trigger download instead of
rendering.
- New `setFileResponseHeaders` utility with an explicit allowlist of
inline-safe MIME types.
## Test plan
- [x] Unit tests pass (9 tests including 2 new ones: header assertions
and attachment-disposition for HTML)
- [x] Lint clean (`lint:diff-with-main`)
- [x] Typecheck clean (`nx typecheck twenty-server`)
- [ ] Manual: upload an HTML file via `uploadWorkflowFile`, access the
returned URL — should download instead of rendering
- [ ] Manual: upload a PNG image, access the URL — should render inline
with correct `Content-Type: image/png`
- [ ] Manual: verify `X-Content-Type-Options: nosniff` header is present
on all file responses
Made with [Cursor](https://cursor.com)
---------
Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com>
Co-authored-by: Etienne <etiennejouan@users.noreply.github.com>
Co-authored-by: Etienne <45695613+etiennejouan@users.noreply.github.com>
## Summary
- The front deploy to S3 is failing because `@ai-sdk/groq` was removed
from `packages/twenty-server/package.json` but the `yarn.lock` was never
updated
## Summary
- Adds a `repository-dispatch` step to the AI catalog sync workflow
(`ci-ai-catalog-sync.yaml`) that notifies `twenty-infra` when a sync PR
is created
- This allows the automerge workflow in `twenty-infra` to reactively
merge the PR instead of waiting for the next cron run
## Test plan
- [ ] Verify `TWENTY_INFRA_TOKEN` secret is available (same one used by
i18n workflows)
- [ ] Trigger the AI catalog sync workflow manually and confirm the
dispatch fires
Made with [Cursor](https://cursor.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
- The `ci-ai-catalog-sync` cron workflow was failing because the
`ai:sync-models-dev` NestJS command bootstraps the full app, which tries
to connect to PostgreSQL — unavailable in CI.
- Converted the sync logic to a standalone `ts-node` script
(`scripts/ai-sync-models-dev.ts`) that runs without NestJS, eliminating
the database dependency.
- Removed the `Build twenty-server` step from the workflow since it's no
longer needed, making the job faster.
## Test plan
- [x] Verified the standalone script runs successfully locally via `npx
nx run twenty-server:ts-node-no-deps-transpile-only --
./scripts/ai-sync-models-dev.ts`
- [x] Verified `--dry-run` flag works correctly
- [x] Verified the output `ai-providers.json` is correctly written with
valid JSON (135 models across 5 providers)
- [x] Verified the script passes linting with zero errors
- [ ] CI should pass without requiring a database service
Fixes:
https://github.com/twentyhq/twenty/actions/runs/23424202182/job/68135439740
Made with [Cursor](https://cursor.com)
## Summary
- **Add `lingui:compile` to Dockerfile** before both the server and
frontend build stages, ensuring compiled translation catalogs are always
fresh regardless of git state
- **Add `repository-dispatch` to i18n workflows** (`i18n-push.yaml` and
`i18n-pull.yaml`) to trigger reactive automerge in `twenty-infra` when
the i18n PR is ready, replacing the 15-minute polling approach
## Context
Users sometimes see "Uncompiled message detected" errors because
releases can be cut from `main` before the i18n PR (with freshly
compiled translation catalogs) has been merged. This creates a race
condition between new translatable strings landing on `main` and their
compiled catalogs being available.
These changes fix this in two ways:
1. **Safety net in builds**: Every Docker build now compiles
translations before building, so even if compiled catalogs in git are
stale, the build artifact is always correct
2. **Faster i18n PR merges**: Instead of a 15-minute cron polling for
i18n PRs, the workflows now notify `twenty-infra` immediately when
translations are ready, reducing merge latency from ~15 minutes to ~1
minute
Companion PR in twenty-infra: twentyhq/twenty-infra
(feat/i18n-reactive-automerge)
## Test plan
- [ ] Verify `TWENTY_INFRA_TOKEN` secret is available to i18n workflows
- [ ] Docker build still succeeds with the added `lingui:compile` steps
- [ ] i18n-push triggers automerge in twenty-infra after pushing changes
- [ ] i18n-pull triggers automerge in twenty-infra after pulling
translations
Made with [Cursor](https://cursor.com)
## Summary
- Removes the `isBillingEnabled` guard that was hiding the Providers and
Custom Providers sections in the Admin AI settings
- The `GET_AI_PROVIDERS` query was being skipped when billing was
enabled, and the provider UI sections were conditionally hidden —
there's no reason to gate provider configuration behind billing status
- Cleans up the now-unused `billingState` and `useAtomStateValue`
imports
## Test plan
- [ ] Verify Providers and Custom Providers sections are visible in
Admin > AI settings on cloud (billing enabled)
- [ ] Verify they remain visible on self-hosted (billing disabled)
Made with [Cursor](https://cursor.com)
## Summary
This PR adds a comprehensive billing usage analytics feature that
provides detailed breakdowns of credit consumption across execution
types, users, resources, and time periods. The implementation includes a
new ClickHouse-backed analytics service, GraphQL API endpoint, and a
frontend dashboard component.
## Key Changes
### Backend
- **New BillingAnalyticsService**: Queries ClickHouse for usage
breakdowns by user, resource, execution type, and time series data
- **BillingEventWriterService**: Writes billing events to ClickHouse for
analytics while maintaining best-effort semantics (never blocks Stripe
billing)
- **ClickHouse Schema**: Added `billingEvent` table with 3-year TTL for
storing detailed billing event data
- **GraphQL Resolver**: New `getBillingAnalytics` query that aggregates
usage data for the current billing period, protected by feature flag and
billing permissions
- **Enhanced BillingUsageEvent**: Added `userWorkspaceId` field to track
per-user credit consumption
- **AI Billing Integration**: Updated AI billing service to pass
`userWorkspaceId` when recording usage events
### Frontend
- **SettingsBillingAnalyticsSection**: New component displaying:
- Usage breakdown by execution type with progress bars
- Daily usage time series chart (28-day view)
- Per-user credit consumption breakdown
- Per-resource (agent/workflow) credit consumption breakdown
- **SettingsUsage Page**: Dedicated page for viewing usage analytics
- **GraphQL Query**: `GetBillingAnalytics` query with generated hooks
- **Navigation**: Added Usage menu item in settings (feature-flagged)
- **Mock Data**: Included screenshot mock data for preview/testing
### Feature Flag
- Added `IS_USAGE_ANALYTICS_ENABLED` feature flag to control visibility
and access to analytics features
## Implementation Details
- Analytics data is queried in parallel for performance
- ClickHouse writes are non-blocking to ensure billing operations never
fail
- Progress bars use dynamic coloring from a predefined palette
- Time series visualization normalizes bar heights relative to max value
- Empty state handling when no analytics data is available
- Responsive UI with proper text truncation for long names
https://claude.ai/code/session_01Y1EqrX6PFq3EJxJq89h7DF
---------
Co-authored-by: Claude <noreply@anthropic.com>
## Summary
- The `MigrateModelIdsToCompositeFormat` migration was setting
`agent.modelId = NULL`, but the column has a `NOT NULL` constraint —
causing the migration to fail on deploy.
- Fixed by setting `modelId` to the `'default-smart-model'` sentinel
value instead of NULL. The runtime already resolves this sentinel
dynamically via `getEffectiveModelConfig` / `isDefaultModelSentinel`, so
agents correctly fall back to the admin-configured smart model.
## Test plan
- [ ] Run `npx nx run twenty-server:database:migrate:prod` — migration
should complete without errors
- [ ] Verify agents still resolve to the correct model at runtime
(sentinel → `getDefaultPerformanceModel()`)
Made with [Cursor](https://cursor.com)
- Renamed FieldConfiguration's layout field to fieldDisplayMode as it
caused issues with the layout field of BarChartConfiguration
- Create relation Field widgets for standard objects
---------
Co-authored-by: Charles Bochet <charles@twenty.com>
## Automated fix for [bug 6848](https://sonarly.com/issue/6848?type=bug)
**Severity:** `critical`
### Summary
WorkflowEditActionCodeFields.tsx calls Object.entries(functionInput)
without a null guard on line 40, crashing when
action.settings.input.logicFunctionInput is undefined. This happens
because workflow version step data in the database may contain the old
field name serverlessFunctionInput (from before the rename in commit
da6f1bbef3), and no data migration was created to update JSON keys
inside the steps JSONB column.
### User Impact
Users viewing or editing a workflow version containing a CODE step that
was created before the serverlessFunction-to-logicFunction rename see a
crash. The page fails to render the workflow step detail panel,
preventing them from viewing or modifying their workflow.
### Root Cause
Proximate cause: Object.entries(functionInput) on line 40 of
WorkflowEditActionCodeFields.tsx throws TypeError because functionInput
is undefined. The stack trace confirms this: the crash occurs during
React rendering (through scheduler -> react-dom render pipeline ->
WorkflowEditActionCodeFields:40:15 -> Object.entries).
1. Why did Object.entries throw? Because the functionInput prop is
undefined. It is initialized from
action.settings.input.logicFunctionInput via useState on line 142 of
WorkflowEditActionCode.tsx, with no fallback value.
2. Why is action.settings.input.logicFunctionInput undefined? Because
the workflow version step data stored in the database JSONB steps column
contains the OLD field name serverlessFunctionInput instead of
logicFunctionInput. When the frontend accesses logicFunctionInput on the
deserialized JSON object, it returns undefined.
3. Why does the database still have the old field name? Because commit
da6f1bbef3 (Rename serverlessFunction to logicFunction, Jan 28 2026)
renamed the Zod schema field from serverlessFunctionInput to
logicFunctionInput, and the database migration
1769556947746-renameServerless.ts only renames SQL tables and columns
(serverlessFunction table to logicFunction, etc.) but does NOT update
the JSON keys inside the steps JSONB column of the workflowVersion
table.
4. Why was no JSON data migration created? The rename was a large-scale
refactoring across the entire codebase. The steps column stores workflow
action data as opaque JSON, and the migration only addressed relational
schema changes (table/column renames) without updating the embedded JSON
document structure. There is no upgrade command in 1-17, 1-18, or 1-19
directories that transforms the JSON keys from serverlessFunctionInput
to logicFunctionInput.
5. Why did the component not handle this gracefully? The
WorkflowEditActionCodeFields component has no null guard on the
functionInput prop before calling Object.entries. Notably, the analogous
WorkflowEditActionLogicFunction component DOES have a null guard on line
52 (action.settings.input.logicFunctionInput ?? {}), showing the team
was aware of this possibility in one component but missed it in the
other.
**Introduced by:** martmull on 2026-02-16 in commit
[`da064d5`](https://github.com/twentyhq/twenty/commit/da064d5e88a62939e0545a37c68381822e6932ef)
### Suggested Fix
Two minimal null guards are added, matching the pattern already used in
the analogous WorkflowEditActionLogicFunction component. First, in
WorkflowEditActionCode.tsx the useState initializer is changed from
action.settings.input.logicFunctionInput to
action.settings.input.logicFunctionInput ?? {} so that functionInput is
always an empty object rather than undefined when the DB record still
uses the old serverlessFunctionInput key. Second, in
WorkflowEditActionCodeFields.tsx the Object.entries call is changed from
Object.entries(functionInput) to Object.entries(functionInput ?? {}) as
a defensive guard at the render layer, ensuring the component renders
safely even if an undefined value is passed from any caller.
### Evidence
- **Code:**
[packages/twenty-front/src/modules/workflow/workflow-steps/workflow-actions/code-action/components/WorkflowEditActionCode.tsx:142
- const [functionInput, setFunctionInput]
=](https://github.com/twentyhq/twenty/blob/main/packages/twenty-front/src/modules/workflow/workflow-steps/workflow-actions/code-action/components/WorkflowEditActionCode.tsx#L142)
- **Code:**
[packages/twenty-front/src/modules/workflow/workflow-steps/workflow-actions/code-action/components/WorkflowEditActionCodeFields.tsx:40
- {Object.entries(functionInput ?? {}).map(([inputKey, inputValue]) =>
{](https://github.com/twentyhq/twenty/blob/main/packages/twenty-front/src/modules/workflow/workflow-steps/workflow-actions/code-action/components/WorkflowEditActionCodeFields.tsx#L40)
---
*Generated by [Sonarly](https://sonarly.com)*
Co-authored-by: Sonarly Claude Code <claude-code@sonarly.com>
- Automatically create and sync command menu items for all workflows
with a manual trigger
- Refactor `useCommandMenuItemsFromBackend`
- Prefill a _Quick Lead_ workflow command menu item during workspace
setup and dev seeding
- Add a ready prop to `HeadlessEngineCommandWrapperEffect` to prevent
premature execution when async data hasn't loaded yet
---------
Co-authored-by: Charles Bochet <charles@twenty.com>
## Summary
Follow-up to #18157. Cherry-picks the useful parts from #18481 (by
@dnplkndll), adapted to align with the current state of `main`:
- **Helm unit tests** for Redis external authentication (secret-based +
plaintext password, both server and worker)
- **Helm unit tests** for `extraEnv` injection (plain values and
`valueFrom` on both server and worker)
- **JSON schema validation** for `server.extraEnv` and `worker.extraEnv`
in `values.schema.json`
- **Fix** `server_url_test.yaml` to use JSONPath filters
(`@.name=="SERVER_URL"`) instead of brittle `env[0]` index selectors
- **Fix** worker `storageEnv` whitespace (missing `-` in `{{-
$storageEnv | nindent 12 }}`)
Stale tests from #18481 (for `disableDbMigrations`, `server.env`
pass-through, and `run-migrations` init container) were dropped since
those features were removed during the #18157 cleanup.
## Test plan
- [ ] `helm lint` passes
- [ ] `helm template` renders cleanly
- [ ] Unit tests pass with `helm unittest` (requires the plugin)
Made with [Cursor](https://cursor.com)
This pull request enhances the Helm chart for the Twenty application by
improving how environment variables and Redis credentials are handled
for both server and worker deployments. The main changes include support
for injecting additional environment variables, improved Redis password
management (including external secrets), and a more robust database
migration workflow.
**Environment Variable Injection:**
- Added support for specifying additional environment variables for both
the server and worker deployments via the `additionalEnv` field in
`values.yaml`. These variables are automatically injected into the
respective pods.
[[1]](diffhunk://#diff-b5d958eae48fd1919e5623bcf0144aac7abb323ae8743e6f31367e383c63c296R55)
[[2]](diffhunk://#diff-b5d958eae48fd1919e5623bcf0144aac7abb323ae8743e6f31367e383c63c296R109-R110)
[[3]](diffhunk://#diff-20bb91909627a12b50b3c165a2a027b663479c0104ed8dbf91d2b9ad8ea8a931R74-R77)
[[4]](diffhunk://#diff-20bb91909627a12b50b3c165a2a027b663479c0104ed8dbf91d2b9ad8ea8a931R157-R172)
[[5]](diffhunk://#diff-20bb91909627a12b50b3c165a2a027b663479c0104ed8dbf91d2b9ad8ea8a931R225-R229)
[[6]](diffhunk://#diff-fb612a3b7a13156aaa607b27d23025e2c6831f111b6a582fd313fad26d2fdb5bR89-R92)
**Redis Credential Management:**
- Introduced support for using external secrets for Redis passwords by
adding `secretName` and `passwordKey` fields under `redis.external` in
`values.yaml`, and logic to inject `REDIS_PASSWORD` from a Kubernetes
secret if configured.
[[1]](diffhunk://#diff-b5d958eae48fd1919e5623bcf0144aac7abb323ae8743e6f31367e383c63c296R180-R182)
[[2]](diffhunk://#diff-5c4fa358b10abd7581188995feb9b4d6be0bc4f06a95bf27bb31b5595d6693d8R92-R100)
[[3]](diffhunk://#diff-20bb91909627a12b50b3c165a2a027b663479c0104ed8dbf91d2b9ad8ea8a931R157-R172)
[[4]](diffhunk://#diff-20bb91909627a12b50b3c165a2a027b663479c0104ed8dbf91d2b9ad8ea8a931R196-R205)
[[5]](diffhunk://#diff-fb612a3b7a13156aaa607b27d23025e2c6831f111b6a582fd313fad26d2fdb5bR70-R79)
- Updated the logic for constructing the `REDIS_URL` to include
authentication information if a password is set or an external secret is
used.
**Database Migration Workflow:**
- Improved the startup command for the server deployment to optionally
skip database migrations (using `DISABLE_DB_MIGRATIONS`), check for an
existing schema before running migrations, and ensure setup scripts are
only run on empty databases.
These changes make the chart more flexible and secure, especially for
production deployments requiring externalized secrets and custom
environment configurations.
---------
Co-authored-by: Charles Bochet <charles@twenty.com>
## Summary
Fixes#13008
Azure AD has a [known
bug](https://learn.microsoft.com/en-us/answers/questions/5570899/our-adminconsent-url-always-fails-with-aadsts65005)
where the first consent attempt for a multi-tenant app fails with
`AADSTS650051` because the service principal is mid-provisioning in the
target tenant. Microsoft's own engineers have acknowledged the issue
with no fix ETA.
This PR adds a transparent retry to the `MicrosoftOAuthGuard`:
1. When the OAuth redirect returns `AADSTS650051`, the guard parses the
`state` parameter to recover the original query params (workspaceId,
invite hash, locale, etc.)
2. Redirects the user back to `/auth/microsoft` with a `msRetry=1`
counter
3. The full OAuth flow restarts — by this point the service principal
has finished provisioning, so consent succeeds
4. Capped at 1 retry (`MAX_MICROSOFT_AUTH_RETRIES`) to prevent infinite
loops. If it still fails, falls through to the normal error page
From the user's perspective, they just see a brief extra redirect
instead of a cryptic error page.
### Context
- `AADSTS650051` is a race condition in Azure AD's service principal
provisioning during multi-tenant app consent
- It's distinct from missing consent (`AADSTS65001`) or admin consent
required (`AADSTS90094`)
- The error is transient — retrying the same flow immediately succeeds
- Microsoft Q&A threads confirm this affects many multi-tenant apps, not
just Twenty
### References
- [Microsoft Q&A: adminconsent always fails with AADSTS650051 on first
attempt](https://learn.microsoft.com/en-us/answers/questions/5570899/our-adminconsent-url-always-fails-with-aadsts65005)
- [Microsoft Q&A: AADSTS650051 for multiple
applications](https://learn.microsoft.com/en-us/answers/questions/5571098/when-customers-attempt-to-sign-in-and-grant-consen)
## Test plan
- [ ] Deploy to a staging environment with Microsoft OAuth enabled
- [ ] Have a user from a new Azure AD tenant attempt Microsoft sign-in
for the first time
- [ ] If AADSTS650051 occurs, verify the user is transparently retried
and signs in successfully
- [ ] Verify `msRetry` counter prevents infinite redirect loops (check
server logs for the warn message)
- [ ] Verify normal Microsoft sign-in flow (no error) is unaffected
🤖 Generated with [Claude Code](https://claude.com/claude-code)
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Co-authored-by: neo773 <62795688+neo773@users.noreply.github.com>
Co-authored-by: neo773 <neo773@protonmail.com>
## Summary
- Renames `ObjectMetadataItem` to `EnrichedObjectMetadataItem` across
the entire frontend (~440 files) to clarify that this type includes
derived fields (`readableFields`, `updatableFields`, nested `fields[]`,
`indexMetadatas[]`) computed at read time from the metadata store
- Creates `splitObjectMetadataGqlResponse` that goes directly from a
GraphQL `ObjectMetadataItemsQuery` response to flat store items
(combining the old
`mapPaginatedObjectMetadataItemsToObjectMetadataItems` +
`splitObjectMetadataItemWithRelated` two-step flow into one call)
- Removes `ObjectMetadataItemWithRelated` type and all "WithRelated"
naming
- Renames `generatedMockObjectMetadataItems` to
`generateTestEnrichedObjectMetadataItemsMock` to make it clear this is
test-only enriched data
- Deletes `useLoadMockedObjectMetadataItems` hook (consolidated into
`useLoadMockedMinimalMetadata`)
- Ensures nothing destined for the metadata store computes
`readableFields`/`updatableFields` (preventing the localStorage bloat
from #18809)
## Type hierarchy (before → after)
**Before:**
```
ObjectMetadataItemsQuery → mapPaginated → ObjectMetadataItemWithRelated → enrich → ObjectMetadataItem
→ split → FlatObjectMetadataItem (store)
```
**After:**
```
ObjectMetadataItemsQuery → splitObjectMetadataGqlResponse → FlatObjectMetadataItem (store)
→ mapPaginated + enrich (tests only) → EnrichedObjectMetadataItem
```
## Test plan
- [x] `npx nx typecheck twenty-front` passes
- [x] `npx nx test twenty-front` passes (767 suites, 4505 tests)
- [x] `npx nx lint twenty-front` passes
- [ ] CI checks pass
Made with [Cursor](https://cursor.com)
## Summary
- On unauthenticated pages (login), mocked object metadata was being
hydrated into the store with `readableFields` and `updatableFields`
attached. These derived arrays duplicate every field per object,
inflating `objectMetadataItems` in localStorage from ~70 KB to ~2 MB.
