RestApiExceptionFilter is used as an exception filter for the core
controller which is used for crud operations on our objects (equivalent
of our dynamic queries findManyPeople etc. on the graphql API).
Exceptions were leading a 400 / BadRequestException response status
which can be confusing to users.
By default we should actually throw a 500 if the error was not handled
priorily, but we have not implemented input validation for the REST api
so we fear to be flooded with errors that should not be 500 but 400 due
to user inputs. A solution should be brought [with this
ticket](https://github.com/twentyhq/core-team-issues/issues/1027) but it
has not been prioritized yet.
This PR aims at improving readability in sentry and user experience with
runtime errors.
**GraphQL errors (and ApolloError)**
1. In sentry we have a lot of "Object captured as exception with keys:
extensions, message" errors (2k over the last 90d), on which we have
zero information. This is because in apollo-factory we were passing on
GraphQL errors to sentry directly why sentry expects the structure of a
JS Error. We are now changing that, rebuilding an Error object and
attempting to help grouping by creating a fingerPrint based on error
code and truncated operationName (same as we do in the back for 500
graphql errors).
2. In sentry we have a lot of ApolloError, who actually correspond to
errors that should not be logged in sentry (Forbidden errors such as
"Email is not verified"), or errors that are already tracked by back-end
(Postgres errors such as "column xxx does not exist"). This is because
ApolloErrors become unhandled rejections errors if they are not caught
and automatically sent to sentry through the basic config. To change
that we are now filtering out ApolloErrors created from GraphQL Errors
before sending error to sentry:
<img width="524" alt="image"
src="https://github.com/user-attachments/assets/02974829-26d9-4a9e-8c4c-cfe70155e4ab"
/>
**Runtime errors**
4. Runtime errors were all caught by sentry with the name "Error",
making them not easy to differentiate on sentry (they were not grouped
together but all appeared in the list as "Error"). We are replacing the
"Error" name with the error message, or the error code if present. We
are introducing a CustomError class that allows errors whose message
contain dynamic text (an id for instance) to be identified on sentry
with a common code. _(TODO: if this approach is validated then I have
yet to replace Error with dynamic error messages with CustomError)_
5. Runtime error messages contain technical details that do not mean
anything to users (for instance, "Invalid folder ID: ${droppableId}",
"ObjectMetadataItem not found", etc.). Let's replace them with "Please
refresh the page." to users and keep the message error for sentry and
our dev experience (they will still show in the console as uncaught
errors).
In morph relation pickers, we were not taking into account permissions
when computing the list of objects to search for, while we should not
search for objects we don't have read permissions on (permission denied
error)
Let's introduce an object-limited role for Tim, to test and/or spot
incompatibilities with restricted permissions in the future.
Our main user tim@apple.dev is now assigned a role that has all settings
permissions, and all object permissions except for update on Pets (to
test read-only view) and read on Rockets.
Since we still need an admin user for each workspace we are introducing
a new member, Jane, who has the admin role
---------
Co-authored-by: Félix Malfait <felix@twenty.com>
Note and task tabs in side panel should only show if user has reading
permission on them.
"Go to companies", "Go to workflows", etc. in command menu should only
show is user has reading permission on related objects.
<img width="507" alt="Capture d’écran 2025-06-17 à 11 09 50"
src="https://github.com/user-attachments/assets/3a2a4c25-0b9b-4ee6-b18f-b019b8a56d47"
/>
<img width="505" alt="Capture d’écran 2025-06-17 à 11 09 56"
src="https://github.com/user-attachments/assets/8a219955-cc8e-4dbf-a4f9-a50e1aaa4b59"
/>
**How to test**
Assign a user with a custom role that has **no** read permissions on
notes/tasks/workflows/companies/opportunities/people (no need to test
them all but at least one between note and tasks; workflows; one between
companies/opportunities/people). Check that you don't see the related
tab / action.