Combined with other metadata keys, this exceeded Safari's 5 MB quota and
caused `QuotaExceededError`, preventing real data from replacing mock
data on login.
- Extracted flat mock data into a dedicated file
(`generatedMockObjectMetadataItemsWithRelated.ts`) and use it for store
hydration, keeping the enriched version
(`generatedMockObjectMetadataItems`) only for tests.
- Completed `splitObjectMetadataItemWithRelated`'s contract by stripping
`readableFields` and `updatableFields` at runtime (not just via
TypeScript's `Omit`), matching the `FlatObjectMetadataItem` return type
that omits all four relational properties.
## Root cause
1. `MinimalMetadataLoadEffect` loads mocked metadata on unauthenticated
pages.
2. `generatedMockObjectMetadataItems` was produced by
`enrichObjectMetadataItemsWithPermissions`, which attaches
`readableFields` and `updatableFields` (full copies of the `fields`
array).
3. `splitObjectMetadataItemWithRelated` only destructured `fields` and
`indexMetadatas` — `readableFields`/`updatableFields` leaked into
`...objectProperties` at runtime because `Omit` is a compile-time-only
guard.
4. These bloated objects were written to localStorage, consuming ~2 MB
instead of ~70 KB.
5. On login, the intermediate state (old composite mock `current` + new
flat real `draft`) exceeded the 5 MB quota, causing a
`QuotaExceededError` deadlock.
## Test plan
- [ ] Open the app in a fresh Safari private window (empty localStorage)
- [ ] Verify the login page loads without errors
- [ ] Log in and verify metadata loads correctly
- [ ] Check `localStorage` size —
`metadataStoreState__objectMetadataItems` should be ~70 KB, not ~2 MB
- [ ] Run existing tests: `npx nx test twenty-front` — no regressions
from the mock data split
Made with [Cursor](https://cursor.com)
---------
Co-authored-by: Charles Bochet <charles@twenty.com>
## Summary
- Store SSO connections (Google, Microsoft, OIDC, SAML) as connected
accounts in the core schema during sign-in/sign-up, gated behind the
`IS_CONNECTED_ACCOUNT_MIGRATED` feature flag
- Add `OIDC` and `SAML` to `ConnectedAccountProvider` enum with
exhaustive switch handling across frontend and backend
- Add `IS_CONNECTED_ACCOUNT_MIGRATED` to `DEFAULT_FEATURE_FLAGS` for new
workspaces, with a fallback check so SSO accounts are created even
before workspace activation
- Always upsert connected accounts to both workspace and core schemas
during messaging OAuth flow, fixing FK constraint violations when
SSO-only accounts exist only in core
- Create message/calendar channels when they don't exist regardless of
new vs reconnect flow
- Filter settings accounts list to only show accounts that have message
or calendar channels
## Test plan
- [ ] Sign up with Google SSO → verify connected account is created in
core schema
- [ ] Connect messaging (Google APIs) after SSO sign-up → verify no FK
errors, channels created, configuration page renders correctly
- [ ] Reconnect an existing messaging account → verify tokens updated,
sync resets triggered
- [ ] Sign in with OIDC/SAML SSO → verify connected account created with
oidcTokenClaims
- [ ] Verify settings accounts page only shows accounts with channels
(SSO-only accounts hidden)
- [ ] Verify typecheck, lint, and unit tests pass
## Summary
Fixes#18744 — The workflow FIND_RECORDS action silently drops filter
conditions when a variable resolves to null/empty, causing the query to
return **all records** instead of erroring.
**Root cause (three compounding layers):**
1. **`variable-resolver.ts`** — `resolveString` returns `undefined` when
a variable lookup fails (e.g., `{{steps.trigger.output.userId}}` where
`userId` doesn't exist in context). The return type says `string` but
`evalFromContext` actually returns `undefined` at runtime.
2. **`checkIfShouldSkipFiltering.ts`** — Treats `undefined`/`null`/`""`
values as "skip this filter." This is correct for the **UI filter
builder** (user hasn't finished typing), but wrong for **workflow
execution** (variable resolution failed = misconfigured workflow).
3. **`find-records.workflow-action.ts`** — When all filters are silently
skipped, `computeRecordGqlOperationFilter` returns `{}` (match
everything). The query runs with no filter, returning all records —
silently succeeding with wrong results.
## Fix
Added validation in `find-records.workflow-action.ts` **after**
`resolveInput` but **before** `computeRecordGqlOperationFilter`. For
each filter with a value-requiring operand (i.e., not IS_EMPTY,
IS_NOT_EMPTY, IS_IN_PAST, IS_IN_FUTURE, IS_TODAY), if the resolved value
is `undefined`, `null`, or `""`, throw `INVALID_STEP_INPUT` with a
descriptive error message.
**Why this approach:**
- Scoped to the workflow executor — does **not** break the UI filter
builder's intentional skip-on-empty behavior
- Does not change shared utilities (`checkIfShouldSkipFiltering`,
`resolveInput`) used across the app
- Fails fast with a clear error instead of silently returning wrong data
- 1 file changed, 23 lines added
## Test plan
- [x] Backend typecheck passes
- [x] oxlint passes (0 warnings, 0 errors)
- [x] Prettier passes
- [ ] Manual: Create a workflow with FIND_RECORDS using a variable that
doesn't exist → should error with "Filter condition has an empty value
after variable resolution" instead of returning all records
- [ ] Manual: Create a workflow with FIND_RECORDS using IS_EMPTY operand
(no value needed) → should still work correctly
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-authored-by: Charles Bochet <charles@twenty.com>
## Summary
- The record board (Kanban view) only loaded the first 10 records per
column, then showed skeleton placeholder cards for the rest without ever
fetching the remaining data
- **Root cause**: The `RecordBoardFetchMoreInViewTriggerComponent`
(IntersectionObserver) is positioned at the bottom of the entire board.
With 10 real cards + 10 skeleton cards per column (~3500px of content),
the trigger div was pushed far beyond the 1600px `rootMargin` detection
zone, so `recordBoardShouldFetchMore` never became `true` and fetch-more
was never triggered
- **Fix**: The initial query now sets `recordBoardShouldFetchMore =
true` when columns need more data, and `triggerRecordBoardFetchMore`
uses an internal `while` loop to load all remaining pages in a single
invocation — making it immune to the InView component racing to reset
the flag between React render cycles
## Summary
- Replaces per-provider TypeScript constant files
(`openai-models.const.ts`, `anthropic-models.const.ts`, etc.) with a
single `ai-providers.json` catalog as the source of truth
- Adds runtime model discovery via AI SDK for self-hosted providers,
with `models.dev` enrichment for pricing/capabilities
- Introduces composite model IDs (`provider/modelId`) for canonical,
conflict-free identification
- Simplifies provider configuration: API keys are injected from
environment variables (e.g., `OPENAI_API_KEY`)
- Adds admin panel UI for provider management (add/remove/test), model
discovery, recommended model configuration, and default fast/smart model
selection per workspace
- Removes deprecated config variables (`AI_DISABLED_MODEL_IDS`,
`AUTO_ENABLE_NEW_AI_MODELS`, etc.)
- Adds database migration for composite model ID format
## Test plan
- [ ] Server typecheck passes
- [ ] Frontend typecheck passes
- [ ] Server unit tests pass
- [ ] Frontend unit tests pass
- [ ] CI pipeline green
- [ ] Admin panel AI tab loads correctly
- [ ] Provider discovery works for configured providers
- [ ] Model recommendation toggles persist
- [ ] Default fast/smart model selection works
Made with [Cursor](https://cursor.com)
## Summary
Fixes#17941 — Saving a blank subdomain causes a redirect to
`.website.com`, effectively breaking the workspace.
**Root cause:** Three layers all fail to reject an empty string `""`:
1. **Frontend (`SettingsDomain.tsx`):** `SaveButton` has both
`onClick={onSave}` and `type="submit"`. The `onClick` fires first,
calling `handleSave()` directly without running Zod validation. So
`isDefined("")` returns `true`, the confirmation modal opens, and the
blank subdomain is submitted.
2. **Backend DTO (`update-workspace-input.ts`):** The `subdomain` field
has `@IsString()` + `@IsOptional()` but no pattern validation, so an
empty string passes the DTO layer.
3. **Backend service (`workspace.service.ts:152`):** `if
(payload.subdomain && ...)` — empty string is falsy in JS, so it skips
`validateSubdomainOrThrow()` entirely and writes `subdomain: ""` to the
database.
**The crash:** After save, the redirect logic does
`"myworkspace.website.com".replace("myworkspace", "")` →
`".website.com"`, sending the user to an invalid URL.
## Fix
- **Frontend:** Call `form.trigger()` at the start of `handleSave` to
run Zod validation regardless of whether the function was invoked via
`onClick` or `form.handleSubmit`. Returns early with validation error if
invalid.
- **Backend DTO:** Add `@Matches(/^[a-z0-9][a-z0-9-]{1,28}[a-z0-9]$/)`
to reject invalid subdomains at the request validation layer
(defense-in-depth).
- **Backend service:** Change `if (payload.subdomain && ...)` to `if
(isDefined(payload.subdomain) && ...)` so empty strings route through
`validateSubdomainOrThrow()` instead of being silently skipped.
## Test plan
- [x] Existing `is-subdomain-valid.util.spec.ts` tests pass (36/36)
- [x] TypeScript type checks pass for both `twenty-server` and
`twenty-front`
- [x] oxlint passes on all changed files
- [x] Prettier passes on all changed files
- [ ] Manual: Navigate to Settings > Domains, clear the subdomain field,
click Save — should show validation error, not redirect
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-authored-by: Charles Bochet <charles@twenty.com>
Co-authored-by: Charles Bochet <charlesBochet@users.noreply.github.com>
This PR:
- Breaks useAgentChatData into focused effect components (streaming,
fetch,
init, auto-scroll, diff sync)
- Splits message list into non-last (stable) + last (streaming/error) to
prevent full re-renders on each stream chunk
- Adds scroll-to-bottom button and MutationObserver-based auto-scroll on
thread switch
- Lifts loading state from context to atoms
- Adds areEqual to selector
factories
We could improve further but this sets up a robust architecture for
further refactoring.
## Messages flow
The flow of messages loading and streaming is now more solid.
Everything goes out from `AgentChatAiSdkStreamEffect`, whether loaded
from the DB or streaming directly, and every consumers is using only one
atom `agentChatMessagesComponentFamilyState`
## Data sync effect with callbacks new hook
See
`packages/twenty-front/src/modules/apollo/hooks/useQueryWithCallbacks.ts`
which allows to fix Apollo v4 migration leftovers and is an
implementation of the pattern we talked about with @charlesBochet
We could refine this pattern in another PR.
# Before
https://github.com/user-attachments/assets/84e7a96f-6790-405d-8a73-2dacbf783be5
# After
https://github.com/user-attachments/assets/4c692e3a-2413-4513-abcc-44d0da311203
Co-authored-by: Charles Bochet <charles@twenty.com>
## Summary
- Restores the original migration timestamp (`1773945207801`) that was
accidentally changed to `1774005903909` during PR #18787
- Keeps the content improvements (default column values) from that PR
intact
## Test plan
- [ ] Verify migration runs correctly with `npx nx run
twenty-server:database:migrate:prod`
Made with [Cursor](https://cursor.com)
## Summary
This preserves percent-encoded payloads when normalizing links fields.
`lowercaseUrlOriginAndRemoveTrailingSlash` was decoding the path and
query string while lowercasing the URL origin. That changes URLs where
encoded payloads are semantically significant, such as Google Maps links
containing `%2F` segments.
Closes#18698.
## Changes
- stop decoding the path/query payload in
`lowercaseUrlOriginAndRemoveTrailingSlash`
- preserve the raw path, query, and hash while still lowercasing the
origin and trimming a trailing slash
- update shared URL normalization tests to assert encoded payloads stay
encoded
- add a server-side regression test covering imported links field
normalization
## Validation
- `corepack yarn jest --config packages/twenty-shared/jest.config.mjs
packages/twenty-shared/src/utils/url/__tests__/lowercaseUrlOriginAndRemoveTrailingSlash.test.ts
--runInBand`
- `corepack yarn jest --config packages/twenty-server/jest.config.mjs
packages/twenty-server/src/engine/core-modules/record-transformer/utils/__tests__/transform-links-value.util.spec.ts
--runInBand`
- `corepack yarn nx test twenty-server --runInBand
--testFile=src/engine/core-modules/record-transformer/utils/__tests__/transform-links-value.util.spec.ts`
---------
Co-authored-by: Charles Bochet <charles@twenty.com>
## Summary
- Reverts the `useSortedNavigationMenuItems` coupling in
`ViewPickerOptionDropdown` introduced by #18791
- The viewPicker now uses `useNavigationMenuItemsData` directly for the
`isFavorite` check, keeping view ordering and navigation menu item
ordering independent
## Why
PR #18791 replaced `useNavigationMenuItemsData` with
`useSortedNavigationMenuItems` in the viewPicker for favorite detection.
While the intent was to filter out stale/orphaned navigation items, this
created an unnecessary coupling: `useSortedNavigationMenuItems`
subscribes to `viewsSelector` and `objectMetadataItemsSelector`, making
the viewPicker transitively dependent on the navigation menu item
ordering system. View ordering in the picker (driven by `view.position`)
and navigation menu item ordering (driven by
`navigationMenuItem.position`) should remain decorrelated.
## Test plan
- [ ] Open the viewPicker dropdown and verify views are listed in
correct order
- [ ] Drag-and-drop to reorder views in the viewPicker — confirm it
works
- [ ] Verify the "Add to Favorite" / "Manage favorite" label still
correctly reflects favorite state
- [ ] Reorder navigation menu items in the sidebar — confirm viewPicker
order is unaffected
Made with [Cursor](https://cursor.com)
## Summary
Fixes#18757
This fixes a set of Favorites / navigation-menu-item integrity problems
related to deleted views, stale hidden items, and upgraded workspaces
with orphaned navigation items.
## What changed
- delete `navigationMenuItem` entries when their favorited view is
deleted
- keep the client metadata store in sync immediately when a view is
deleted
- determine whether a view is already favorited from visible valid
navigation items instead of raw stale items
- add a `1.20.0` upgrade repair command that deletes orphan navigation
menu items and normalizes positions
- add regression coverage for deletion of both record-based and
view-based navigation menu items
## Details
Server:
- extend `NavigationMenuItemDeletionService` so cleanup applies to
deleted views as well as deleted records
- add regression tests covering record-based deletion, view-based
deletion, and no-op behavior
- add `DeleteOrphanNavigationMenuItemsCommand` to remove orphaned items
pointing to:
- deleted views
- deleted records
- missing folders
- normalize positions per scope (`userWorkspaceId + folderId`) after
repair
- wire the new repair command into the `1.20.0` upgrade flow
Frontend:
- add `useRemoveNavigationMenuItemByViewId`
- remove the related navigation item from client metadata immediately
when deleting a view
- use sorted / visible navigation items for favorite detection so stale
hidden rows do not block re-adding a favorite
## Why
Issue `#18757` reports mismatches between Favorites shown in the UI and
rows users can still find in the database. We found that current
Favorites behavior is driven by `navigationMenuItem`, not the legacy
`favorite` table, and that stale / orphaned `navigationMenuItem` rows
could:
- remain after deleting a favorited view
- stay hidden from the UI if they point to invalid targets
- still cause the UI to think a view was already favorited
- persist in workspaces with migration damage from skipped sequential
upgrades
This patch addresses those cases directly and adds an upgrade-time
repair path for older corrupted workspaces.
## Validation
Passed:
- `./node_modules/.bin/jest --config
packages/twenty-server/jest.config.mjs --runInBand
packages/twenty-server/src/engine/metadata-modules/navigation-menu-item/services/__tests__/navigation-menu-item-deletion.service.spec.ts`
- `./node_modules/.bin/tsc -p packages/twenty-front/tsconfig.json
--noEmit --pretty false`
Known unrelated existing failure:
- `./node_modules/.bin/tsc -p packages/twenty-server/tsconfig.json
--noEmit --pretty false`
The server typecheck failure is pre-existing and unrelated to this
branch. Current errors are around `@file-type/pdf` module resolution and
`is-psl-parsed-domain.type.ts`.
---------
Co-authored-by: Charles Bochet <charles@twenty.com>
Co-authored-by: Charles Bochet <charlesBochet@users.noreply.github.com>
On top of [previous closed
PR](https://github.com/twentyhq/twenty/pull/18713) from @FelixMalfait :
- add a schema-creation-skipping optimization
- extract a handler-per-operation pattern,
- add runtime input validation guards,
- integrate with the standard workspace cache
- add gql-style error handling
To do/optimize/check :
- gql parsing and null backfilling
## Intro
This PR introduces a **direct GraphQL execution path** that bypasses
per-workspace GraphQL schema generation for workspace data queries (CRUD
on user-defined objects like companies, people, tasks, etc.).
## Why
In the current architecture, every workspace gets its own
dynamically-generated GraphQL schema reflecting its custom objects and
fields. This costs **~20MB of RAM per workspace per pod** and takes time
to build. For a multi-tenant SaaS with thousands of workspaces, this is
a significant infrastructure cost and a latency bottleneck (especially
on cold starts or cache misses).
The insight is that most workspace queries (`findMany`, `createOne`,
`updateOne`, etc.) don't actually *need* the full schema — they can be
routed directly to the existing Common API query runners by parsing the
GraphQL AST and matching resolver names against object metadata. The
schema is only truly needed for introspection, subscriptions, or queries
that mix core and workspace resolvers.
## How It Works
1. A Yoga `onRequest` plugin intercepts incoming GraphQL requests
2. It parses the query AST and checks if all top-level fields map to
generated workspace resolvers (e.g. `findManyCompanies`,
`createOnePerson`)
3. If yes, it executes them directly against the query runners, skipping
schema generation entirely
4. If the query contains introspection, subscriptions, or core-only
resolvers, it falls through to the normal path
5. Even for mixed queries it can't fully handle, it sets
`skipWorkspaceSchemaCreation` to avoid building the schema when
unnecessary
The whole thing is gated behind the
`IS_DIRECT_GRAPHQL_EXECUTION_ENABLED` feature flag for safe incremental
rollout.
**Net effect**: dramatically lower memory footprint and faster response
times for the vast majority of workspace API calls.
---------
Co-authored-by: Félix Malfait <felix@twenty.com>
Fixes https://github.com/twentyhq/private-issues/issues/432
## Problem
When a user's invoice goes unpaid, Stripe moves their subscription to
`unpaid` status, and Twenty suspends the workspace. But if the user pays
that invoice while a new billing period has started, Stripe has already
generated a new **draft** invoice for that period. Since the draft isn't
finalized or paid, Stripe doesn't reactivate the subscription — the
workspace stays suspended indefinitely.
## What was missing
- No handler for the `invoice.paid` Stripe webhook event
- No mechanism to finalize draft invoices that accumulated during the
`unpaid` period
- No way to reset the workspace deletion countdown (`suspendedAt`) when
a user shows payment intent
## What was added
### 1. `StripeInvoiceService` — new Stripe SDK wrapper
- `listDraftInvoices(stripeSubscriptionId)` — lists all draft invoices
for a subscription
- `finalizeInvoice(invoiceId)` — finalizes a draft with `auto_advance:
true` so Stripe auto-charges it
### 2. `INVOICE_PAID` enum value
Added to `BillingWebhookEvent` so the controller can route it.