---------
Co-authored-by: Charles Bochet <charles@twenty.com>
Closes https://github.com/twentyhq/core-team-issues/issues/992
Occasionaly, users can have a subscription created but still have their
workspace not activated and therefore not have a role yet, if they did
not go through the whole flow the first time. This causes a permission
check error while calling checkoutSession, while it shouldn't.
We detected the error through sentry. Since there has been no occurences
for the past three weeks in aggregateCompanies and getCurrentUser
transactions (while we have daily errors in checkoutSession), I assume
it has been fixed in the meantime. If not it will pop again on sentry
anyway !
<img width="798" alt="Capture d’écran 2025-06-16 à 18 38 43"
src="https://github.com/user-attachments/assets/2067c166-8b19-4c83-9270-6e49ee7ae0f5"
/>
Closes https://github.com/twentyhq/core-team-issues/issues/868
We should not allow to grant any writing permission (update, soft
delete, delete) on an object or at role-level without the reading
permission at the same level.
This has been implemented in the front-end at role level, and is yet to
be done at object level (@Weiko)
We need to use twentyORMManager and not twentyORMGlobalManager in rest
api base handler, because we don't want to bypass permissions using
`shouldBypassPermissions` parameter (which we would have to do to use
twentyORMGlobalManager).
ScopedWorkspaceContextFactory was not adapted to rest api requests which
form differs from graphql request.
If permissionsV2 feature flag is toggled, we should recompute the
permissions.
We decided to make each WorkspaceXxCacheService Xx-specific (feature
flag, permissions...), so we are not recomputing permission cache from
workspaceFeatureFlagCacheService where feature flags are recomputed,
even if that would be a lower level than FeatureFlagService. This allows
to avoid complex circuclar dependency and keeps a clear purpose for each
service.
In this PR
1. fix workflow step creation by adding forgotten
`shouldBypassPermissionChecks` in WorkflowVersionStepWorkspaceService
2. clarify the rule for twentyORMGlobalManager: do not add unnecessary
`shouldBypassPermissionChecks` for system objects (there are no
object-records permission checks on system objects, they are dealt with
at resolver level)
In this PR
- Determine object record permissions on workflows objects (workflow,
workflowVersion, workflowRun) base on settings permissions @Weiko
- Add Workflow permission guards on workflow resolvers @thomtrp . **Any
method within a resolver that has the SettingsPermission Guard is only
callable by a apiKey or a user that has the permission** (so not by
external parties).
- Add checks bypass in workflow services since 1) for actions gated by
settings permissions, the gate should be done at resolver level, so it
will have been done before the call to the service 2) some service
methods may be called by workflowTriggerController which is callable by
external parties without permissions (ex:
workflowCommonWorkspaceService.getWorkflowVersionOrFail). This is
something we may want to change in the future (still to discuss), by
removing the guard at resolver-level and relying on
shouldBypassPermissionChecks at getRepository and made in a way that we
only bypass for external parties.
- Add checks bypass for actions performed by workflows since they should
not be restricted in our current vision
- Add tests
In this PR
1. Add missing override of insert() method on
WorkspaceSelectQueryBuilder to return our custom
WorkspaceInsertQueryBuilder with permission checks.
2. Replace override implementation of methods on WorkspaceEntityManager
that call createQueryBuilder at a nested internal layer of typeORM (i.e.
not directly in the initial implementation of EntityManager - unlike
findBy for instance -, but in calls done under the hood at a level which
would force us to override entire other classes to pass on our
permissionOptions. It is the case for methods which call typeORM's
EntityPersistExecutor for instance.), to validate permissions and then
allow the subsequent calls to be made without permission checks
3. adapt tests
---------
Co-authored-by: Charles Bochet <charles@twenty.com>
In this PR
1. adding tests on relations and nested relations to make sure that if
any permission is missing, the query fails
2. adding tests on objectRecord permissions to make sure that
permissions granted or restricted by objectPermissions take precedence
on the role's allObjectRecords permissions
In the frame of https://github.com/twentyhq/core-team-issues/issues/924
- Rename dataSource -> workspaceDataSource when relevant to ease
understandability
- override workspaceDataSource.createQueryBuilder, because we don't want
developers to use it directly since it does not run permission checks at
this level. Indeed, we cannot do so because 1) datasources are shared
between roles so we would need to re-think its implementation to make
that possible, while for now we never call
workspaceDatasource.createQueryBuilder in our codebase 2)
workspaceEntityManager.createQueryBuilder, that we have overriden with
permission checks, then performs a call to
workspaceDataSource.createQueryBuilder so that would make two permission
checks.