### 3. `processInvoicePaid()` in `BillingWebhookInvoiceService`
New private handler that:
- Fetches all draft invoices for the subscription
- Filters to only those whose `period_end` is in the past (already
overdue)
- Finalizes each one (with error handling per invoice to avoid blocking
the webhook)
- If the workspace is suspended, resets `suspendedAt` to now (buys time
before deletion)
### 4. Controller routing
`INVOICE_FINALIZED` and `INVOICE_PAID` are now grouped in the same
switch case, both calling `processStripeEvent(data, eventType)`, which
forks internally in the service.
## Expected recovery flow
User pays overdue invoice
→ Stripe fires invoice.paid
→ Handler finalizes past-due draft invoices (auto_advance: true)
→ Stripe auto-charges them
→ Subscription becomes active
→ customer.subscription.updated fires
→ Existing logic unsuspends workspace
→ suspendedAt refreshed (resets deletion countdown while payments
cascade)
## Summary
Fixes#18610
After completing the CreateProfile onboarding step,
`currentWorkspaceMembersState` (plural, used by `useActorFieldDisplay`
to resolve "Created by" names) was never updated with the new name —
only `currentWorkspaceMemberState` (singular) was. This caused "Created
by" fields to display "Untitled" for the remainder of the session.
**Root cause analysis:**
The issue reporter attributed the bug to empty
`core.user.firstName/lastName`, but `createdBy` display doesn't use
`core.user` at all. The actual flow:
1. Sign-up creates workspace member with empty names (copied from empty
`core.user`)
2. `GetCurrentUserDocument` loads `currentWorkspaceMembersState` with
empty names
3. CreateProfile step updates workspace member in DB +
`currentWorkspaceMemberState` (singular) ✓
4. `currentWorkspaceMembersState` (plural) is **never refreshed** — the
query is skipped once `currentUser` exists
5. `useActorFieldDisplay` looks up from the stale plural state → empty
name → "Untitled"
**Fix:** Update `currentWorkspaceMembersState` alongside
`currentWorkspaceMemberState` in the CreateProfile submit handler.
## Summary
Fixes "missing FROM-clause entry for table" SQL error when using
`orderByForRecords` with a relation field (e.g. `{ company: { name:
"AscNullsFirst" } }`) in a `groupBy` query.
### Root cause
This is a regression from #18005 (Feb 17). That PR correctly removed a
duplicate `applyOrderToBuilder()` call on the groupBy subquery (which
was conflicting with the `ROW_NUMBER() OVER (... ORDER BY ...)` window
function), but in doing so it also removed the LEFT JOINs that
`applyOrderToBuilder` was adding for relation fields.
After that change, ordering relied solely on `getOrderByRawSQL()` inside
`applyPartitionByToBuilder()`, which builds raw SQL for the window
function but never added the required JOINs. Scalar field ordering (e.g.
`name`, `position`) kept working since those don't need JOINs, but
relation field ordering (e.g. `company.name`) broke.
### Fix
- `getOrderByRawSQL` now returns `relationJoins` alongside the SQL
string so callers get the join info they need
- `applyPartitionByToBuilder` in `GroupByWithRecordsService` adds the
required LEFT JOINs before building the `ROW_NUMBER()` window function,
mirroring the pattern used by `applyOrderToBuilder` in the `findMany`
path
## Test plan
- [x] Manually tested locally with a GraphQL query matching the
customer's failing pattern (`orderByForRecords: [{ company: { name:
"AscNullsFirst" } }]`)
- [x] Added integration tests for ascending and descending relation
field ordering in `group-by-with-records-resolver.integration-spec.ts`
- [x] Typecheck passes
- [x] Lint passes
- [x] CI green
## Summary
- Some production workspaces have `SELECT` or `MULTI_SELECT`
fieldMetadata options that are missing the `id` property (e.g.
`[{"color":"yellow","label":"Draft","value":"DRAFT","position":0},
...]`).
- Adds a new upgrade command
`upgrade:1-20:backfill-select-field-option-ids` that queries all
SELECT/MULTI_SELECT fieldMetadata per workspace, detects options missing
an `id`, and backfills them with a UUID v4.
- The command is idempotent (no-op when all options already have ids),
supports `--dry-run`, and invalidates workspace caches after patching.
## Summary
### Fix 1: Autogrow input unclickable when value is empty string
- Fixes the last name input on the Person record show page being
unclickable on regular screens
### Fix 2: Surface nested migration errors in add-missing-system-fields
command
- `AddMissingSystemFieldsToStandardObjectsCommand` calls
`workspaceMigrationRunnerService.run()` directly and was re-throwing
`WorkspaceMigrationRunnerException` without reading its nested `errors`
- Extracted `getNestedErrorMessages()` helper (reused by both
`isUniqueViolationError` and `enrichErrorMessage`)
- Both catch sites now call `enrichErrorMessage()` before re-throwing,
appending nested error details to the message
**Before:**
```
ERROR [UpgradeCommand] Error in workspace ...: Migration action 'create' for 'fieldMetadata' failed
ERROR [UpgradeCommand] undefined
```
**After:**
```
ERROR [UpgradeCommand] Error in workspace ...: Migration action 'create' for 'fieldMetadata' failed (metadata: duplicate key value violates unique constraint "...")
```
## Test plan
- [x] Lint passes
- [x] Typecheck passes
- [x] Verify upgrade command errors now include nested error details
- [x] Verify autogrow input is clickable when value is empty string
## Summary
- Removes a redundant direct `cookieStorage.setItem('tokenPair', ...)`
call in `handleSetAuthTokens` that was overwriting the Jotai-managed
cookie (which has a 180-day expiry) with a session cookie (no expiry)
- This caused users to be logged out whenever their browser fully
closed, instead of staying authenticated for 180 days
## Root cause
In April 2025 (`a7e6564017`), a direct `cookieStorage.setItem` call was
added alongside `setTokenPair()` as a workaround because Recoil's
`onSet` effect fired too late for the Apollo client to read the token
synchronously.
In February 2026 (`674f4353cd`), `tokenPairState` was migrated from
Recoil to Jotai's `atomWithStorage`, which writes the cookie
**synchronously** with a 180-day `expires`. The old direct write was
left in place and now runs *after* the Jotai write, overwriting the
cookie without an `expires` — making it a session cookie.
## Test plan
- [ ] Log in to the app
- [ ] Inspect the `tokenPair` cookie in DevTools → Application → Cookies
- [ ] Verify the cookie has an expiration date ~180 days from now (not
"Session")
- [ ] Close and reopen the browser — confirm you remain logged in
Made with [Cursor](https://cursor.com)
## Summary
- Migrates 4 entities (`connectedAccount`, `messageChannel`,
`calendarChannel`, `messageFolder`) from per-workspace schemas to the
shared `core` metadata schema
- Introduces a `IS_CONNECTED_ACCOUNT_MIGRATED` feature flag to control
the migration: when enabled, reads come from core metadata and all
writes are dual-written to both workspace and core
- Extracts 12 enums from workspace entity files to `twenty-shared` for
reuse across frontend and backend
- Creates new TypeORM entities, metadata services, GraphQL
resolvers/DTOs, and exception interceptors per entity
- Each entity owns its own data access module
(`ConnectedAccountDataAccessModule`, `MessageChannelDataAccessModule`,
`CalendarChannelDataAccessModule`, `MessageFolderDataAccessModule`) — no
umbrella infrastructure module
- Adds a 1.20 upgrade command that backfills data from workspace schemas
to core (preserving UUIDs) and enables the feature flag
- Replaces direct repository access with data access service calls
across ~50 files in messaging, calendar, and connected-account modules
- Adds `lastSignedInAt` and `oidcTokenClaims` fields to the new
`ConnectedAccountEntity`
- Drops unused `lastSyncHistoryId` field from the migrated connected
account entity
## Test plan
- [x] Lint passes (`npx nx lint:diff-with-main twenty-server`)
- [x] Typecheck passes (`npx nx typecheck twenty-server`)
- [x] All unit tests pass (477 suites, 4267 tests, 0 failures)
- [ ] Manual test: verify messaging sync works with feature flag
disabled (existing behavior)
- [ ] Manual test: run upgrade command on a workspace, verify data
backfilled to core tables
- [ ] Manual test: verify messaging/calendar sync works with feature
flag enabled (dual-write path)
- [ ] Manual test: verify GraphQL metadata resolvers return correct data
when flag enabled
Previously, the Opened section was always hidden for all workflow
objects. We now base the “in nav” check on the full set of object
metadata IDs that have a workspace nav item, instead of only active
non-system objects. So: if an object has a nav item (e.g. workflow
runs), it no longer appears under Opened; if it doesn’t, it can appear
there. The hook was simplified to only expose
`objectMetadataIdsInWorkspaceNav`, and the unused return value was
removed.
## Summary
When clicking the ✕ button on an uploaded file in the AI chatbot context
preview, the click event bubbled up to the `StyledClickableContainer`
parent, which triggered `handleClick` (opening the file preview modal)
instead of — or in addition to — calling `onRemove`.
**Root cause:** `StyledClickableContainer` has `onClick={handleClick}`
at the div level. The `AvatarOrIcon` (X button) `onClick={onRemove}`
callback didn't stop propagation, so the event continued to bubble and
opened the preview.
**Fix:** Wrap `onRemove` in a `handleRemove` callback that calls
`e.stopPropagation()` before invoking the original handler.
Fixes#18298
---------
Co-authored-by: victorjzq <zhiqiangjia@users.noreply.github.com>
Co-authored-by: Charles Bochet <charles@twenty.com>
Co-authored-by: Charles Bochet <charlesBochet@users.noreply.github.com>
Related to issue #17711
Follow-up to PR #17774 which fixed the frontend link normalization only
## Summary
- Normalize `primaryEmail` (lowercase) and `primaryLinkUrl` in relation
`connect.where` composite values
- Applied in both frontend spreadsheet import and backend
`DataArgProcessorService` to cover all entry points (UI import, GraphQL
API)
---------
Co-authored-by: Etienne <45695613+etiennejouan@users.noreply.github.com>
## Summary
- Fixes merge queue PRs blocking each other by changing the concurrency
group in `ci-merge-queue.yaml`
- The old concurrency group used `merge_group.base_ref` which resolves
to `refs/heads/main` for every PR, causing all merge queue entries to
serialize behind a single concurrency slot
- Now uses `github.ref` (unique per entry:
`refs/heads/gh-readonly-queue/main/pr-NUMBER-SHA`), matching what all
other CI workflows already do
## Recommended ruleset changes (in GitHub Settings > Rules > Rulesets >
"CI Status Checks")
- **Grouping strategy**: Switch `ALLGREEN` to `NONE` -- each PR is still
tested against the correct base (including all PRs ahead of it in the
queue), but failures only affect the failing PR instead of ejecting the
entire group. `max_entries_to_build: 5` still allows parallel
speculative testing.
- **`min_entries_to_merge_wait_minutes`**: Reduce from 5 to 1 -- the
5-minute wait adds unnecessary latency to every merge.
## Test plan
- [ ] Enqueue 2+ PRs in the merge queue and verify both trigger e2e
tests in parallel instead of one blocking the other
## Summary
- **Restores `convertViewFilterValueToString()` calls** that were lost
when the converter layer was removed in #18667. The GraphQL
`ViewFilter.value` is typed as `JSON` (can be a string, array, or
object), but the frontend type system expects a `string`. Without
stringification, SELECT/MULTI_SELECT filter values (e.g. `['LOST']`)
reach `arrayOfStringsOrVariablesSchema` as raw arrays, causing a
`ZodError: expected string, received array`.
- **Fixes applied at two data boundary points**: `splitViewWithRelated`
(primary entry from metadata store) and `mapViewFiltersToFilters` (which
also accepts `GqlViewFilter[]` directly).
Fixes a production regression introduced by #18667.
## Test plan
- [ ] Apply a SELECT filter (e.g. filter Opportunities by Stage =
"Lost") — should no longer throw ZodError
- [ ] Apply a MULTI_SELECT filter — should work correctly
- [ ] Verify filters with multiple selected values work (e.g. Stage is
"Lost" or "Won")
- [ ] Verify empty filters and "is not" operands still work
- [ ] Verify filters loaded from saved views still work after page
refresh
Made with [Cursor](https://cursor.com)
fix CSS selector > button not reaching Button component's inner element
### Reproduce
- when you click Data model and create a new field, you will see this
bug.
- When you import record and check the height of remove button in the
final validation step.
### Root Reason
the Button component internally renders a wrapper div around the
<button> element, so > button only reaches the intermediate div, not the
actual button.
### Fix
- padding-right not applied → chevron overlapped with text
- ValidationStep: height: 24px not applied → Remove button was 32px
instead of 24px
<img width="1288" height="376" alt="image"
src="https://github.com/user-attachments/assets/885cd8b0-1fe2-484a-8425-70f52b784ecb"
/>
<img width="1001" height="291" alt="image"
src="https://github.com/user-attachments/assets/0b489478-8fbc-4f7e-886a-38012eeb07ef"
/>
## Summary
- Migrates `LogicFunctionModule`, `CodeInterpreterModule`, and
`CaptchaModule` from the `forRootAsync` + injection token pattern to the
`DriverFactoryBase` lazy-loading pattern (matching `EmailModule` and
`FileStorageModule`)
- Fixes#18724 where `LOGIC_FUNCTION_TYPE` was not respected in worker
processes because the driver was created at module boot time before the
DB config cache was loaded
- Removes `isEnvOnly` from `LOGIC_FUNCTION_TYPE`,
`CODE_INTERPRETER_TYPE`, `CAPTCHA_DRIVER`, `IS_MULTIWORKSPACE_ENABLED`,
and `FRONTEND_URL` — these can now be safely configured via the database
at runtime
## How it works
Each migrated module now uses a `DriverFactory` (extending
`DriverFactoryBase`) instead of a module-level async factory + Symbol
injection token:
1. **Lazy creation**: `getCurrentDriver()` creates the driver on first
call, after `DatabaseConfigDriver.onModuleInit()` has loaded the DB
cache
2. **Auto-recreation**: If config changes in the DB, the next
`getCurrentDriver()` call detects the key mismatch and creates a new
driver instance
3. **Unified config**: Both server and worker read from the same
database — driver config only needs to be set once
### Files deleted (old pattern)
- `logic-function-module.factory.ts`,
`logic-function-drivers.module.ts`, `logic-function-driver.constants.ts`
- `code-interpreter-module.factory.ts`
- `captcha.module-factory.ts`, `captcha-driver.constants.ts`
### Files created (new pattern)
- `logic-function-driver.factory.ts`
- `code-interpreter-driver.factory.ts`
- `captcha-driver.factory.ts`
Net: **-150 lines**
## Test plan
- [x] `npx nx typecheck twenty-server` passes
- [x] `npx nx lint:diff-with-main twenty-server` passes
- [ ] Integration tests pass (`npx nx run
twenty-server:test:integration:with-db-reset`)
- [ ] Verify logic functions execute in workflow runs (the original bug)
- [ ] Verify code interpreter works in workflow code steps
- [ ] Verify captcha validation works on sign-up (when captcha is
configured)
Made with [Cursor](https://cursor.com)
- Fix duplication of tabs in dashboards that didn't open the duplicated
tab in the side panel: replaced the logic that used Math.round with an
algorithm that switches the positions of the elements. We will keep
relying on integers, but it will work well in all cases.
- Make dnd work with record page layouts, where there is a pinned tab
- Make tab movements work with pinned tab, too
- Allow user to set a tab as the pinned tab
- Make backend changes to be able to save tabs with `layoutMode`
https://github.com/user-attachments/assets/ce1130fa-71df-49ba-ba2f-6f971e15dd49
---------
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
Introduces a progress indicator for command menu items (both engine
commands and front components). When a command is running, the menu item
now displays a percentage alongside the loader spinner instead of just a
spinner.
- Added `commandMenuItemProgressFamilyState` to track per-item progress
and `CommandListItemLoader` to render it
- Exposed `updateProgress` in the twenty-sdk public API so front
components can report execution progress back to the host
- Wired progress reporting into `ExportMultipleRecordsCommand` as the
first consumer, showing CSV export progress
- Progress state is cleaned up on unmount for both engine commands and
headless front components
- Refactor
## Summary
- **Dynamic SSE headers**: The `graphql-sse` client was created with a
static `Authorization` header captured at creation time. When the access
token refreshed, the SSE client kept using the expired token on every
reconnection attempt, causing up to 10 wasted retries before the client
was disposed and recreated. Now `headers` is passed as a function that
reads the latest token from the Jotai store on each connection attempt.
- **Token renewal retry with error classification**:
`handleTokenRenewal` previously had zero retry tolerance — any failure
during `renewToken` (including transient network errors, server 500s, or
timeouts) triggered an immediate full logout via
`onUnauthenticatedError()`. Now the renewal retries up to 3 times with
linear backoff for transient errors. Only explicit server rejections
(`CombinedGraphQLErrors`, e.g. expired/revoked refresh token) skip
retries and proceed to logout immediately.
- **Preserved error types in `renewTokenMutation`**: The old code caught
all errors and re-threw them as a generic `new Error('Something went
wrong...')`, destroying the original error type. Callers couldn't
distinguish a GraphQL auth rejection from a network failure. Now errors
propagate with their original type.
- **Simplified SSE retry handler**: With dynamic headers handling token
freshness automatically, the retry handler no longer needs the
`initialTokenForSseClient` comparison to detect token mismatches. It now
only resets the SSE client when the user has logged out (no token) or
after 10+ consecutive failures.
## Test plan
- [ ] Log in, wait for access token to expire (~30 min or configure
shorter expiry), verify no unexpected logout occurs
- [ ] Simulate transient network failure during token renewal (e.g.
throttle network in devtools), verify the retry logic recovers without
logging out
- [ ] Verify SSE real-time updates continue working after a token
refresh
- [ ] Verify genuine logout still works when refresh token is actually
invalid/expired
- [ ] Open multiple browser tabs, verify token refresh works correctly
across all tabs without "suspicious activity" revocation
Made with [Cursor](https://cursor.com)
## Summary
- Move `BackfillNavigationMenuItemTypeCommand` from the 1-19 to the 1-20
upgrade path and split the DB transaction into two phases (data
backfill, then schema changes) to avoid the PostgreSQL error "cannot
ALTER TABLE because it has pending trigger events."
- Fix backfill logic to prefer `OBJECT` over `VIEW` for navigation menu
items that have `targetObjectMetadataId`, and correct already mis-typed
items. Tighten the `CHECK` constraint to enforce `viewId IS NULL` for
`OBJECT` type items.
- On the frontend, force `navigationMenuItems` into `staleEntityKeys`
when the server's `minimalMetadata` response omits the collection hash
(happens when the Redis cache hasn't been warmed after an upgrade),
ensuring the sidebar loads navigation items.
## Test plan
- [ ] Upgrade from 1.18 or 1.19 to 1.20 and verify the migration
completes without errors
- [ ] Verify navigation menu items of type `OBJECT` do not have a
`viewId` set in the database
- [ ] Sign out and sign in — confirm navigation menu items appear in the
sidebar on first load
- [ ] Verify `VIEW`-typed items also appear correctly in the sidebar
Made with [Cursor](https://cursor.com)
This PR adds a `workspace:export` server command for ongoing debug
tooling/administrative work
Demo Video shows
1. Exporting a sample workspace YC
2. Restoring Local DB Docker volume snapshot to Dropped YC Workspace
3. Importing exported SQL
4. Booting and navigating the imported Workspace
https://github.com/user-attachments/assets/0e1ac6cb-8ce1-440b-8b56-f81dcb27a9c8
- Adds template variable interpolation (`${...}`) to command menu item
labels and short labels, enabling dynamic text like `Create new
${capitalize(objectMetadataItem.labelSingular)}` instead of static
`Create new record`.
- Supports `capitalize` and `lowercase` transform functions within
template expressions.
Graph SDK occasionally returns a 400 with a null error message which is
not a real bad request but a transient hiccup.
Classify these as temporary errors so they get retried instead of
flooding Sentry.
Fixes TWENTY-SERVER-D3X
## Summary
- Fixes object `color` to use the `standardOverrides` mechanism for
standard objects, matching how `label`, `description`, and `icon`
already work
- Previously, color was written directly to the `objectMetadata.color`
column for **both** standard and custom objects, which meant user color
customizations on standard objects could be overwritten during metadata
syncs
- Custom objects continue to have `color` updated directly on the entity
(no change)
## Changes
| File | What changed |
|------|-------------|
| `object-metadata-standard-overrides-properties.constant.ts` | Added
`'color'` to `OBJECT_METADATA_STANDARD_OVERRIDES_PROPERTIES` so
`sanitizeRawUpdateObjectInput` routes color into `standardOverrides` for
standard objects |
| `resolve-object-metadata-standard-override.util.ts` | Extended to
support `'color'` as a key — handled like `icon` (no i18n/translation,
just direct override check) |
| `object-metadata.resolver.ts` | Added `@ResolveField` for `color` that
resolves through `resolveObjectMetadataStandardOverride`, matching the
existing `labelSingular`/`labelPlural`/`description`/`icon` resolve
fields |
| `flat-object-metadata-validator.service.ts` | Removed `'color'` from
`allowedOverrideKeys` for system objects since it now flows through
`standardOverrides` |
| `resolve-object-metadata-standard-override.util.spec.ts` | Added test
cases for custom object color, standard object color override, and
standard object color fallback |
| `successful-update-one-standard-object-metadata.integration-spec.ts` |
Added `'when updating color'` test case, included `color` in GraphQL
queries and `standardOverrides` fragment, reset color in `afterEach`
cleanup |
## How it works now
| Object type | Color update flow |
|---|---|
| **Custom** | Written directly to `objectMetadata.color` column |
| **Standard** | Stored in `objectMetadata.standardOverrides.color`,
resolved via `@ResolveField` at query time |
This is identical to how `label`, `description`, and `icon` have always
worked.
## Test plan
- [x] Unit tests pass
(`resolve-object-metadata-standard-override.util.spec.ts` — 21 tests)
- [x] Typecheck passes (`npx nx typecheck twenty-server`)
- [x] Lint passes (`npx nx lint:diff-with-main twenty-server`)
- [ ] Integration test snapshot regenerates correctly
(`successful-update-one-standard-object-metadata`)
- [ ] Verify standard object color editing from sidebar persists via
`standardOverrides`
- [ ] Verify custom object color editing from sidebar persists directly
on entity
Made with [Cursor](https://cursor.com)
Threads with no FROM participants have no entry in the participants map,
causing extractParticipantSummary to receive undefined and crash
Fixes TWENTY-SERVER-FB8
## Summary
- **BackfillMissingStandardViewsCommand**: When view validation fails
(e.g. a viewField references a field metadata that doesn't exist in the
workspace), log a warning and skip instead of throwing — so the
workspace upgrade continues with the remaining commands.
- **AddMissingSystemFieldsToStandardObjectsCommand**: Wrap both the
non-tsVector batch migration and each individual tsVector migration in
try-catch. If a field already exists (e.g. duplicate key on `name +
objectMetadataId + workspaceId`), the error is logged as a warning and
the command moves on to the next field.
These errors were observed during the 1.18 → 1.19 production upgrade for
workspaces with non-standard state (missing "owner" field metadata on
Opportunity, or searchVector fields already present with a different
universalIdentifier).
## Test plan
- [ ] Re-run upgrade on the affected production workspaces
- [ ] Verify upgrade completes successfully with warnings instead of
failures
- [ ] Confirm that workspaces which were already upgrading cleanly are
unaffected
Made with [Cursor](https://cursor.com)
## Summary
- **Optimistic metadata store updates**: Replace `refetchQueries` with
direct `addToDraft`/`applyChanges` calls in create, update, and delete
navigation menu item mutation hooks for instant UI feedback. Client-side
UUID generation enables optimistic creates before the server responds.
- **SSE event enrichment with `targetRecordIdentifier`**: Introduce
`NavigationMenuItemRecordIdentifierService` to resolve record display
info (label, image) and enrich SSE metadata events at emission time, so
the sidebar shows record names immediately without a page refresh.
- **Centralized role permission resolution**: Add
`resolveRolePermissionConfigFromAuthContext` to `PermissionsService`,
removing duplicated role resolution logic from individual services.
- **Mutation fragments include `targetRecordIdentifier`**: Switch
create/update/delete mutations from `NavigationMenuItemFields` to
`NavigationMenuItemQueryFields` so the mutation response includes
`targetRecordIdentifier`, preventing a brief gap where RECORD favorites
are invisible in the sidebar.
- **Folder UI fixes**: Remove transparent border on
`StyledFolderContainer` that caused a 1px size inconsistency between
folder and non-folder items in Favorites. Make the folder kebab menu
hover-only instead of always visible.
## Summary
- Upgrade `ink` from 5.1.1 to 6.8.0 in twenty-sdk (React 19 required, no
API breaking changes)
- Upgrade `react`/`react-dom` from 18 to 19 and
`@types/react`/`@types/react-dom` to 19 in twenty-sdk
- Enable `incrementalRendering` — only redraws changed lines instead of
full output, reducing flickering
- Pause the animation timer when the pipeline is not actively building
or syncing, so Ink stops re-rendering and the terminal becomes
scrollable (fixes inability to scroll up to read errors)
- Remove `AnimationProvider` context — derive animation frames from
`Date.now()` directly in `useStatusIcon`
- Export `NavigationMenuItemType` from `twenty-sdk` (re-exported from
`twenty-shared/types`)
- Add dedicated NavigationMenuItem integration tests to postcard-app
(unique positions, unique identifiers, valid object references)
## Test plan
- [ ] Run `twenty dev` and verify the TUI renders normally during
build/sync
- [ ] Trigger an error and verify the terminal output freezes and
becomes scrollable
- [ ] Verify that after fixing the error, the TUI resumes animating on
next build cycle
- [ ] Verify `import { NavigationMenuItemType } from "twenty-sdk"` works
- [ ] Run postcard-app integration tests and verify new
NavigationMenuItem tests pass
- enrich SSE events with relations
- remove queries from sse metadata events
- on sse event, manage store
- on object/field metadata changes, manage store
TODO left:
- fix other metadata items
---------
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Co-authored-by: github-actions <github-actions@twenty.com>
Co-authored-by: Charles Bochet <charles@twenty.com>
## Description
- Introduces a new engine command execution model that replaces the
previous approach of mapping `EngineComponentKey` to React components.
Instead, engine commands are now mounted headlessly via
`HeadlessEngineCommandMountRoot`, with their execution context populated
synchronously before mounting.
- Creates new headless command components
- Moves error handling from the SDK layer to the host app by wrapping
all mounted commands with a new `CommandMenuItemErrorBoundary`
The new flow works as follows:
- When a command menu item with an `engineComponentKey` is clicked,
`useCommandMenuItemFrontComponentCommands` calls
`useMountEngineCommand`, which synchronously reads the current context
store (object metadata, selected records, filters, view ID, etc.) and
writes a `MountedEngineCommandContext` into
`mountedEngineCommandsState`.
- The command is then mounted into `mountedEngineCommandsState`, which
triggers `HeadlessEngineCommandMountRoot` to render the corresponding
headless component from `ENGINE_COMPONENT_KEY_HEADLESS_COMPONENT_MAP`,
wrapped in `CommandMenuItemErrorBoundary`,
`ContextStoreComponentInstanceContext.Provider`, and
`EngineCommandComponentInstanceContext.Provider`.
- Each command component reads its execution context and delegates to
one of the 4 execution patterns: `HeadlessEngineCommandWrapperEffect`
(simple actions), `HeadlessConfirmationModalEngineCommandEffect`
(destructive actions needing confirmation),
`HeadlessNavigateEngineCommand` (GO_TO_* commands), or
`HeadlessOpenSidePanelPageEngineCommand` (SEARCH_RECORDS, ASK_AI,
VIEW_PREVIOUS_AI_CHATS).
- After execution, the command self-unmounts via
`useUnmountEngineCommand`, which removes the entry from
`mountedEngineCommandsState` and stops rendering the component.
### What
Unifies record page layout editing and navigation menu editing into a
single global "layout customization" session. Dashboard editing stays
separate.
### How it works
**Two edit mode systems, one context-based read:**
- `isLayoutCustomizationModeEnabledState` -- global atom for record
pages + navigation
- `isDashboardInEditModeComponentState` -- dashboard-only, independent
per-component atom
- `PageLayoutEditModeProvider` -- context that dispatches to
`RecordPageLayoutEditModeProvider` (reads global atom) or
`DashboardPageLayoutEditModeProvider` (reads component atom), one
component per file
**Session registry + independent atoms:**
- `activeCustomizationPageLayoutIdsState` -- accumulates page layout IDs
as user navigates during customization (`string[]`)
- Save/cancel iterate the ID list and read each layout's draft/persisted
atoms independently
- Follows the same pattern as `settingsRoleIdsState` +
`settingsDraftRoleFamilyState`
**Unified UI:**
- `LayoutCustomizationBar` replaces the old `NavigationMenuEditModeBar`
- Enter once -- edit record layouts + navigation -- save/cancel
everything together
- `useSaveLayoutCustomization` orchestrates sequential save: navigation
draft -- page layouts -- field widget groups
- Error snackbar on partial save failure (with TODO for future atomic
server mutation)
**Draft protection during customization:**
- `PageLayoutRelationWidgetsSyncEffect` guarded -- only updates
persisted state from server, skips draft/currentLayouts while
customization is active
- `useExecuteTasksOnAnyLocationChange` skips draft reset when
customization mode is enabled
- Command execution blocked during layout customization
### Cleanup
- Deleted `NavigationMenuEditModeBar`,
`isNavigationMenuInEditModeState`,
`isPageLayoutInEditModeComponentState`,
`useIsGlobalLayoutCustomizationActive`
- `DraftPageLayout` type changed from `Omit` to `Pick` (explicit fields)
- Removed save/cancel from `DefaultRecordCommandMenuItemsConfig` (bar
handles it now)
- Extracted `useSaveFieldsWidgetGroups` from save orchestration
- Split `PageLayoutEditModeProvider` into 3 separate files (one
component per file, Twenty convention)
### Known issues
- **Stale deleted widget after save (pre-existing on `main`)**: Delete
widget -- save -- exit customization -- Apollo cache stale -- sync
effect overwrites Jotai from stale data -- widget reappears until
refresh. Separate PR needed, likely tied to the planned server-side
`saveLayoutCustomization` atomic endpoint.
### Open questions
- **Module location**: Layout customization hooks/states live in `/app`
-- should they move to their own `modules/layout-customization/`?
- **Atomic server mutation**: All save mutations are on metadata schema
(`createNavigationMenuItem`, `deleteNavigationMenuItem`,
`updateNavigationMenuItem`, `updatePageLayoutWithTabsAndWidgets`,
`upsertFieldsWidget`). A single `saveLayoutCustomization` endpoint could
make saves truly atomic.
https://github.com/user-attachments/assets/036ef542-97f3-485b-a68f-3726002c81fb
Fixes#18607
So previously i followed the frontend approach which led to
architectural mismatch.
- We already have the correct things implemented in the
`processViewNameWithTemplate` and in object metadata service. Found that
the seeder files had the hardcoded names instead of {objectLabelPlural}.
So changed all the hardcoded string to contain {objectLabelPlural} to
seed the new workspaces accurately.
- it will have a follow up PR for typeORM migration to migrate the
existing workspaces
https://github.com/user-attachments/assets/3b460f9f-c59d-49d1-8baa-17322d696ecc
I hope i am correct this time :)
Co-authored-by: Arun kumar <arunkumar@Aruns-MacBook-Air.local>
## Summary
- Reorganizes the `navigation-menu-item` frontend module from a flat
structure into `common/`, `display/`, and `edit/` subfolders with
type-specific subdirectories (`link/`, `folder/`, `object/`, `view/`,
`record/`)
- Every file is now in a leaf folder describing its type: `components/`,
`hooks/`, `utils/`, `types/`, or `constants/`
- Moves 14 NavigationMenuItem-related components out of
`object-metadata/` and `side-panel/pages/` into the
`navigation-menu-item` module where they belong
- Creates type-specific display utility functions (e.g.,
`getLinkNavigationMenuItemLabel`,
`getObjectNavigationMenuItemComputedLink`) to replace generic
switch-based functions
- Unifies the Favorites section drag-and-drop from `@hello-pangea/dnd`
to `@dnd-kit/react`, matching the Workspace section's DnD library
- Renames Favorites section components from `CurrentWorkspaceMember*` to
`Favorites*` for clarity
- Deletes unused `FavoritesDragDropProviderContent` and
`NavbarDragProvider`
## Test plan
- [x] `npx nx typecheck twenty-front` passes
- [x] `npx nx lint:diff-with-main twenty-front` passes (0 warnings, 0
errors)
- [x] `npx nx test twenty-front` passes (763 suites, 4467 tests)
- [ ] Verify favorites drag-and-drop still works in the UI (reorder
items, move between folders)
- [ ] Verify workspace edit mode drag-and-drop still works
- [ ] Verify "add to navigation" drag from command menu/side panel still
works
## Summary
- **Delete 10 unused files**: 7 hooks (`useWorkflowRunUnsafe`,
`useGetViewById`, `useCreateViewFieldGroup`, `useDeleteViewFieldGroup`,
`useUpdateViewFieldGroup`, `useCreateManyViewFieldGroups`,
`useMoveViewColumns` + test), 1 component (`SettingsSummaryCard`), 1
utility (`createEventContext`)
- **Rename `objectMetadataItemsState` → `objectMetadataItemsSelector`**
across ~85 files to accurately reflect it is a derived selector (via
`createAtomSelector`), not a base Jotai atom
## Details
### Dead code removed
| Type | Name | Reason |
|------|------|--------|
| Hook | `useWorkflowRunUnsafe` | Never imported — duplicate of
`useWorkflowRun` without schema validation |
| Hook | `useGetViewById` | Never imported — `useViewById` is used
instead |
| Hook | `useCreateViewFieldGroup` | Never imported — CRUD done via
`usePerformViewFieldGroupAPIPersist` |
| Hook | `useDeleteViewFieldGroup` | Same as above |
| Hook | `useUpdateViewFieldGroup` | Same as above |
| Hook | `useCreateManyViewFieldGroups` | Same as above |
| Hook | `useMoveViewColumns` | Only imported by its own test — no
production usage |
| Component | `SettingsSummaryCard` | Never imported anywhere |
| Utility | `createEventContext` | Never imported anywhere |
### Rename
`objectMetadataItemsState` is created via `createAtomSelector` (it
derives from `objectMetadataItemsWithFieldsSelector`), so naming it
`*State` is misleading. Renamed to `objectMetadataItemsSelector` for
consistency with sibling selectors like
`objectMetadataItemsByNamePluralMapSelector`.
## Summary
- Removes `ProcessedNavigationMenuItem` type and the
`sortNavigationMenuItems` enrichment pipeline that pre-computed display
fields (label, link, icon, avatarUrl, etc.) for every navigation menu
item upfront
- Replaces with `filterAndSortNavigationMenuItems` (pure filter+sort
returning raw `NavigationMenuItem[]`) and small utility functions
(`getNavigationMenuItemLabel`, `getNavigationMenuItemComputedLink`,
`getNavigationMenuItemObjectNameSingular`) that components call on
demand
- Eliminates the redundant `itemType` field (was identical to the raw
`type` field) across ~30 consumer files
- Each type-specific renderer
(`NavigationDrawerItemForObjectMetadataItem`, `NavigationMenuItemIcon`,
link/folder components) now derives only the 1-2 display fields it
actually needs from the raw item + globally available Jotai atoms
Net result: -1199 / +1177 lines, 7 files deleted, 4 new utility files.
## Summary
- Adds a `color` column to `ObjectMetadataEntity` with full GraphQL
support so object icon colors are persisted at the metadata level
- Adds a `type` column to `NavigationMenuItemEntity` (enum: `OBJECT`,
`VIEW`, `FOLDER`, `LINK`, `RECORD`) replacing field-based type inference
- Updates frontend to read object colors from `objectMetadata.color`
(falling back to standard defaults) in the sidebar nav, record index
header, and record show breadcrumb
- Simplifies `NavigationMenuItemIcon` color resolution via
`getEffectiveNavigationMenuItemColor` util
## Color rules
| Item type | Color source | Editable in sidebar? |
|-----------|-------------|---------------------|
| **Object** | `objectMetadata.color` | Yes — persisted to
`objectMetadata.color` on Save |
| **Folder** | `navigationMenuItem.color` | Yes |
| **Link** | Fixed default (`DEFAULT_NAVIGATION_MENU_ITEM_COLOR_LINK`) |
No |
| **View** | `objectMetadata.color` (from the parent object) | No |
| **Record** | None | No |
- **Object** items represent the whole object (e.g. "Companies") and
point to the INDEX view. Changing their color updates
`objectMetadata.color` via `useSaveObjectMetadataColorsFromDraft`.
- **View** items represent specific non-INDEX views. Their color comes
from the parent object's metadata (read-only).
- Only **folders** store their color on `navigationMenuItem.color` —
enforced by `hasNavigationMenuItemOwnColor` util.
- `getEffectiveNavigationMenuItemColor` returns `objectColor` for both
OBJECT and VIEW items, folder's own color for folders, and the fixed
default for links.
## NavigationMenuItemType enum
- Shared enum created in `twenty-shared` with values: `OBJECT`, `VIEW`,
`FOLDER`, `LINK`, `RECORD`
- Registered as a GraphQL enum on the backend
- Replaces string literals across entity, DTOs, input, converters, and
frontend hooks
- Migration backfills existing rows: INDEX views → `OBJECT`, non-INDEX
views → `VIEW`, based on join with the view table
## Design decisions
- **OBJECT vs VIEW distinction**: Items pointing to INDEX views are
typed as `OBJECT` (represent the whole object, color editable). Items
pointing to non-INDEX views are typed as `VIEW` (specific view, color
read-only from parent object).
- **Dual color storage**: `navigationMenuItem.color` is preserved for
folders only. Objects use `objectMetadata.color` as their source of
truth.
- **Type discriminator**: The `type` column replaces field-based
inference (checking `viewId`, `link`, `targetRecordId` presence) with an
explicit enum, simplifying `isNavigationMenuItemLink` /
`isNavigationMenuItemFolder` to simple `item.type ===` checks.
- **No settings page color picker**: Object color editing is done from
the sidebar edit panel, not the data model settings page.
## Test plan
- [ ] Verify objects display their default standard colors in the
sidebar
- [ ] Verify object color editing works in the sidebar edit panel
(persists to objectMetadata.color)
- [ ] Verify folder color editing works in the sidebar edit panel
- [ ] Verify views, links, and records do NOT show a color picker in the
sidebar edit panel
- [ ] Run `npx nx typecheck twenty-front` and `npx nx typecheck
twenty-server`
- [ ] Verify the database migrations add `color` to `objectMetadata` and
`type` to `navigationMenuItem`
Made with [Cursor](https://cursor.com)
## Summary
Closes#18673
Some languages (e.g., German "Unternehmen") and even English words
(sheep, deer, aircraft, series) have identical singular and plural
forms. Twenty previously blocked saving when labels matched, making it
impossible to correctly name objects in these cases.
- **Labels** are purely display strings — removed the equality
validation from both the frontend Zod schema and backend validator
- **API names** (nameSingular/namePlural) must stay different since they
generate distinct GraphQL resolvers (`findOne` vs `findMany`,
`createOne` vs `createMany`, etc.) and REST endpoints — this validation
is preserved
- Added a shared `computeMetadataNamesFromLabels` util in
`twenty-shared` that auto-appends `'s'` to the plural API name when both
labels produce the same camelCase name (e.g., "Unternehmen" →
`unternehmen` / `unternehmens`)
- Both the frontend form and backend sync-check use the same shared util
— single source of truth, no duplicated logic
**No retroactive impact**: since the old code prevented identical labels
from ever being saved, no existing workspace has `labelSingular ===
labelPlural`.