Closes https://github.com/twentyhq/core-team-issues/issues/605
Actually settingsPermissions checks were already implemented, but we had
no tests on them.
In the ticket we had mentioned
_TO DO: in pemissions.service we should stop calling
userRoleService.getRolesByUserWorkspaces and call
getRoleIdForUserWorkspace instead which relies on the cache._
But actually roleId is not enough for settings permissions because we
don't store them in the cache (unlien object records permissions - which
I think we had forgotten about when adding that TODO.), so we will still
need to make a db call to load the role's settingsPermissions. I think
it's better to make just one db call to get the role and
settingsPermissions from userWorkspaceId (as currently) than to make one
redis call to get roleId for userWorksapce then one db call to get role
and its settingsPermissions).
Closes https://github.com/twentyhq/core-team-issues/issues/748
In the frame of the work on permissions we
- remove all raw queries possible to use repositories instead
- forbid usage workspaceDataSource.executeRawQueries()
- restrict usage of workspaceDataSource.query() to force developers to
pass on shouldBypassPermissionChecks to use it.
---------
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
I encountered a bug where I was missing permissions while calling
searchResolver because the repository from
`twentyORMManager.getRepository` was missing permissions itself.
The repository was returned from the cached repositories map using a
repository key feature the roleId, the rolesVersion and
featureFlagMapVersion.
I was not able to reproduce but this error should not go unnoticed: we
always expect to find objectPermissions for every roleId in the
datasource now.
I was not able to understand what happened for now but I think throwing
the error will help keeping an eye on it
Since https://github.com/twentyhq/twenty/pull/12286 we are now capturing
in sentry graphql queries errors in the FE.
We want to exclude InternalServerErrors because they are already
captured in sentry from the BE.
We should capture graphQL exceptions thrown in the FE in Sentry.
All the more so as we have just cleaned back-end errors in sentry,
preventing 4xx errors from being wrongfully sent to sentry.
Those 4xx errors should, except for `Unauthenticated` and `Forbidden`
errors (for now - this list can evolve), trigger a sentry FE error, as
we are not suppose to let users of the product interface trigger queries
that will fail with 4xx errors (for instance a malformed input).
We still miss an efficient way to group those errors together in sentry.
It could be the message but the message may be different for each user
if it contains user-specific data, and we don't always have control on
the message.
This can be done later as we iterate on improving sentry
Fixes https://github.com/twentyhq/twenty/issues/11566 + translates dates
- Date display bug: We had an issue with date (not date time) display
depending on the timezone the user had selected. The date is stored in
the db as yyyy-mm-dd, eg. 2025-05-01 for **May 1st, 2025**. When
returned this date is formatted in a UTC DateTime at midnight, so
2025-05-1 00:00:00. Then when displaying the date we were converting
this date using the timeZone, so 2025-04-30 17:00:00, thus displaying
**April 30th, 2025**. The fix chosen is that we should not take into
account the timezone for date (not date time!) displays as we always
want to show the same date.
- Date translation: dates were not translated, not in their default
display (_May 1st, 2025_) nor in their relative display (_about a month
ago_). The lib we use for date formatting, date-fns, offers a
translation option with pre-built `Locale`s from their lib.
Unfortunately and surprisingly we cannot just use directly string locale
codes (like `fr-FR`), cf [open issue on
date-fns](https://github.com/date-fns/date-fns/issues/3660). A util was
introduced to offset this by dynamically importing the right date-fns
Locale based on the locale code.