## Test plan
- [x] New unit tests for `computeMetadataNamesFromLabels` (7 tests:
standard labels, Sheep, Unternehmen, Aircraft, empty labels, different
labels, applyCustomSuffix)
- [x] Updated frontend schema validation tests (identical labels with
different names now passes; identical names still fails)
- [x] Updated backend integration test cases (removed identical-label
failing cases)
- [ ] Manual: create a new object with identical singular/plural labels
(e.g. "Sheep" / "Sheep") — should save successfully with API names
`sheep` / `sheeps`
- [ ] Manual: verify existing objects with different labels still work
unchanged
Made with [Cursor](https://cursor.com)
## Fix: Standard object rename ignored when UI language is not English
(Closes#18650)
### Problem
When a user renames a standard object (for example, changing **"People"
→ "Contacts"**), the custom name is only respected when the UI language
is set to **English**.
For any other locale, the resolver ignores the user-defined override and
falls back to the i18n translation of the original default label.
As a result, the custom name defined by the user is not displayed when
the UI language changes.
### Root Cause
`resolveObjectMetadataStandardOverride` only applied direct overrides
when the locale matched `SOURCE_LOCALE` (English):
```ts
if (
safeLocale === SOURCE_LOCALE &&
isNonEmptyString(objectMetadata.standardOverrides?.[labelKey])
) {
return objectMetadata.standardOverrides[labelKey] ?? '';
}
## Summary
- Removes the redundant `DATABASE_STATEMENT_TIMEOUT_MS` config variable
(default 15s) from `ConfigVariables`
- Updates the core TypeORM datasource to use
`PG_DATABASE_PRIMARY_TIMEOUT_MS` (default 10s) for its `query_timeout`,
aligning it with the workspace datasource which already uses this
variable
- This consolidates two separate env vars that controlled the same
concern (database query timeout) into a single one
## Summary
Fixes#18524
Fixes the MCP response contract for non-`initialize` methods.
Previously, `/mcp` returned initialize-style metadata for methods like
`tools/list`, which caused strict MCP clients to reject the response
shape. The endpoint also returned `201 Created` for RPC calls even
though no resource was being created.
## Changes
- return only method-specific payloads for MCP list methods
- `tools/list` -> `{ tools: [...] }`
- `prompts/list` -> `{ prompts: [] }`
- `resources/list` -> `{ resources: [] }`
- keep MCP server metadata only on `initialize`
- make `/mcp` return `200 OK` instead of `201 Created`
- add regression tests for:
- `tools/list` response shape
- `prompts/list` response shape
- `resources/list` response shape
## Why
Strict MCP clients expect:
- standard RPC transport semantics over HTTP
- method-specific JSON-RPC result payloads
Returning initialize metadata for non-`initialize` methods breaks that
expectation and can cause client deserialization or protocol validation
failures.
## Verification
- reproduced the issue locally against `/mcp`
- verified `tools/list` was previously returning initialize-style fields
- verified `tools/list` now returns only `result.tools`
- verified `/mcp` now returns `200 OK`
- ran targeted Jest tests:
```bash
cd /Users/apple/MyProjects/OpenSource/twenty/packages/twenty-server
npx jest --runInBand src/engine/api/mcp/services/__tests__/mcp-protocol.service.spec.ts src/engine/api/mcp/services/__tests__/mcp-tool-executor.service.spec.ts
---------
Co-authored-by: Félix Malfait <felix@twenty.com>
## Summary
- **Remove all "core" prefixes** from the views system — the
metadata-based storage migration is complete, so `CoreView`,
`coreViewsSelector`, `getCoreViews`, etc. are now just `View`,
`viewsSelector`, `getViews`
- **Eliminate the entire converter layer** (15 files, ~850 lines
deleted) — `convertCoreViewToView` and all sub-converters were either
no-ops or trivially adding `__typename` / mapping identical enum values.
Local enums now re-export from generated GraphQL types directly (single
source of truth)
- **Unify `View` and `ViewWithRelations`** into one type —
`ViewWithRelations` is now a type alias for `View`, selectors return
data directly without conversion
### Backend
- Rename `@ObjectType('CoreView')` → `@ObjectType('View')` (and all
sub-entities)
- Rename resolver methods: `getCoreViews` → `getViews`, `createCoreView`
→ `createView`, etc.
- Rename `FIND_ALL_CORE_VIEWS_GRAPHQL_OPERATION` →
`FIND_ALL_VIEWS_GRAPHQL_OPERATION`
### Frontend
- Delete 15 converter files (`convertGqlView*ToView*`,
`convertView*ToGql`, `convertViewWithRelationsToView`)
- Re-export `ViewType`, `ViewKey`, `ViewFilterGroupLogicalOperator` from
generated enums (no more duplicate enum definitions with different
casing)
- Replace `ViewOpenRecordInType` with `ViewOpenRecordIn` from generated
- Remove `__typename` from all local view sub-types
- Remove unused `variant` from `ViewFilter`, make `displayValue` and
`definition` optional
- Rename ~45 GraphQL query/mutation files and all selectors to drop
"core" prefix
- Delete unused `viewsWithRelationsSelector`
## Summary
This PR implements OAuth 2.0 Dynamic Client Registration (RFC 7591) and
OAuth 2.0 Protected Resource Metadata (RFC 9728) support, enabling
third-party applications to dynamically register as OAuth clients
without manual configuration.
## Key Changes
### OAuth Dynamic Client Registration
- **New Controller**: `OAuthRegistrationController` at `POST
/oauth/register` endpoint
- Validates client metadata according to RFC 7591 specifications
- Enforces PKCE-only public client model (no client secrets)
- Supports only `authorization_code` grant type and `code` response type
- Rate limits registrations to 10 per hour per IP address
- Returns `client_id` and registration metadata in response
- **Input Validation**: `OAuthRegisterInput` DTO with constraints on:
- Client name (max 256 chars)
- Redirect URIs (max 20, validated for security)
- Grant types, response types, scopes, and auth methods
- Logo and client URIs (max 2048 chars)
- **Discovery Endpoint Update**: Added `registration_endpoint` to OAuth
discovery metadata
### Stale Registration Cleanup
- **Cleanup Service**: Automatically removes OAuth-only registrations
older than 30 days that have no active installations
- **Cron Job**: Runs daily at 02:30 AM UTC with batch processing (100
records per batch)
- **CLI Command**: `cron:stale-registration-cleanup` to manually trigger
cleanup
### MCP (Model Context Protocol) Authentication
- **New Guard**: `McpAuthGuard` implements RFC 9728 compliance
- Wraps JWT authentication with proper error responses
- Returns `WWW-Authenticate` header with protected resource metadata URL
on 401
- Enables OAuth-protected MCP endpoints
### Protected Resource Metadata
- **New Endpoint**: `GET /.well-known/oauth-protected-resource` (RFC
9728)
- Advertises MCP resource as OAuth-protected
- Lists supported scopes and bearer token methods
- Enables OAuth clients to discover authorization requirements
### Application Registration Updates
- **New Source Type**: `OAUTH_ONLY` enum value for OAuth-only
registrations
- **Install Service**: Skips artifact installation for OAuth-only apps
(no code artifacts)
### Frontend Updates
- **Authorization Page**: Support both snake_case (standard OAuth) and
camelCase (legacy) query parameters
- `client_id` / `clientId`
- `code_challenge` / `codeChallenge`
- `redirect_uri` / `redirectUrl`
## Implementation Details
- **Rate Limiting**: Uses token bucket algorithm with 10 registrations
per 3,600,000ms window per IP
- **Scope Validation**: Requested scopes are capped to allowed OAuth
scopes; defaults to all scopes if not specified
- **Redirect URI Validation**: Uses existing `validateRedirectUri`
utility for security
- **Cache Headers**: Registration responses include `Cache-Control:
no-store` and `Pragma: no-cache`
- **Batch Processing**: Cleanup operations process 100 records at a time
to avoid memory issues
- **Grace Period**: 30-day grace period before cleanup to allow time for
client activation
https://claude.ai/code/session_01PxcuWFFRuXMASMaMGTLYk2
---------
Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com>
Co-authored-by: github-actions <github-actions@twenty.com>
Co-authored-by: cubic-dev-ai[bot] <191113872+cubic-dev-ai[bot]@users.noreply.github.com>
## Summary
Replace the single `metadataVersion` integer with per-entity-type
**collection hashes** for granular metadata staleness detection. The
backend already generates a UUID per flat entity map on each cache
recompute (`crypto.randomUUID()` in `WorkspaceCacheService`); we now
expose these via the minimal metadata endpoint and SSE events so the
frontend can compare and know exactly which entity types are stale.
### Key changes
**Backend:**
- `WorkspaceCacheService.getCacheHashes()` — new public method that
reads only `:hash` keys from Redis without fetching full data
- `MinimalMetadataDTO` — added `collectionHashes: Record<string,
string>` (JSON scalar mapping `AllMetadataName` → collection hash),
removed `metadataVersion`
- `MetadataEventDTO` — added optional `updatedCollectionHash` field to
SSE events
- `MetadataEventsToDbListener` — reads the collection hash for the
affected entity type after cache invalidation and attaches it to the SSE
event before publishing
- `MinimalMetadataService` — no longer queries the workspace table; uses
`getCacheHashes()` for all flat entity maps and maps cache keys to
`AllMetadataName` locally
**Frontend:**
- `metadataCollectionHashesState` — new Jotai atom with
`atomWithStorage` + `getOnInit: true` storing
`Partial<Record<MetadataEntityKey, string>>`
- `mapAllMetadataNameToEntityKey()` — explicit mapping from backend
`AllMetadataName` to frontend `MetadataEntityKey` (23 entries)
- `useLoadMinimalMetadata` — stores `collectionHashes` from server,
computes `staleEntityKeys` by comparing local vs server hashes
- `patchMetadataStoreFromSSEEvent()` — accepts optional
`updatedCollectionHash` and updates `metadataCollectionHashesState`
- All 11 SSE effect components — pass
`eventDetail.updatedCollectionHash` through to the patch function
- `useStaleMetadataEntities` — new hook returning entity keys missing
from collection hashes (not yet loaded/synced)
- `resetMetadataStore()` — also clears collection hashes
- Deleted `metadataVersionState` (superseded by collection hashes)
### Design decisions
- **No change to hash generation** — existing `crypto.randomUUID()` is
sufficient. Hashes are persisted in Redis, survive server restarts, and
change only on `invalidateAndRecompute`.
- **"Collection hash" naming** — used consistently to clarify the hash
represents an entire entity collection (e.g., all views), not a single
record.
- **Mapping localized** — backend `WorkspaceCacheKeyName` →
`AllMetadataName` mapping lives in the minimal metadata service.
Frontend `AllMetadataName` → `MetadataEntityKey` mapping lives in a
local utility. Nothing in `twenty-shared`.
- **Backward compatible** — `collectionHashes` is additive;
`updatedCollectionHash` is nullable.
2026-03-14 23:38:37 +01:00
3060 changed files with 105511 additions and 49210 deletions
-`skills/example-skill.ts` — Example AI agent skill definition
-`__tests__/app-install.integration-test.ts` — Integration test that builds, installs, and verifies the app (includes `vitest.config.ts`, `tsconfig.spec.json`, and a setup file)
## Local server
The scaffolder can start a local Twenty dev server for you (all-in-one Docker image with PostgreSQL, Redis, server, and worker). You can also manage it manually:
```bash
yarn twenty server start # Start (pulls image if needed)
yarn twenty server status # Check if it's healthy
yarn twenty server logs # Stream logs
yarn twenty server stop # Stop (data is preserved)
yarn twenty server reset # Wipe all data and start fresh
```
The server is pre-seeded with a workspace and user (`tim@apple.dev` / `tim@apple.dev`).
## Next steps
- Run `yarn twenty help` to see all available commands.
- Use `yarn twenty auth:login` to authenticate with your Twenty workspace.
- Explore the generated project and add your first entity with `yarn twenty entity:add` (logic functions, front components, objects, roles, views, navigation menu items, skills).
- Use `yarn twenty app:dev` while you iterate — it watches, builds, and syncs changes to your workspace in real time.
-`CoreApiClient` (for workspace data via `/graphql`) is auto-generated by `yarn twenty app:dev`. `MetadataApiClient` (for workspace configuration and file uploads via `/metadata`) ships pre-built with the SDK. Both are available via `import { CoreApiClient, MetadataApiClient } from 'twenty-sdk/clients'`.
- Use `yarn twenty remote add --local` to authenticate with your Twenty workspace via OAuth.
- Explore the generated project and add your first entity with `yarn twenty add` (logic functions, front components, objects, roles, views, navigation menu items, skills).
- Use `yarn twenty dev` while you iterate — it watches, builds, and syncs changes to your workspace in real time.
-`CoreApiClient` (for workspace data via `/graphql`) is auto-generated by `yarn twenty dev`. `MetadataApiClient` (for workspace configuration and file uploads via `/metadata`) ships pre-built with the SDK. Both are available via `import { CoreApiClient, MetadataApiClient } from 'twenty-sdk/clients'`.
## Build and publish your application
@@ -121,19 +135,19 @@ Once your app is ready, build and publish it using the CLI:
```bash
# Build the app (output goes to .twenty/output/)
yarn twenty app:build
yarn twenty build
# Build and create a tarball (.tgz) for distribution
yarn twenty app:build --tarball
yarn twenty build --tarball
# Publish to npm (requires npm login)
yarn twenty app:publish
yarn twenty publish
# Publish with a dist-tag (e.g. beta, next)
yarn twenty app:publish --tag beta
yarn twenty publish --tag beta
# Publish directly to a Twenty server (builds, uploads, and installs in one step)
yarn twenty auth:list # List all configured workspaces
# Application
yarn twenty app:dev # Start dev mode (watch, build, sync, and auto-generate typed client)
yarn twenty entity:add # Add a new entity (object, field, function, front-component, role, view, navigation-menu-item)
yarn twenty function:logs # Stream function logs
yarn twenty function:execute # Execute a function with JSON payload
yarn twenty app:uninstall # Uninstall app from workspace
```
## Integration Tests
If your project includes the example integration test (`src/__tests__/app-install.integration-test.ts`), you can run it with:
```bash
# Make sure a Twenty server is running at http://localhost:3000
yarn test
```
The test builds and installs the app, then verifies it appears in the applications list. Test configuration (API URL and API key) is defined in `vitest.config.ts`.
## LLMs instructions
Main docs and pitfalls are available in LLMS.md file.
Run `yarn twenty help` to list all available commands.
## Learn More
To learn more about Twenty applications, take a look at the following resources:
- [twenty-sdk](https://www.npmjs.com/package/twenty-sdk) - learn about `twenty-sdk` tool.
IF NOT EXISTS (SELECT 1 FROM pg_roles WHERE rolname = :'app_user') THEN
EXECUTE format('CREATE USER %I WITH PASSWORD %L', :'app_user', :'app_password');
END IF;
END
$do$;
EOSQL
echo "Creating core schema and granting permissions..."
psql -h {{ include "twenty.fullname" . }}-db -p 5432 -U postgres -d "${DBNAME}" -v app_user="${APP_USER}" -c 'CREATE SCHEMA IF NOT EXISTS core'
psql -h {{ include "twenty.fullname" . }}-db -p 5432 -U postgres -d "${DBNAME}" -v db="${DBNAME}" -v app_user="${APP_USER}" -c 'GRANT ALL PRIVILEGES ON DATABASE :"db" TO :"app_user";'
@@ -106,31 +106,6 @@ spec:
psql -h {{ include "twenty.fullname" . }}-db -p 5432 -U postgres -d "${DBNAME}" -v app_user="${APP_USER}" -c 'ALTER DEFAULT PRIVILEGES IN SCHEMA public GRANT ALL ON SEQUENCES TO :"app_user";'
echo "Database ${DBNAME} is ready."
{{- end }}
- name:run-migrations
{{- $img := include "twenty.server.image" . }}
image:{{include "twenty.image.repository" $img }}:{{ include "twenty.image.tag" $img }}
LABEL org.opencontainers.image.description="All-in-one Twenty image for local development and SDK usage. Includes PostgreSQL, Redis, server, and worker."
@@ -20,19 +20,51 @@ Apps let you build and manage Twenty customizations **as code**. Instead of conf
## Prerequisites
- Node.js 24+ and Yarn 4
- A Twenty workspace and an API key (create one at https://app.twenty.com/settings/api-webhooks)
- Docker (for the local Twenty dev server)
## Getting Started
Create a new app using the official scaffolder, then authenticate and start developing:
Create a new app using the official scaffolder. It can automatically start a local Twenty instance for you:
```bash filename="Terminal"
# Scaffold a new app (includes all examples by default)
# Scaffold a new app — the CLI will offer to start a local Twenty server
npx create-twenty-app@latest my-twenty-app
cd my-twenty-app
# Start dev mode: automatically syncs local changes to your workspace
yarn twenty app:dev
yarn twenty dev
```
### Local Server Management
The SDK includes commands to manage a local Twenty dev server (all-in-one Docker image with PostgreSQL, Redis, server, and worker):
```bash filename="Terminal"
# Start the local server (pulls the image if needed)
yarn twenty server start
# Check server status
yarn twenty server status
# Stream server logs
yarn twenty server logs
# Stop the server
yarn twenty server stop
# Reset all data and start fresh
yarn twenty server reset
```
The local server comes pre-seeded with a workspace and user (`tim@apple.dev` / `tim@apple.dev`), so you can start developing immediately without any manual setup.
### Authentication
Connect your app to the local server using OAuth:
```bash filename="Terminal"
# Authenticate via OAuth (opens browser)
yarn twenty remote add --local
```
The scaffolder supports two modes for controlling which example files are included:
Run `UPDATE core."user" SET "canAccessFullAdminPanel" = TRUE WHERE email = 'you@yourdomain.com';` in database container to get access to admin panel.
#### When running workflow, workflow run fails with "Logic function execution is disabled. Set LOGIC_FUNCTION_TYPE to LOCAL or LAMBDA to enable."
In production, logic functions are disabled by default. Set the `LOGIC_FUNCTION_TYPE` environment variable to `LOCAL` or `LAMBDA` to enable them. This can be configured via environment variables or through the admin panel database variables. See the [Logic Functions setup guide](/developers/self-host/capabilities/setup#logic-functions-available-drivers) for details.
قم بتشغيل `UPDATE core."user" SET "canAccessFullAdminPanel" = TRUE WHERE email = 'you@yourdomain.com';` في حاوية قاعدة البيانات للحصول على الوصول إلى لوحة الإدارة.
#### عند تشغيل سير العمل، يفشل تشغيل سير العمل مع الرسالة "تم تعطيل تنفيذ دوال المنطق. عيّن LOGIC_FUNCTION_TYPE إلى LOCAL أو LAMBDA لتمكين ذلك."
في بيئة الإنتاج، يتم تعطيل دوال المنطق افتراضيًا. قم بتعيين متغير البيئة `LOGIC_FUNCTION_TYPE` إلى `LOCAL` أو `LAMBDA` لتمكينها. يمكن تكوين ذلك عبر متغيرات البيئة أو من خلال متغيرات قاعدة البيانات في لوحة الإدارة. راجع [دليل إعداد دوال المنطق](/l/ar/developers/self-host/capabilities/setup#logic-functions-available-drivers) للحصول على التفاصيل.
description: اربط مساعدي الذكاء الاصطناعي بمساحة عمل Twenty الخاصة بك باستخدام بروتوكول سياق النموذج.
---
<Warning>
MCP حاليًا في مرحلة **ألفا** وهو متاح فقط في بعض مساحات العمل. قد لا يكون مفعّلًا لمساحة عملك بعد.
</Warning>
تعرض Twenty خادم [MCP](https://modelcontextprotocol.io/) بحيث تتمكّن مساعدات الذكاء الاصطناعي — Claude Desktop وClaude Code وCursor وChatGPT وغيرها — من قراءة وكتابة بيانات نظام إدارة علاقات العملاء (CRM) لديك باستخدام اللغة الطبيعية.
استخدم **عنوان URL لمساحة العمل** (عنوان URL الذي تستخدمه للوصول إلى Twenty) كنقطة نهاية MCP. على Twenty Cloud، قد يكون عنوان URL لمساحة العمل هو `https://{mycompany}.twenty.com` أو نطاق مخصص. الخادم متاح على:
باستخدام OAuth، يفتح عميل MCP لديك نافذة متصفح لتسجيل الدخول. لا يتم تخزين أي أسرار في ملفات الإعداد، ويتم تحديث الرموز المميِّزة تلقائيًا.
<Note>
يتطلب OAuth عميل MCP يدعم [مواصفة تفويض MCP](https://modelcontextprotocol.io/specification/2025-03-26/basic/authorization). تدعمه Claude Desktop وClaude Code وCursor وChatGPT.
</Note>
أضِف ما يلي إلى تهيئة عميل MCP لديك، واستبدِل `{your-workspace-url}` بمضيف مساحة العمل لديك (على سبيل المثال: `mycompany.twenty.com`):
```json
{
"mcpServers": {
"twenty": {
"type": "streamable-http",
"url": "https://{your-workspace-url}/mcp"
}
}
}
```
هذا كل شيء — لا حاجة إلى مفتاح API. عند اتصال العميل للمرة الأولى، سيفعل ما يلي:
1. اكتشاف بيانات التعريف الخاصة بـ OAuth لدى Twenty عبر `/.well-known/oauth-protected-resource` و`/.well-known/oauth-authorization-server`
2. تسجيل نفسه كعميل OAuth عبر التسجيل الديناميكي للعميل (RFC 7591)
3. فتح متصفحك لتفويض الوصول
4. استلام الرموز المميِّزة والاتصال بخادم MCP
تعيد الاتصالات اللاحقة استخدام الرموز المميِّزة المخزنة وتحدِّثها تلقائيًا.
### الخيار ب — مفتاح API
إذا كان عميل MCP لديك لا يدعم OAuth، أو كنت تفضّل بيانات اعتماد ثابتة، فمرِّر مفتاح API في ترويسة `Authorization`:
```json
{
"mcpServers": {
"twenty": {
"type": "streamable-http",
"url": "https://{your-workspace-url}/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}
```
<Warning>
يمنح مفتاح API الخاص بك حق الوصول إلى بيانات مساحة العمل. أبعِده عن أنظمة التحكم في الإصدارات وملفات dotfiles المشتركة.
</Warning>
لإنشاء مفتاح API، انتقل إلى **Settings > APIs & Webhooks > + Create key**. راجع [واجهات برمجة التطبيقات](/l/ar/developers/extend/api#create-an-api-key) للتفاصيل.
## البدء السريع
### 1. انسخ الإعداد
انتقل إلى **Settings > AI > More > MCP Server** في Twenty. اختر طريقة المصادقة (OAuth أو مفتاح API)، وانسخ مقطع JSON (سيستخدم بالفعل عنوان URL لمساحة العمل لديك)، ثم الصقه في ملف إعدادات عميل MCP لديك.
| **Claude Code** | `~/.claude.json` (المستخدم) أو `.mcp.json` (المشروع) |
| **Cursor** | `.cursor/mcp.json` ضمن مشروعك، أو `~/.cursor/mcp.json` عالميًا |
| **ChatGPT** | فعِّل وضع المطوّر في **Settings > Apps & Connectors > Advanced settings**، ثم استخدم **Create** في **Settings > Apps & Connectors** لإضافة خادم MCP |
### 2. الاتصال
أعِد تشغيل عميل MCP لديك (أو أعد تحميل ملف الإعداد). إذا كنت تستخدم OAuth فسيتم توجيهك إلى Twenty لتفويض الوصول. إذا كنت تستخدم مفتاح API فسيكون الاتصال فوريًا.
### 3. ابدأ باستخدامه
اطلب من مساعد الذكاء الاصطناعي التفاعل مع نظام إدارة علاقات العملاء (CRM) لديك:
* *"أرني أحدث 5 شركات تم إنشاؤها"*
* *"أنشئ شخصًا جديدًا باسم Jane Doe في Acme Corp"*
* *"اعثر على جميع الفرص المفتوحة التي تزيد قيمتها عن 10 آلاف دولار"*
## الأدوات المتاحة
بعد الاتصال، يوفّر خادم MCP أدوات تعكس واجهة برمجة تطبيقات Twenty (API). سير العمل الموصى به هو:
1. **`get_tool_catalog`** — اكتشف جميع الأدوات المتاحة
2. **`learn_tools`** — احصل على مخطط الإدخال لأدوات محددة
3. **`execute_tool`** — شغّل أداة
لا تحتاج إلى تذكّر أسماء الأدوات. اسأل مساعد الذكاء الاصطناعي عمّا يمكنه فعله وسيستدعي `get_tool_catalog` تلقائيًا.
## الصلاحيات
ترث اتصالات MCP أذونات المستخدم المُصادَق عليه (OAuth) أو الدور المُعيَّن لمفتاح API. لتقييد ما يمكن لخادم MCP القيام به:
* **OAuth**: ينطبق دور المستخدم في مساحة العمل.
* **API Key**: عيِّن دورًا لمفتاح API ضمن **Settings > Roles**. راجع [الأذونات](/l/ar/user-guide/permissions-access/capabilities/permissions).
## التكوين ذاتي الاستضافة
في حالات الاستضافة الذاتية، استبدِل `{your-workspace-url}` بعنوان URL الخاص بالخادم لديك. تأكّد من أن قيمة `SERVER_URL` في بيئتك تطابق عنوان URL العام لمثيل Twenty لديك — إذ يُستخدَم ذلك لإنشاء بيانات تعريف اكتشاف OAuth.
```bash
SERVER_URL=https://twenty.yourcompany.com
```
تُشتق نقطة نهاية MCP ونقاط نهاية OAuth وبيانات تعريف الاكتشاف جميعها من هذه القيمة.
## استكشاف الأخطاء وإصلاحها
**أخطاء "Unauthorized" أو 401**
* OAuth: أعد التفويض عبر مسح الرموز المميِّزة المخزنة في عميل MCP لديك ثم أعد الاتصال.
* API Key: تحقّق من أن المفتاح صالح ولم تنتهِ صلاحيته. أعِد توليده إذا لزم الأمر.
**عملية OAuth لا تفتح متصفحًا**
* تأكّد من أن عميل MCP لديك يدعم تفويض MCP. ارجع إلى طريقة مفتاح API إذا لم يكن كذلك.
**انتهاء مهلة الاتصال**
* تحقّق من إمكانية الوصول إلى عنوان URL لنقطة نهاية MCP من جهازك. بالنسبة لحالات الاستضافة الذاتية، تحقّق من أن الخادم يعمل وأن `SERVER_URL` مُعيَّن بشكل صحيح.
title: الرد التلقائي على رسائل البريد الإلكتروني الواردة
description: أنشئ سير عمل يستخدم الذكاء الاصطناعي لفرز رسائل البريد الإلكتروني الواردة وإرسال ردود ضمن سلسلة المحادثة تلقائيًا.
---
استجب لرسائل البريد الإلكتروني الواردة خلال ثوانٍ — وليس ساعات. يستخدم سير العمل هذا وكيلاً للذكاء الاصطناعي لتصفية الضوضاء (النشرات الإخبارية، الرسائل غير المرغوب فيها، الردود التلقائية) وصياغة رد مخصص للرسائل الحقيقية، ثم يرسله كرد مُسلسل داخل المحادثة الأصلية.
## كيف يعمل تسلسل رسائل البريد الإلكتروني
تحمل كل رسالة بريد إلكتروني ترويسة `Message-ID` مخفية — بصمة فريدة يعيّنها خادم بريد المرسل. عند الرد على بريد إلكتروني، يقوم عميل البريد لديك بتعيين ترويسة `In-Reply-To` التي تُشير إلى تلك البصمة. بهذه الطريقة يقوم Gmail وOutlook وكل عميل آخر بتجميع الرسائل في سلاسل المحادثات.
في Twenty، تُخزَّن تلك البصمة كـ `headerMessageId` على كائن Message. يلتقطه سير العمل لديك ويمرّره إلى حقل In-Reply-To لإجراء Send Email.
## بناء سير العمل
### الخطوة 1: إنشاء سير عمل جديد
توجّه إلى **Settings -> Workflows** وانقر **+ New Workflow**.
### الخطوة 2: التشغيل عند الرسائل الواردة
اختر **When a Record is Created** ثم **Messages**.
في كل مرة يصل فيها بريد إلكتروني إلى Twenty، يُشغَّل هذا.
**In-Reply-To** يتوقّع `message.headerMessageId` من المشغّل — إنها البصمة الفريدة للبريد الإلكتروني، وليست عنوان المستلم. إذا تركته فارغًا، فسيُرسَل البريد الإلكتروني على أي حال، ولكن كرسالة مستقلة.
</Tip>
<Warning>
يستخدم Gmail سطر الموضوع لتجميع الرسائل في سلاسل المحادثات. يجب أن يبدأ الموضوع بـ `Re:` (بما في ذلك النقطتان والمسافة) ليعرض Gmail الرد داخل سلسلة المحادثة الأصلية. بدون ذلك، سيظهر الرد كمحادثة منفصلة — حتى إذا تم تعيين ترويسة In-Reply-To بشكل صحيح.
</Warning>
### الخطوة 7: الاختبار والتفعيل
اضغط **Test**، ثم تحقّق من عميل البريد لديك. يجب أن يظهر الرد متداخلًا تحت الرسالة الأصلية.
فعِّل عندما تكون راضيًا عنه.
## أفكار للبناء عليها
* **الرد على كبار الشخصيات فقط** — أضِف فرعًا يتحقق من مجال المُرسِل أو مما إذا كان موجودًا كجهة اتصال في Twenty
* **التوجيه حسب النية** — استخدم موجهات AI Agent منفصلة للتعامل مع استفسارات المبيعات بشكل مختلف عن طلبات الدعم
* **الإثراء قبل الرد** — أضِف خطوة Search Records لجلب شركة المُرسِل أو سجل الصفقات إلى الموجه الخاص بالذكاء الاصطناعي للحصول على ردود أكثر تخصيصًا
@@ -21,19 +21,51 @@ Mit Apps können Sie Twenty-Anpassungen **als Code** erstellen und verwalten. An
## Voraussetzungen
* Node.js 24+ und Yarn 4
* Ein Twenty-Workspace und ein API-Schlüssel (unter https://app.twenty.com/settings/api-webhooks erstellen)
* Docker (für den lokalen Twenty-Dev-Server)
## Erste Schritte
Erstellen Sie mit dem offiziellen Scaffolder eine neue App, authentifizieren Sie sich und beginnen Sie mit der Entwicklung:
Erstelle eine neue App mit dem offiziellen Scaffolder. Der Scaffolder kann für dich automatisch eine lokale Twenty-Instanz starten:
```bash filename="Terminal"
# Eine neue App erstellen (enthält standardmäßig alle Beispiele)
# Eine neue App erstellen — die CLI bietet an, einen lokalen Twenty-Server zu starten
npx create-twenty-app@latest my-twenty-app
cd my-twenty-app
# Dev-Modus starten: synchronisiert lokale Änderungen automatisch mit deinem Arbeitsbereich
yarn twenty app:dev
yarn twenty dev
```
### Lokale Serververwaltung
Das SDK enthält Befehle zur Verwaltung eines lokalen Twenty-Dev-Servers (All-in-One-Docker-Image mit PostgreSQL, Redis, Server und Worker):
```bash filename="Terminal"
# Den lokalen Server starten (lädt das Image bei Bedarf herunter)
yarn twenty server start
# Serverstatus prüfen
yarn twenty server status
# Serverprotokolle streamen
yarn twenty server logs
# Server stoppen
yarn twenty server stop
# Alle Daten zurücksetzen und neu starten
yarn twenty server reset
```
Der lokale Server ist bereits mit einem Arbeitsbereich und einem Benutzer (`tim@apple.dev` / `tim@apple.dev`) vorbefüllt, sodass Sie ohne manuelle Einrichtung sofort mit der Entwicklung beginnen können.
### Authentifizierung
Verbinden Sie Ihre App mithilfe von OAuth mit dem lokalen Server:
```bash filename="Terminal"
# Authenticate via OAuth (opens browser)
yarn twenty remote add --local
```
Das Scaffolding-Tool unterstützt zwei Modi, um zu steuern, welche Beispieldateien enthalten sind:
Führen Sie den folgenden Befehl im Datenbankcontainer aus, um Zugriff auf das Admin-Panel zu erhalten: `UPDATE core."user" SET "canAccessFullAdminPanel" = TRUE WHERE email = 'you@yourdomain.com';`
#### Beim Ausführen eines Workflows schlägt die Workflow-Ausführung fehl mit "Logic function execution is disabled. Set LOGIC_FUNCTION_TYPE to LOCAL or LAMBDA to enable."
In der Produktion sind Logikfunktionen standardmäßig deaktiviert. Setzen Sie die Umgebungsvariable `LOGIC_FUNCTION_TYPE` auf `LOCAL` oder `LAMBDA`, um sie zu aktivieren. Dies kann über Umgebungsvariablen oder über die Datenbankvariablen des Admin-Panels konfiguriert werden. Details finden Sie im [Leitfaden zur Einrichtung von Logikfunktionen](/l/de/developers/self-host/capabilities/setup#logic-functions-available-drivers).
description: Verbinden Sie KI-Assistenten mit Ihrem Twenty-Workspace über das Model Context Protocol.
---
<Warning>
MCP befindet sich derzeit in **alpha** und ist nur in einigen Workspaces verfügbar. Möglicherweise ist es für Ihren Workspace noch nicht aktiviert.
</Warning>
Twenty stellt einen [MCP](https://modelcontextprotocol.io/)-Server bereit, damit KI-Assistenten — Claude Desktop, Claude Code, Cursor, ChatGPT und andere — Ihre CRM-Daten in natürlicher Sprache lesen und schreiben können.
Verwenden Sie Ihre **Workspace-URL** (die URL, mit der Sie auf Twenty zugreifen) als MCP-Endpunkt. In Twenty Cloud kann Ihre Workspace-URL `https://{mycompany}.twenty.com` oder eine benutzerdefinierte Domain sein. Der Server ist verfügbar unter:
Sie haben zwei Möglichkeiten, Ihren MCP-Client zu authentifizieren: **OAuth** (empfohlen) oder **API-Schlüssel**.
### Option A — OAuth (empfohlen)
Mit OAuth öffnet Ihr MCP-Client ein Browserfenster, damit Sie sich anmelden können. Es werden keine geheimen Informationen in Konfigurationsdateien gespeichert, und Token werden automatisch erneuert.
<Note>
OAuth erfordert einen MCP-Client, der die [MCP-Autorisierungsspezifikation](https://modelcontextprotocol.io/specification/2025-03-26/basic/authorization) unterstützt. Claude Desktop, Claude Code, Cursor und ChatGPT unterstützen dies.
</Note>
Fügen Sie dies zu Ihrer MCP-Client-Konfiguration hinzu und ersetzen Sie `{your-workspace-url}` durch den Host Ihrer Workspace-URL (z. B. `mycompany.twenty.com`):
```json
{
"mcpServers": {
"twenty": {
"type": "streamable-http",
"url": "https://{your-workspace-url}/mcp"
}
}
}
```
Das ist alles — kein API-Schlüssel erforderlich. Wenn der Client sich zum ersten Mal verbindet, wird er:
1. Die OAuth-Metadaten von Twenty über `/.well-known/oauth-protected-resource` und `/.well-known/oauth-authorization-server` ermitteln
2. Sich über die dynamische Client-Registrierung (RFC 7591) als OAuth-Client registrieren
3. Ihren Browser öffnen, um den Zugriff zu autorisieren
4. Token empfangen und eine Verbindung zum MCP-Server herstellen
Nachfolgende Verbindungen verwenden die gespeicherten Token erneut und erneuern sie automatisch.
### Option B — API-Schlüssel
Wenn Ihr MCP-Client OAuth nicht unterstützt oder Sie statische Anmeldeinformationen bevorzugen, übergeben Sie einen API-Schlüssel im `Authorization`-Header:
```json
{
"mcpServers": {
"twenty": {
"type": "streamable-http",
"url": "https://{your-workspace-url}/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}
```
<Warning>
Ihr API-Schlüssel gewährt Zugriff auf Workspace-Daten. Halten Sie es von der Versionskontrolle und von gemeinsam genutzten Dotfiles fern.
</Warning>
Um einen API-Schlüssel zu erstellen, gehen Sie zu **Settings > APIs & Webhooks > + Create key**. Details finden Sie unter [APIs](/l/de/developers/extend/api#create-an-api-key).
## Schnellstart
### 1. Konfiguration kopieren
Gehen Sie in Twenty zu **Settings > AI > More > MCP Server**. Wählen Sie Ihre Authentifizierungsmethode (OAuth oder API-Schlüssel), kopieren Sie das JSON-Snippet (es verwendet bereits Ihre Workspace-URL) und fügen Sie es in die Konfigurationsdatei Ihres MCP-Clients ein.
| **Claude Code** | `~/.claude.json` (Benutzer) oder `.mcp.json` (Projekt) |
| **Cursor** | `.cursor/mcp.json` in Ihrem Projekt oder `~/.cursor/mcp.json` global |
| **ChatGPT** | Aktivieren Sie den Entwicklermodus in **Settings > Apps & Connectors > Advanced settings** und verwenden Sie dann **Create** in **Settings > Apps & Connectors**, um den MCP-Server hinzuzufügen |
### 2. Verbinden
Starten Sie Ihren MCP-Client neu (oder laden Sie die Konfiguration neu). Bei Verwendung von OAuth werden Sie zu Twenty weitergeleitet, um den Zugriff zu autorisieren. Bei Verwendung eines API-Schlüssels wird die Verbindung sofort hergestellt.
### 3. Jetzt loslegen
Bitten Sie Ihren KI-Assistenten, mit Ihrem CRM zu interagieren:
* *"Zeige mir die 5 zuletzt erstellten Unternehmen"*
* *"Erstelle eine neue Person namens Jane Doe bei Acme Corp"*
* *"Finde alle offenen Verkaufschancen mit einem Wert von mehr als $10k"*
## Verfügbare Tools
Nach der Verbindung stellt der MCP-Server Tools bereit, die die Twenty-API widerspiegeln. Der empfohlene Workflow ist:
1. **`get_tool_catalog`** — alle verfügbaren Tools entdecken
2. **`learn_tools`** — das Eingabeschema für bestimmte Tools abrufen
3. **`execute_tool`** — ein Tool ausführen
Sie müssen sich die Tool-Namen nicht merken. Fragen Sie Ihren KI-Assistenten, was er tun kann, und er ruft `get_tool_catalog` automatisch auf.
## Berechtigungen
MCP-Verbindungen erben die Berechtigungen des authentifizierten Benutzers (OAuth) oder die dem API-Schlüssel zugewiesene Rolle. So beschränken Sie, was der MCP-Server tun darf:
* **OAuth**: Es gilt die Workspace-Rolle des Benutzers.
* **API-Schlüssel**: Weisen Sie dem API-Schlüssel unter **Settings > Roles** eine Rolle zu. Siehe [Berechtigungen](/l/de/user-guide/permissions-access/capabilities/permissions).
## Selbstgehostete Konfiguration
Für selbstgehostete Instanzen ersetzen Sie `{your-workspace-url}` durch die URL Ihres Servers. Stellen Sie sicher, dass `SERVER_URL` in Ihrer Umgebung der öffentlichen URL Ihrer Twenty-Instanz entspricht — dieser Wert wird verwendet, um die OAuth-Discovery-Metadaten zu generieren.
```bash
SERVER_URL=https://twenty.yourcompany.com
```
Der MCP-Endpunkt, die OAuth-Endpunkte und die Discovery-Metadaten leiten sich alle von diesem Wert ab.
## Fehlerbehebung
**"Unauthorized"- oder 401-Fehler**
* OAuth: Autorisieren Sie erneut, indem Sie die gespeicherten Token in Ihrem MCP-Client löschen und die Verbindung wiederherstellen.
* API-Schlüssel: Überprüfen Sie, ob der Schlüssel gültig ist und nicht abgelaufen ist. Generieren Sie ihn bei Bedarf neu.
**Der OAuth-Flow öffnet keinen Browser**
* Stellen Sie sicher, dass Ihr MCP-Client MCP Authorization unterstützt. Wechseln Sie andernfalls zur API-Schlüssel-Methode.
**Verbindungszeitüberschreitung**
* Stellen Sie sicher, dass die MCP-Endpunkt-URL von Ihrem Rechner aus erreichbar ist. Bei selbstgehosteten Instanzen prüfen Sie, ob der Server läuft und `SERVER_URL` korrekt gesetzt ist.
title: Automatische Antwort auf eingehende E-Mails
description: Erstellen Sie einen Workflow, der KI nutzt, um eingehende E-Mails zu sichten und automatisch Antworten im Thread zu senden.