Introducing a class of RelationException extending CustomException to
help grouping those exception in sentries by ExceptionCode.
I did not introduce a filter as these are thrown in utils that can be
used in multiple places now or in the future, and filters are to be
added at resolver-level.
We have approvedAccessDomain custom exceptions, but they were never
filtered while some of them reflects 4xx errors which we don't want to
be captured as 5xx errors
Follow-up on https://github.com/twentyhq/twenty/pull/12007
In this PR
- adding a filter on HttpExceptionHandlerService to filter out 4xx
errors from driver handling (as we do for graphQL errors: see
useGraphQLErrorHandler hook - only filteredIssues are sent to`
exceptionHandlerService.captureExceptions()`.)
- grouping together more missing metadata issues
- attempting to use error codes as issues names in sentry to improve UI;
for now it says "Error" all the time
Fixes https://github.com/twentyhq/twenty/issues/11864 and
https://github.com/twentyhq/core-team-issues/issues/908
We should not send `createManyXXX` mutations with FE-forged ids in the
payload if we want to do an upsert, because that 1) prevents records
from being merged 2) triggers optimistic rendering while we can't know
before-hand which records will actually be created and which records
will only be updated
Also noticed createdBy was being overriden even for records we are
updating and not creating, which did not seem right, so fixed that too
This PR attemps at improving sentry grouping and filtering by
- Using the exceptionCode as the fingerprint when the error is a
customException. For this to work in this PR we are now throwing
customExceptions instead of internalServerError deprived of their code.
They will still be converted to Internal server errors when sent back as
response
- Filtering 4xx issues where it was missing (for emailVerification
because errors were not handled, for invalid captcha and billing errors
because they are httpErrors and not graphqlErrors)
---------
Co-authored-by: Félix Malfait <felix@twenty.com>
In this PR we are
1. cleaning typeORM service by removing connectToDataSource method
2. using workspaceDataSource instead of mainDataSource when possible,
and replacing raw SQL with workspaceRepository methods to use
First and main step of
https://github.com/twentyhq/core-team-issues/issues/747
We are implementing a permission check layer in our custom
WorkspaceEntityManager by overriding all the db-executing methods (this
PR only overrides some as a POC, the rest will be done in the next PR).
Our custom repositories call entity managers under the hood to interact
with the db so this solves the repositories case too.
This is still behind the feature flag IsPermissionsV2Enabled.
In the next PR
- finish overriding all the methods required in WorkspaceEntityManager
- add tests
Closes https://github.com/twentyhq/core-team-issues/issues/857
The issue was caused by the fact that the preview chip was accidentally
made clickable while not linked to any record id:
<img width="763" alt="Capture d’écran 2025-04-28 à 15 17 32"
src="https://github.com/user-attachments/assets/c1d9bf61-edcb-442f-a914-eccc627ee190"
/>
this was fixed by [this
PR](https://github.com/twentyhq/twenty/pull/11745) (@etiennejouan)
It was causing the side panel to open while the record id was empty,
while this recordId is used in query filters (as it should be to fetch
record data), leading the queries to fail.
Let's early return with an error instead as it does not make sense to
open the record page with an empty recordId.
---------
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
In this PR we are
- introducing a cached map `{ userworkspaceId: roleId } `to reduce calls
to get a userWorkspace's role (we were having N+1 around that with
combinedFindMany queries and generally having a lot of avoidable
queries)
- using the roles permissions cache (`{ roleId: { objectNameSingular:
{ canRead: bool, canUpdate: bool, ...} } `) in Permissions V1's
userHasObjectPermission, in order to 1) improve performances to avoid
calls to get roles 2) start using our permissions cache
In this PR we are
- (if permissionsV2 is enabled) executing permission checks at query
builder level. To do so we want to override the query builders methods
that are performing db calls (.execute(), .getMany(), ... etc.) For now
I have just overriden some of the query builders methods for the poc. To
do so I created custom query builder classes that extend typeorm's query
builder (selectQueryBuilder and updateQueryBuilder, for now and later I
will tackle softDeleteQueryBuilder, etc.).
- adding a notion of roles permissions version and roles permissions
object to datasources. We will now use one datasource per roleId and
rolePermissionVersion. Both rolesPermissionsVersion and rolesPermissions
objects are stored in redis and recomputed at role update or if queried
and found empty. Unlike for metadata version we don't need to store a
version in the db that stands for the source of truth. We also don't
need to destroy and recreate the datasource if the rolesPermissions
version changes, but only to update the value for rolesPermissions and
rolesPermissionsVersions on the existing datasource.
What this PR misses
- computing of roles permissions should take into account
objectPermissions table (for now it only looks at what's on the roles
table)
- pursue extension of query builder classes and overriding of their db
calling-methods
- what should the behaviour be for calls from twentyOrmGlobalManager
that don't have a roleId?
Part of https://github.com/twentyhq/core-team-issues/issues/526
An active workspace's defaultRoleId should never be null.
We can't rely on a simple postgres NOT NULL constraint as defaultRoleId
will always be initially null when the workspace is first created since
the roles do not exist at that time.
Since a suspended workspace can be active again, we should maintain that
rule during the workspace suspension. The only moment defaultRoleId can
have a null value is during the onboarding. this pr enforces that
Closes https://github.com/twentyhq/core-team-issues/issues/526
(for reminder:
1. Make defaultRoleId non-nullable for an active workspace
2. Remove permissions V1 feature flag
3. Set member role as default role for new workspaces
About 1.:
An active workspace's defaultRoleId should never be null.
We can't rely on a simple postgres NOT NULL constraint as defaultRoleId
will always be initially null when the workspace is first created since
the roles do not exist at that time.
Let's add a more complex rule to ensure that
About 3.:
In the first phase of our deploy of permissions, we chose to assign
admin role to all existing users, not to break any existing behavior
with the introduction of the feature (= existing users have less rights
than before).
As we deploy permissions to all existing and future workspaces, let's
set the member role as default role for future workspaces.
)
In [a previous PR](https://github.com/twentyhq/twenty/pull/11023) I
fixed the issue where users where re-assigned the default role when
signin up using SSO.
The "fix" actually introduced another error, calling
`assignRoleToUserWorkspace` only when a user is signing up to twenty,
forgetting about the case where an existing twenty user signs up to a
different workspace. They should still be assigned the role in that
case.
To fix this and improve clarity, I moved assignRoleToUserWorkspace to
addUserToWorkspace and renamed addUserToWorkspace to
addUserToWorkspaceIfUserNotInWorkspace since this is at each sign in but
does nothing if user is already in the workspace.
I think ideally we should refactor this part to improve readability and
understandability, maybe with different flows for each case: signIn and
signUp, to twenty or to a workspace
When logging using a SSO method, we call signInUp service in which we
were wrongfully assigning a role to the user even if the user is signing
in and not signin up.
This went unnoticed during our QA as a different sign-in method is
called when logging with the credentials.
---------
Co-authored-by: Weiko <corentin@twenty.com>
Closes https://github.com/twentyhq/core-team-issues/issues/469 and
https://github.com/twentyhq/core-team-issues/issues/500
In this PR
1. stop conditioning permission initialization for a workspace to env
variable value. Instead we want to create and assign permissions and
roles in all new workspaces. For now that will be totally silent.
2. temporarily, the default role is set to the admin role for new
workspaces. it will also be the case for existing workspaces through the
backfill command. Member role is still being created though. (when we
will do the final roll-out we will update this so that future workspaces
have the member role as default role. our goal here is not to break any
current behaviour for users, that today have all have the equivalent of
admin rights).
During sign-up, new users setting new workspaces are required to choose
a billing plan (30-days or 7 days). They are then redirected to
billing.twenty.com to confirm and configure their plan choice.
Then they are quickly redirected to /settings/billing before being
redirected to /create-workspace.