---
Antworten Sie auf eingehende E-Mails in Sekunden — nicht in Stunden. Dieser Workflow verwendet einen KI-Agenten, um Störendes auszufiltern (Newsletter, Spam, automatische Antworten) und eine personalisierte Antwort auf echte Nachrichten zu verfassen und sie dann als Antwort im Thread innerhalb der ursprünglichen Unterhaltung zu senden.
## Wie E-Mail-Threading funktioniert
Jede E-Mail enthält einen verborgenen `Message-ID`-Header — einen eindeutigen Fingerabdruck, der vom Mailserver des Absenders zugewiesen wird. Wenn Sie auf eine E-Mail antworten, setzt Ihr E-Mail-Client einen `In-Reply-To`-Header, der auf diesen Fingerabdruck verweist. So gruppieren Gmail, Outlook und alle anderen Clients Nachrichten in Threads.
In Twenty wird dieser Fingerabdruck als `headerMessageId` am Message-Objekt gespeichert. Ihr Workflow greift ihn auf und übergibt ihn an das Feld In-Reply-To der Aktion Send Email.
## Den Workflow erstellen
### Schritt 1: Neuen Workflow erstellen
Gehen Sie zu **Settings -> Workflows** und klicken Sie auf **+ New Workflow**.
### Schritt 2: Bei eingehenden Nachrichten auslösen
Wählen Sie **When a Record is Created** und dann **Messages**.
Jedes Mal, wenn eine E-Mail in Twenty eingeht, wird dies ausgelöst.
Fügen Sie eine **AI Agent**-Aktion hinzu. Dieser einzelne Schritt erledigt zwei Dinge: Er entscheidet, ob die E-Mail eine Antwort verdient, und falls ja, verfasst er eine.
Verwenden Sie einen Prompt wie:
```
You are an email triage assistant for a sales team. Read the following
Das Feld In-Reply-To sorgt dafür, dass dies eine Antwort und keine neue Unterhaltung ist. Der Empfänger sieht sie als Teil des Threads unter der ursprünglichen E-Mail in Gmail, Outlook oder jedem anderen Client.
**In-Reply-To** erwartet eine `message.headerMessageId` aus dem Trigger — das ist der eindeutige Fingerabdruck der E-Mail, keine Empfängeradresse. Wenn Sie es leer lassen, wird die E-Mail trotzdem gesendet, nur als eigenständige Nachricht.
</Tip>
<Warning>
Gmail verwendet die Betreffzeile, um Nachrichten in Threads zu gruppieren. Der Betreff **muss** mit `Re:` beginnen (einschließlich Doppelpunkt und Leerzeichen), damit Gmail die Antwort im ursprünglichen Thread anzeigt. Ohne dies erscheint die Antwort als separate Unterhaltung — selbst wenn der In-Reply-To-Header korrekt gesetzt ist.
</Warning>
### Schritt 7: Testen und aktivieren
Klicken Sie auf **Test**, und prüfen Sie dann Ihren E-Mail-Client. Die Antwort sollte unter der ursprünglichen Nachricht verschachtelt erscheinen.
Aktivieren Sie ihn, wenn Sie zufrieden sind.
## Ideen zum Ausbau
* **Nur an VIPs antworten** — fügen Sie einen Zweig hinzu, der die Domain des Absenders prüft oder ob er in Twenty als Kontakt existiert
* **Nach Absicht routen** — verwenden Sie separate AI Agent-Prompts, um Vertriebsanfragen anders zu bearbeiten als Supportanfragen
* **Vor der Antwort anreichern** — fügen Sie einen Schritt Search Records hinzu, um das Unternehmen oder die Deal-Historie des Absenders in den AI-Prompt zu übernehmen, für persönlichere Antworten
@@ -21,19 +21,51 @@ Le app ti consentono di creare e gestire le personalizzazioni di Twenty **come c
## Prerequisiti
* Node.js 24+ e Yarn 4
* Uno spazio di lavoro Twenty e una chiave API (creane una su https://app.twenty.com/settings/api-webhooks)
* Docker (per il server di sviluppo locale di Twenty)
## Per iniziare
Crea una nuova app utilizzando lo scaffolder ufficiale, quindi autenticati e inizia a sviluppare:
Crea una nuova app utilizzando lo scaffolder ufficiale. Può avviare automaticamente un'istanza locale di Twenty per te:
```bash filename="Terminal"
# Crea lo scaffold di una nuova app (include tutti gli esempi per impostazione predefinita)
# Crea lo scaffold di una nuova app — la CLI offrirà di avviare un server locale di Twenty
npx create-twenty-app@latest my-twenty-app
cd my-twenty-app
# Avvia la modalità di sviluppo: sincronizza automaticamente le modifiche locali con il tuo workspace
yarn twenty app:dev
yarn twenty dev
```
### Gestione del server locale
L'SDK include comandi per gestire un server di sviluppo locale di Twenty (immagine Docker all-in-one con PostgreSQL, Redis, server e worker):
```bash filename="Terminal"
# Avvia il server locale (scarica l'immagine se necessario)
yarn twenty server start
# Verifica lo stato del server
yarn twenty server status
# Segui i log del server
yarn twenty server logs
# Arresta il server
yarn twenty server stop
# Reimposta tutti i dati e riparti da zero
yarn twenty server reset
```
Il server locale è preconfigurato con uno spazio di lavoro e un utente (`tim@apple.dev` / `tim@apple.dev`), così puoi iniziare a sviluppare immediatamente senza alcuna configurazione manuale.
### Autenticazione
Collega la tua app al server locale tramite OAuth:
```bash filename="Terminal"
# Autenticati tramite OAuth (apre il browser)
yarn twenty remote add --local
```
Lo strumento di scaffolding supporta due modalità per controllare quali file di esempio vengono inclusi:
Esegui `UPDATE core."user" SET "canAccessFullAdminPanel" = TRUE WHERE email = 'you@yourdomain.com';` nel container del database per accedere al pannello di amministrazione.
#### Durante l'esecuzione di un workflow, l'esecuzione del workflow non riesce con "L'esecuzione delle funzioni logiche è disabilitata. Imposta LOGIC_FUNCTION_TYPE su LOCAL o LAMBDA per abilitarle."
In produzione, le funzioni logiche sono disabilitate per impostazione predefinita. Imposta la variabile d'ambiente `LOGIC_FUNCTION_TYPE` su `LOCAL` o `LAMBDA` per abilitarle. Questo può essere configurato tramite variabili d'ambiente o tramite le variabili del database del pannello di amministrazione. Consulta la [guida alla configurazione delle funzioni logiche](/l/it/developers/self-host/capabilities/setup#logic-functions-available-drivers) per dettagli.
description: Collega gli assistenti AI al tuo spazio di lavoro di Twenty utilizzando il Model Context Protocol.
---
<Warning>
MCP è attualmente in **alpha** ed è disponibile solo su alcuni spazi di lavoro. Potrebbe non essere ancora abilitato per il tuo spazio di lavoro.
</Warning>
Twenty espone un server [MCP](https://modelcontextprotocol.io/) affinché gli assistenti AI — Claude Desktop, Claude Code, Cursor, ChatGPT e altri — possano leggere e scrivere i dati del tuo CRM in linguaggio naturale.
Usa l'**URL dello spazio di lavoro** (l'URL che usi per accedere a Twenty) come endpoint MCP. Su Twenty Cloud, l'URL del tuo spazio di lavoro potrebbe essere `https://{mycompany}.twenty.com` oppure un dominio personalizzato. Il server è disponibile all'indirizzo:
Hai due modi per autenticare il tuo client MCP: **OAuth** (consigliato) o **API Key**.
### Opzione A — OAuth (Consigliato)
Con OAuth, il tuo client MCP apre una finestra del browser per effettuare l'accesso. Nessun segreto viene archiviato nei file di configurazione e i token si rinnovano automaticamente.
<Note>
OAuth richiede un client MCP che supporti la [specifica MCP Authorization](https://modelcontextprotocol.io/specification/2025-03-26/basic/authorization). Claude Desktop, Claude Code, Cursor e ChatGPT lo supportano.
</Note>
Aggiungi questo alla configurazione del tuo client MCP, sostituendo `{your-workspace-url}` con l'host del tuo spazio di lavoro (ad es. `mycompany.twenty.com`):
```json
{
"mcpServers": {
"twenty": {
"type": "streamable-http",
"url": "https://{your-workspace-url}/mcp"
}
}
}
```
È tutto — non è necessaria alcuna chiave API. Quando il client si connette per la prima volta, eseguirà:
1. Scoprire i metadati OAuth di Twenty tramite `/.well-known/oauth-protected-resource` e `/.well-known/oauth-authorization-server`
2. Registrarsi come client OAuth tramite registrazione dinamica del client (RFC 7591)
3. Aprire il browser per autorizzare l'accesso
4. Ricevere i token e connettersi al server MCP
Le connessioni successive riutilizzano i token memorizzati e li rinnovano automaticamente.
### Opzione B — Chiave API
Se il tuo client MCP non supporta OAuth, o preferisci credenziali statiche, passa una chiave API nell'header `Authorization`:
```json
{
"mcpServers": {
"twenty": {
"type": "streamable-http",
"url": "https://{your-workspace-url}/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}
```
<Warning>
La tua chiave API concede l'accesso ai dati dello spazio di lavoro. Mantienila fuori dal controllo versione e dai dotfile condivisi.
</Warning>
Per creare una chiave API, vai su **Impostazioni > API e Webhook > + Crea chiave**. Vedi [API](/l/it/developers/extend/api#create-an-api-key) per i dettagli.
## Avvio rapido
### 1. Copia la configurazione
Vai su **Impostazioni > AI > Altro > MCP Server** in Twenty. Scegli il metodo di autenticazione (OAuth o Chiave API), copia lo snippet JSON (userà già l'URL del tuo spazio di lavoro) e incollalo nel file di configurazione del tuo client MCP.
| **Claude Code** | `~/.claude.json` (utente) o `.mcp.json` (progetto) |
| **Cursor** | `.cursor/mcp.json` nel tuo progetto, oppure `~/.cursor/mcp.json` globalmente |
| **ChatGPT** | Attiva la Modalità sviluppatore in **Impostazioni > App e Connettori > Impostazioni avanzate**, quindi usa **Crea** in **Impostazioni > App e Connettori** per aggiungere il server MCP |
### 2. Connetti
Riavvia il tuo client MCP (o ricarica la configurazione). Se usi OAuth verrai reindirizzato a Twenty per autorizzare l'accesso. Se usi una chiave API la connessione è immediata.
### 3. Inizia a usarlo
Chiedi al tuo assistente AI di interagire con il tuo CRM:
* *"Mostrami le 5 aziende create più di recente"*
* *"Crea una nuova persona di nome Jane Doe presso Acme Corp"*
* *"Trova tutte le opportunità aperte di valore superiore a $10k"*
## Strumenti disponibili
Una volta connesso, il server MCP espone strumenti che rispecchiano l'API di Twenty. Il flusso di lavoro consigliato è:
1. **`get_tool_catalog`** — scoprire tutti gli strumenti disponibili
2. **`learn_tools`** — ottenere lo schema di input per strumenti specifici
3. **`execute_tool`** — eseguire uno strumento
Non è necessario ricordare i nomi degli strumenti. Chiedi al tuo assistente AI cosa può fare e chiamerà `get_tool_catalog` automaticamente.
## Permessi
Le connessioni MCP ereditano le autorizzazioni dell'utente autenticato (OAuth) o il ruolo assegnato alla chiave API. Per limitare ciò che il server MCP può fare:
* **OAuth**: si applica il ruolo dell'utente nello spazio di lavoro.
* **Chiave API**: assegna un ruolo alla chiave API in **Impostazioni > Ruoli**. Vedi [Autorizzazioni](/l/it/user-guide/permissions-access/capabilities/permissions).
## Configurazione self-hosted
Per le istanze self-hosted, sostituisci `{your-workspace-url}` con l'URL del tuo server. Assicurati che `SERVER_URL` nel tuo ambiente corrisponda all'URL pubblico della tua istanza Twenty — viene utilizzato per generare i metadati di discovery OAuth.
```bash
SERVER_URL=https://twenty.yourcompany.com
```
L'endpoint MCP, gli endpoint OAuth e i metadati di discovery derivano tutti da questo valore.
## Risoluzione dei problemi
**Errori "Unauthorized" o 401**
* OAuth: esegui nuovamente l'autorizzazione eliminando i token memorizzati nel tuo client MCP e riconnettiti.
* Chiave API: verifica che la chiave sia valida e non sia scaduta. Rigenerala se necessario.
**Il flusso OAuth non apre il browser**
* Assicurati che il tuo client MCP supporti MCP Authorization. In caso contrario, ricorri al metodo con Chiave API.
**Timeout di connessione**
* Verifica che l'URL dell'endpoint MCP sia raggiungibile dalla tua macchina. Per le istanze self-hosted, controlla che il server sia in esecuzione e che `SERVER_URL` sia impostato correttamente.
description: Crea un flusso di lavoro che utilizza l'IA per smistare le email in arrivo e inviare automaticamente risposte in thread.
---
Rispondi alle email in arrivo in pochi secondi — non ore. Questo flusso di lavoro utilizza un agente IA per filtrare il rumore (newsletter, spam, risposte automatiche) e redigere una risposta personalizzata ai messaggi reali, quindi la invia come risposta in thread all'interno della conversazione originale.
## Come funziona il threading delle email
Ogni email include un'intestazione nascosta `Message-ID` — un'impronta univoca assegnata dal server di posta del mittente. Quando rispondi a un'email, il tuo client di posta imposta un'intestazione `In-Reply-To` che fa riferimento a quell'impronta. È così che Gmail, Outlook e tutti gli altri client raggruppano i messaggi nei thread.
In Twenty, quell'impronta è memorizzata come `headerMessageId` sull'oggetto Message. Il tuo flusso di lavoro la recupera e la passa al campo In-Reply-To dell'azione Send Email.
## Creazione del flusso di lavoro
### Passaggio 1: Crea un nuovo flusso di lavoro
Vai a **Impostazioni -> Flussi di lavoro** e fai clic su **+ New Workflow**.
### Passaggio 2: Imposta il trigger sui messaggi in arrivo
Scegli **When a Record is Created** e seleziona **Messages**.
Ogni volta che un'email arriva in Twenty, questo si attiva.
Il campo In-Reply-To è ciò che rende questo un messaggio di risposta invece di una nuova conversazione. Il destinatario lo vede nel thread sotto l'email originale in Gmail, Outlook o qualsiasi altro client.
**In-Reply-To** si aspetta un `message.headerMessageId` dal trigger — è l'impronta univoca dell'email, non un indirizzo del destinatario. Se lo lasci vuoto, l'email viene comunque inviata, solo come messaggio autonomo.
</Tip>
<Warning>
Gmail usa l'oggetto per raggruppare i messaggi nei thread. L'oggetto **deve** iniziare con `Re:` (inclusi i due punti e lo spazio) affinché Gmail visualizzi la risposta all'interno del thread originale. Senza di esso, la risposta apparirà come una conversazione separata — anche se l'intestazione In-Reply-To è impostata correttamente.
</Warning>
### Passaggio 7: Testa e attiva
Fai clic su **Test**, quindi controlla il tuo client di posta. La risposta dovrebbe comparire annidata sotto il messaggio originale.
Attivalo quando sei soddisfatto.
## Idee per sviluppare ulteriormente
* **Rispondi solo ai VIP** — aggiungi un ramo che controlli il dominio del mittente o se esiste come Contatto in Twenty
* **Instrada in base all'intento** — usa prompt distinti di AI Agent per gestire le richieste commerciali in modo diverso rispetto a quelle di supporto
* **Arricchisci prima di rispondere** — aggiungi un passaggio Search Records per inserire nel prompt dell'AI l'azienda del mittente o la cronologia delle trattative per risposte più personalizzate
@@ -21,19 +21,51 @@ Os aplicativos permitem criar e gerenciar personalizações do Twenty **como có
## Pré-requisitos
* Node.js 24+ e Yarn 4
* Um espaço de trabalho do Twenty e uma chave de API (crie uma em https://app.twenty.com/settings/api-webhooks)
* Docker (para o servidor de desenvolvimento local do Twenty)
## Primeiros passos
Crie um novo aplicativo usando o gerador oficial, depois autentique-se e comece a desenvolver:
Crie um novo app usando o gerador oficial de estrutura. Ele pode iniciar automaticamente uma instância local do Twenty para você:
```bash filename="Terminal"
# Criar a estrutura de um novo app (inclui todos os exemplos por padrão)
# Criar a estrutura de um novo app — a CLI oferecerá iniciar um servidor local do Twenty
npx create-twenty-app@latest my-twenty-app
cd my-twenty-app
# Iniciar modo de desenvolvimento: sincroniza automaticamente as alterações locais com seu workspace
yarn twenty app:dev
yarn twenty dev
```
### Gerenciamento do Servidor Local
O SDK inclui comandos para gerenciar um servidor de desenvolvimento local do Twenty (imagem Docker all-in-one com PostgreSQL, Redis, servidor e worker):
```bash filename="Terminal"
# Iniciar o servidor local (faz pull da imagem se necessário)
yarn twenty server start
# Verificar o status do servidor
yarn twenty server status
# Transmitir os logs do servidor
yarn twenty server logs
# Parar o servidor
yarn twenty server stop
# Redefinir todos os dados e começar do zero
yarn twenty server reset
```
O servidor local já vem pré-configurado com um espaço de trabalho e um usuário (`tim@apple.dev` / `tim@apple.dev`), para que você possa começar a desenvolver imediatamente, sem qualquer configuração manual.
### Autenticação
Conecte seu aplicativo ao servidor local usando OAuth:
```bash filename="Terminal"
# Authenticate via OAuth (opens browser)
yarn twenty remote add --local
```
O gerador de estrutura oferece suporte a dois modos para controlar quais arquivos de exemplo são incluídos:
Execute `UPDATE core."user" SET "canAccessFullAdminPanel" = TRUE WHERE email = 'você@seudominio.com';` no contêiner de banco de dados para obter acesso ao painel administrativo.
#### Ao executar um fluxo de trabalho, a execução do fluxo de trabalho falha com "A execução da função de lógica está desativada. Defina LOGIC_FUNCTION_TYPE como LOCAL ou LAMBDA para ativar."
Em produção, as funções de lógica estão desativadas por padrão. Defina a variável de ambiente `LOGIC_FUNCTION_TYPE` como `LOCAL` ou `LAMBDA` para ativá-las. Isso pode ser configurado por meio de variáveis de ambiente ou pelas variáveis de banco de dados do painel de administração. Veja o [guia de configuração de Funções de Lógica](/l/pt/developers/self-host/capabilities/setup#logic-functions-available-drivers) para obter detalhes.
description: Conecte assistentes de IA ao seu espaço de trabalho do Twenty usando o Model Context Protocol.
---
<Warning>
O MCP está atualmente em **alfa** e está disponível apenas em alguns espaços de trabalho. O MCP pode ainda não estar ativado no seu espaço de trabalho.
</Warning>
O Twenty expõe um servidor [MCP](https://modelcontextprotocol.io/) para que assistentes de IA — Claude Desktop, Claude Code, Cursor, ChatGPT e outros — possam ler e escrever seus dados de CRM por meio de linguagem natural.
Use o **URL do espaço de trabalho** (o URL que você usa para acessar o Twenty) como o endpoint do MCP. Na Twenty Cloud, o URL do seu espaço de trabalho pode ser `https://{mycompany}.twenty.com` ou um domínio personalizado. O servidor está disponível em:
Você tem duas maneiras de autenticar seu cliente MCP: **OAuth** (recomendado) ou **Chave de API**.
### Opção A — OAuth (recomendado)
Com o OAuth, seu cliente MCP abre uma janela do navegador para você fazer login. Nenhum segredo é armazenado em arquivos de configuração, e os tokens são atualizados automaticamente.
<Note>
O OAuth requer um cliente MCP que ofereça suporte à [especificação de Autorização MCP](https://modelcontextprotocol.io/specification/2025-03-26/basic/authorization). Claude Desktop, Claude Code, Cursor e ChatGPT oferecem suporte.