The problem is that even though /create-workspace is not a gated path,
/settings/billing is: it is a path gated by WORKSPACE permission which
has not been granted to the user at this stage. therefore the user is
not redirected to /create-workspace since they do not reach
/settings/billing path but is redirected to the profile page instead.
(To see this feature flag must be removed + billing enabled: you can use
this branch [from this closed
PR](https://github.com/twentyhq/twenty/pull/10570)).
The chosen workaround is to bypass, in the FE, the permission check for
WORKSPACE permission if the workspace's activation status is
PENDING_CREATION. There is no need for any BE change.
A user should not be able to delete their account if they are the last
admin of a workspace.
It means that if a user wants to sign out of twenty, they should delete
their workspace, not their account
In this PR
- updateWorkspaceMemberRole api was changed to stop allowing null as a
valid value for roleId. it is not possible anymore to just unassign a
role from a user. instead it is only possible to assign a different role
to a user, which will unassign them from their previous role. For this
reason in the FE the bins icons next to the workspaceMember on a role
page were removed
- updateWorkspaceMemberRole will throw if a user attempts to update
their own role
- tests tests tests!
- Adding permission gates on workspaceMember to only allow user with
admin permissions OR users attempting to update or delete themself to
perform write operations on workspaceMember object
- Reverting some changes to treat workflow objects as regular metadata
objects (any user can interact with them)
- (fix) Block updates on soft deleted records
Adding permission gates on all workspace-invitations endpoints:
sendInvitation, resendInvitation, deleteWorkspaceInvitation,
findWorkspaceInvitations (the latter being from my understanding only
used to list the invitations to then re-send them or detee them).
+ tests on Api & webhooks permission gates
Following a conversation with @etiennejouan and @martmull, we are adding
a permission gate on billing resolver's checkoutSession, which should
only be accessible to entitled users or at workspace creation (when
there are no roles yet), when the subscription is incomplete
Closes https://github.com/twentyhq/core-team-issues/issues/393
- enforcing object-records permission checks in resolvers for now. we
will move the logic to a lower level asap
- add integration tests that will still be useful when we have moved the
logic
- introduce guest seeded role to test limited permissions on
object-records
This was updated a few weeks ago and went unnoticed since 1) integration
tests were broken + 2) we don't have actionnable updateMany mutations in
the product at the moment
It will fix some tests, at least
all-people-resolvers.integration-spec.ts for instance
After introducing userWorkspaceId into JWTs, we were wrongfully
executing
```
const userWorkspace = await this.userWorkspaceRepository.findOne({
where: {
id: payload.userWorkspaceId,
},
});
```
which would return a random userWorkpace if `payload.userWorkspaceId` is
undefined.
All generated JWTs have had a userWorkspaceId for more than a week now,
but in tests we had not modified the accessToken in use, which did not
have a userWorkspaceId, until [this
pr](https://github.com/twentyhq/twenty/pull/10204)
In this PR
- closing https://github.com/twentyhq/core-team-issues/issues/313
- adding permission gates on workspace settings and security settings
- adding integration tests for each of the protected setting and
security
Adding SettingsPermissionsGuard to execute permission check.
The guard is added directly in resolver, either at resolver level (ex:
roles) or resolver-endpoint level (ex: metadata). this can be challenged
!
In this PR, we are implementing the updateRole endpoint with the
following rules
1. A user can only update a member's role if they have the permission (=
the admin role)
2. Admin role can't be unassigned if there are no other admin in the
workspace
3. (For now) as members can only have one role for now, when they are
assigned a new role, they are first unassigned the other role (if any)
4. (For now) removing a member's admin role = leaving the member with no
role = calling updateRole with a null roleId
In this PR
- introducing roles module to separate roles logic (assign a Role, get a
workspace's roles etc.) from permission logic (check if a user has a
permission)
- Introduces getRoles endpoint to fetch a workspace's roles
- introduces the first permission check: getRoles in only accessible to
users with permission on ROLE setting. Implemented
validatesUserHasWorkspaceSettingPermissionOrThrow
Marie