</Note>
Adicione isto à configuração do seu cliente MCP, substituindo `{your-workspace-url}` pelo host do seu espaço de trabalho (por exemplo, `mycompany.twenty.com`):
```json
{
"mcpServers": {
"twenty": {
"type": "streamable-http",
"url": "https://{your-workspace-url}/mcp"
}
}
}
```
É isso — nenhuma chave de API é necessária. Quando o cliente se conectar pela primeira vez, ele irá:
1. Descobrir os metadados de OAuth do Twenty por meio de `/.well-known/oauth-protected-resource` e `/.well-known/oauth-authorization-server`
2. Registrar-se como um cliente OAuth por meio de registro dinâmico de cliente (RFC 7591)
3. Abrir seu navegador para autorizar o acesso
4. Receber tokens e conectar-se ao servidor MCP
Conexões subsequentes reutilizam os tokens armazenados e os atualizam automaticamente.
### Opção B — Chave de API
Se o seu cliente MCP não oferecer suporte a OAuth, ou se você preferir credenciais estáticas, passe uma chave de API no cabeçalho `Authorization`:
```json
{
"mcpServers": {
"twenty": {
"type": "streamable-http",
"url": "https://{your-workspace-url}/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}
```
<Warning>
Sua chave de API concede acesso aos dados do espaço de trabalho. Mantenha-a fora do controle de versão e de dotfiles compartilhados.
</Warning>
Para criar uma chave de API, vá em **Settings > APIs & Webhooks > + Create key**. Veja [APIs](/l/pt/developers/extend/api#create-an-api-key) para detalhes.
## Início rápido
### 1. Copie a configuração
Vá até **Settings > AI > More > MCP Server** no Twenty. Escolha seu método de autenticação (OAuth ou Chave de API), copie o trecho de JSON (ele já usará o URL do seu espaço de trabalho) e cole-o no arquivo de configuração do seu cliente MCP.
| **Claude Code** | `~/.claude.json` (usuário) ou `.mcp.json` (projeto) |
| **Cursor** | `.cursor/mcp.json` no seu projeto, ou `~/.cursor/mcp.json` globalmente |
| **ChatGPT** | Ative o Modo Desenvolvedor em **Settings > Apps & Connectors > Advanced settings** e, em seguida, use **Create** em **Settings > Apps & Connectors** para adicionar o servidor MCP |
### 2. Conectar
Reinicie seu cliente MCP (ou recarregue a configuração). Se estiver usando OAuth, você será redirecionado ao Twenty para autorizar o acesso. Se estiver usando uma chave de API, a conexão é imediata.
### 3. Comece a usá-lo
Peça ao seu assistente de IA para interagir com seu CRM:
* *"Mostre-me as 5 empresas criadas mais recentemente"*
* *"Crie uma nova pessoa chamada Jane Doe na Acme Corp"*
* *"Encontre todas as oportunidades em aberto com valor superior a $10k"*
## Ferramentas disponíveis
Depois de conectado, o servidor MCP expõe ferramentas que refletem a API do Twenty. O fluxo de trabalho recomendado é:
1. **`get_tool_catalog`** — descobrir todas as ferramentas disponíveis
2. **`learn_tools`** — obter o esquema de entrada para ferramentas específicas
3. **`execute_tool`** — executar uma ferramenta
Você não precisa se lembrar dos nomes das ferramentas. Pergunte ao seu assistente de IA o que ele pode fazer e ele chamará `get_tool_catalog` automaticamente.
## Permissões
As conexões MCP herdam as permissões do usuário autenticado (OAuth) ou da função atribuída à chave de API. Para restringir o que o servidor MCP pode fazer:
* **OAuth**: Aplica-se à função do espaço de trabalho do usuário.
* **Chave de API**: Atribua uma função à chave de API em **Settings > Roles**. Veja [Permissões](/l/pt/user-guide/permissions-access/capabilities/permissions).
## Configuração auto-hospedada
Para instâncias auto-hospedadas, substitua `{your-workspace-url}` pelo URL do seu servidor. Certifique-se de que `SERVER_URL` no seu ambiente corresponda ao URL público da sua instância do Twenty — isso é usado para gerar os metadados de descoberta do OAuth.
```bash
SERVER_URL=https://twenty.yourcompany.com
```
O endpoint do MCP, os endpoints de OAuth e os metadados de descoberta derivam todos desse valor.
## Resolução de Problemas
**Erros "Unauthorized" ou 401**
* OAuth: autorize novamente limpando os tokens armazenados no seu cliente MCP e reconectando.
* Chave de API: verifique se a chave é válida e não expirou. Gere-a novamente, se necessário.
**O fluxo do OAuth não abre um navegador**
* Garanta que seu cliente MCP ofereça suporte à Autorização MCP. Se não oferecer, utilize o método de Chave de API.
**Tempo limite de conexão**
* Confirme que o URL do endpoint MCP é acessível a partir da sua máquina. Para instâncias auto-hospedadas, verifique se o servidor está em execução e se `SERVER_URL` está definido corretamente.
description: Crie um fluxo de trabalho que use IA para fazer a triagem dos e-mails de entrada e enviar respostas encadeadas automaticamente.
---
Responda a e-mails de entrada em segundos — não em horas. Este fluxo de trabalho usa um Agente de IA para filtrar o ruído (boletins informativos, spam, respostas automáticas) e redigir uma resposta personalizada para mensagens reais e, em seguida, enviá-la como uma resposta encadeada dentro da conversa original.
## Como funciona o encadeamento de e-mails
Cada e-mail contém um cabeçalho oculto `Message-ID` — uma impressão digital única atribuída pelo servidor de e-mail do remetente. Quando você responde a um e-mail, seu cliente de e-mail define um cabeçalho `In-Reply-To` que faz referência a essa impressão digital. É assim que o Gmail, o Outlook e todos os outros clientes agrupam as mensagens em conversas.
No Twenty, essa impressão digital é armazenada como `headerMessageId` no objeto Message. Seu fluxo de trabalho obtém isso e o passa para o campo In-Reply-To da ação Enviar e-mail.
## Criando o fluxo de trabalho
### Etapa 1: Criar um novo fluxo de trabalho
Vá para **Configurações -> Fluxos de trabalho** e clique em **+ Novo fluxo de trabalho**.
### Etapa 2: Acionar com mensagens recebidas
Escolha **When a Record is Created** e selecione **Messages**.
Sempre que um e-mail chega ao Twenty, isto é acionado.
O campo In-Reply-To é o que faz disso uma resposta em vez de uma nova conversa. O destinatário a vê encadeada sob o e-mail original no Gmail, Outlook ou qualquer outro cliente.
**In-Reply-To** espera um `message.headerMessageId` do acionador — é a impressão digital única do e-mail, não um endereço do destinatário. Se você deixá-lo vazio, o e-mail ainda será enviado, apenas como uma mensagem independente.
</Tip>
<Warning>
O Gmail usa a linha de assunto para agrupar mensagens em conversas. O assunto **deve** começar com `Re:` (incluindo os dois-pontos e o espaço) para que o Gmail exiba a resposta dentro da conversa original. Sem isso, a resposta aparecerá como uma conversa separada — mesmo que o cabeçalho In-Reply-To esteja definido corretamente.
</Warning>
### Etapa 7: Testar e ativar
Clique em **Test** e, em seguida, verifique seu cliente de e-mail. A resposta deve aparecer aninhada sob a mensagem original.
Ative quando estiver satisfeito com isso.
## Ideias para desenvolver
* **Responder apenas a VIPs** — adicione um ramo que verifique o domínio do remetente ou se ele existe como um Contato no Twenty
* **Roteie por intenção** — use prompts separados do Agente de IA para lidar com consultas de vendas de forma diferente de solicitações de suporte
* **Enriquecer antes de responder** — adicione uma etapa Pesquisar Registros para trazer a empresa do remetente ou o histórico de negócios para o prompt de IA, obtendo respostas mais personalizadas
@@ -21,19 +21,51 @@ Aplicațiile vă permit să construiți și să gestionați personalizările Twe
## Cerințe
* Node.js 24+ și Yarn 4
* Un spațiu de lucru Twenty și o cheie API (creați una la https://app.twenty.com/settings/api-webhooks)
* Docker (pentru serverul local de dezvoltare Twenty)
## Începeți
Creați o aplicație nouă folosind generatorul oficial, apoi autentificați-vă și începeți să dezvoltați:
Creează o aplicație nouă folosind generatorul oficial. Poate porni automat o instanță Twenty locală pentru tine:
```bash filename="Terminal"
# Creează scheletul unei aplicații noi (include toate exemplele în mod implicit)
# Creează scheletul unei aplicații noi — CLI-ul îți va oferi opțiunea de a porni un server Twenty local
npx create-twenty-app@latest my-twenty-app
cd my-twenty-app
# Pornește modul de dezvoltare: sincronizează automat modificările locale cu spațiul tău de lucru
yarn twenty app:dev
yarn twenty dev
```
### Gestionarea serverului local
SDK-ul include comenzi pentru a gestiona un server local de dezvoltare Twenty (imagine Docker all-in-one cu PostgreSQL, Redis, server și worker):
```bash filename="Terminal"
# Pornește serverul local (descarcă imaginea dacă este necesar)
yarn twenty server start
# Verifică starea serverului
yarn twenty server status
# Afișează în timp real jurnalele serverului
yarn twenty server logs
# Oprește serverul
yarn twenty server stop
# Resetează toate datele și pornește de la zero
yarn twenty server reset
```
Serverul local vine preconfigurat cu un spațiu de lucru și un utilizator (`tim@apple.dev` / `tim@apple.dev`), astfel încât să poți începe să dezvolți imediat, fără nicio configurare manuală.
### Autentificare
Conectează-ți aplicația la serverul local folosind OAuth:
```bash filename="Terminal"
# Autentifică-te prin OAuth (se deschide browserul)
yarn twenty remote add --local
```
Generatorul de schelet acceptă două moduri pentru a controla ce fișiere de exemplu sunt incluse:
Rulați `UPDATE core."user" SET "canAccessFullAdminPanel" = TRUE WHERE email = 'tine@domeniultău.com';` în containerul de baze de date pentru a obține acces la panoul de administrare.
#### Când rulați un flux de lucru, rularea fluxului de lucru eșuează cu "Execuția funcțiilor logice este dezactivată. Setați LOGIC_FUNCTION_TYPE la LOCAL sau LAMBDA pentru a le activa."
În producție, funcțiile logice sunt dezactivate în mod implicit. Setați variabila de mediu `LOGIC_FUNCTION_TYPE` la `LOCAL` sau `LAMBDA` pentru a le activa. Acest lucru poate fi configurat prin variabile de mediu sau prin variabilele bazei de date din panoul de administrare. Consultați [Ghidul de configurare a funcțiilor logice](/l/ro/developers/self-host/capabilities/setup#logic-functions-available-drivers) pentru detalii.
description: Conectați asistenți AI la spațiul dvs. de lucru Twenty folosind Model Context Protocol.
---
<Warning>
MCP este în prezent în **alpha** și este disponibil doar în unele spații de lucru. Este posibil să nu fie activat încă pentru spațiul dvs. de lucru.
</Warning>
Twenty expune un server [MCP](https://modelcontextprotocol.io/) astfel încât asistenții AI — Claude Desktop, Claude Code, Cursor, ChatGPT și alții — să poată citi și scrie datele tale din CRM prin limbaj natural.
Folosește **URL-ul spațiului de lucru** (URL-ul pe care îl folosești pentru a accesa Twenty) drept punct final MCP. Pe Twenty Cloud, URL-ul spațiului tău de lucru poate fi `https://{mycompany}.twenty.com` sau un domeniu personalizat. Serverul este disponibil la:
Ai două moduri de a-ți autentifica clientul MCP: **OAuth** (recomandat) sau **Cheie API**.
### Opțiunea A — OAuth (Recomandat)
Cu OAuth, clientul tău MCP deschide o fereastră de browser pentru a te autentifica. Nicio informație secretă nu este stocată în fișierele de configurare, iar tokenurile se reîmprospătează automat.
<Note>
OAuth necesită un client MCP care suportă [specificația MCP Authorization](https://modelcontextprotocol.io/specification/2025-03-26/basic/authorization). Claude Desktop, Claude Code, Cursor și ChatGPT o suportă.
</Note>
Adaugă asta în configurația clientului tău MCP, înlocuind `{your-workspace-url}` cu gazda spațiului tău de lucru (de ex. `mycompany.twenty.com`):
```json
{
"mcpServers": {
"twenty": {
"type": "streamable-http",
"url": "https://{your-workspace-url}/mcp"
}
}
}
```
Atât — nu este necesară nicio cheie API. Când clientul se conectează pentru prima dată, acesta va:
1. Va descoperi metadatele OAuth ale Twenty prin `/.well-known/oauth-protected-resource` și `/.well-known/oauth-authorization-server`
2. Se va înregistra ca un client OAuth prin înregistrare dinamică a clientului (RFC 7591)
3. Îți va deschide browserul pentru a autoriza accesul
4. Va primi tokenurile și se va conecta la serverul MCP
Conexiunile ulterioare reutilizează tokenurile stocate și le reîmprospătează automat.
### Opțiunea B — Cheie API
Dacă clientul tău MCP nu suportă OAuth sau preferi acreditări statice, furnizează o cheie API în antetul `Authorization`:
```json
{
"mcpServers": {
"twenty": {
"type": "streamable-http",
"url": "https://{your-workspace-url}/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}
```
<Warning>
Cheia ta API oferă acces la datele spațiului de lucru. Ține-o în afara controlului versiunilor și a dotfiles partajate.
</Warning>
Pentru a crea o cheie API, mergi la **Settings > APIs & Webhooks > + Create key**. Vezi [API-uri](/l/ro/developers/extend/api#create-an-api-key) pentru detalii.
## Pornire rapidă
### 1. Copiază configurația
Mergi la **Settings > AI > More > MCP Server** în Twenty. Alege metoda de autentificare (OAuth sau Cheie API), copiază fragmentul JSON (va folosi deja URL-ul spațiului tău de lucru) și lipește-l în fișierul de configurare al clientului tău MCP.
| **Claude Code** | `~/.claude.json` (utilizator) sau `.mcp.json` (proiect) |
| **Cursor** | `.cursor/mcp.json` în proiectul tău sau `~/.cursor/mcp.json` global |
| **ChatGPT** | Activează Modul pentru dezvoltatori în **Settings > Apps & Connectors > Advanced settings**, apoi folosește **Create** în **Settings > Apps & Connectors** pentru a adăuga serverul MCP |
### 2. Conectează-te
Repornește clientul tău MCP (sau reîncarcă configurația). Dacă folosești OAuth, vei fi redirecționat către Twenty pentru a autoriza accesul. Dacă folosești o cheie API, conexiunea este imediată.
### 3. Începe să-l folosești
Roagă-ți asistentul AI să interacționeze cu CRM-ul tău:
* *"Arată-mi cele 5 companii create cel mai recent"*
* *"Creează o persoană nouă numită Jane Doe la Acme Corp"*
* *"Găsește toate oportunitățile deschise cu o valoare mai mare de $10k"*
## Instrumente disponibile
După conectare, serverul MCP expune instrumente care oglindesc API-ul Twenty. Fluxul de lucru recomandat este:
1. **`get_tool_catalog`** — descoperă toate instrumentele disponibile
2. **`learn_tools`** — obține schema de intrare pentru instrumente specifice
3. **`execute_tool`** — rulează un instrument
Nu este nevoie să reții numele instrumentelor. Întreabă-ți asistentul AI ce poate face și va apela `get_tool_catalog` automat.
## Permisiuni
Conexiunile MCP moștenesc permisiunile utilizatorului autentificat (OAuth) sau rolul atribuit cheii API. Pentru a restricționa ce poate face serverul MCP:
* **OAuth**: Se aplică rolul utilizatorului din spațiul de lucru.
* **Cheie API**: Atribuie un rol cheii API în **Settings > Roles**. Vezi [Permisiuni](/l/ro/user-guide/permissions-access/capabilities/permissions).
## Configurație pentru auto-găzduire
Pentru instanțele auto-găzduite, înlocuiește `{your-workspace-url}` cu URL-ul serverului tău. Asigură-te că `SERVER_URL` din mediul tău corespunde URL-ului public al instanței tale Twenty — acesta este folosit pentru a genera metadatele de descoperire OAuth.
```bash
SERVER_URL=https://twenty.yourcompany.com
```
Punctul final MCP, punctele finale OAuth și metadatele de descoperire derivă toate din această valoare.
## Depanare
**Erori "Unauthorized" sau 401**
* OAuth: reautorizează ștergând tokenurile stocate din clientul tău MCP și reconectează-te.
* Cheie API: verifică dacă cheia este validă și nu a expirat. Regenereaz-o dacă este necesar.
**Fluxul OAuth nu deschide un browser**
* Asigură-te că clientul tău MCP suportă MCP Authorization. Dacă nu, revino la metoda cu Cheie API.
**Timeout de conexiune**
* Confirmă că URL-ul punctului final MCP este accesibil de pe calculatorul tău. Pentru instanțele auto-găzduite, verifică faptul că serverul rulează și că `SERVER_URL` este setat corect.
description: Creează un flux de lucru care folosește IA pentru a tria e-mailurile primite și pentru a trimite automat răspunsuri în același fir.
---
Răspunde la e-mailurile primite în câteva secunde — nu în ore. Acest flux de lucru folosește un Agent IA pentru a filtra zgomotul (newslettere, spam, răspunsuri automate) și pentru a redacta un răspuns personalizat la mesajele reale, apoi îl trimite ca răspuns în același fir, în cadrul conversației originale.
## Cum funcționează firele de e-mail
Fiecare e-mail are un antet ascuns `Message-ID` — o amprentă unică atribuită de serverul de e-mail al expeditorului. Când răspundeți la un e-mail, clientul dvs. de e-mail setează un antet `In-Reply-To` care face referire la acea amprentă. Astfel, Gmail, Outlook și orice alt client grupează mesajele în fire de discuție.
În Twenty, acea amprentă este stocată ca `headerMessageId` pe obiectul Message. Fluxul dvs. de lucru o preia și o transmite în câmpul In-Reply-To al acțiunii Send Email.
## Construirea fluxului de lucru
### Pasul 1: Creați un flux de lucru nou
Accesați **Settings -> Workflows** și faceți clic pe **+ New Workflow**.
### Pasul 2: Declanșare la mesajele primite
Alegeți **When a Record is Created** și selectați **Messages**.
De fiecare dată când un e-mail ajunge în Twenty, aceasta se declanșează.
Câmpul In-Reply-To este cel care face ca acesta să fie un răspuns, nu o conversație nouă. Destinatarul îl vede afișat în fir sub e-mailul original în Gmail, Outlook sau orice alt client.
**In-Reply-To** necesită un `message.headerMessageId` din declanșator — este amprenta unică a e-mailului, nu o adresă a destinatarului. Dacă îl lăsați necompletat, e-mailul se trimite în continuare, doar ca un mesaj independent.
</Tip>
<Warning>
Gmail folosește subiectul pentru a grupa mesajele în fire de discuție. Subiectul **trebuie** să înceapă cu `Re:` (inclusiv două puncte și spațiu) pentru ca Gmail să afișeze răspunsul în cadrul firului original. Fără acesta, răspunsul va apărea ca o conversație separată — chiar dacă antetul In-Reply-To este setat corect.
</Warning>
### Pasul 7: Testați și activați
Apăsați **Test**, apoi verificați clientul de e-mail. Răspunsul ar trebui să apară sub mesajul original.
Activați când sunteți mulțumit(ă) de rezultat.
## Idei pe care să le dezvoltați
* **Răspundeți doar VIP-urilor** — adăugați o ramură care verifică domeniul expeditorului sau dacă acesta există ca Contact în Twenty
* **Dirijați în funcție de intenție** — folosiți prompturi AI Agent separate pentru a trata diferit solicitările de vânzări față de cererile de asistență
* **Îmbogățiți înainte de a răspunde** — adăugați un pas Search Records pentru a include compania expeditorului sau istoricul tranzacțiilor în promptul AI, pentru răspunsuri mai personalizate
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.