⚠️ **AI-generated PR — not ready for review** ⚠️
cc @FelixMalfait
---
## Changes
### System prompt improvements
- Explicit skill-before-tools workflow to prevent the model from calling
tools without loading the matching skill first
- Data efficiency guidance (default small limits, use filters)
- Pluralized `load_skill` → `load_skills` for consistency with
`load_tools`
### Token usage reduction
- Output serialization layer: strips null/undefined/empty values from
tool results
- Lowered default `find_*` limit from 100 → 10, max from 1000 → 100
### System object tool generation
- System objects (calendar events, messages, etc.) now generate AI tools
- Only workflow-related and favorite-related objects are excluded
### Context window display fix
- **Bug**: UI compared cumulative tokens (sum of all turns) against
single-request context window → showed 100% after a few turns
- **Fix**: Track `conversationSize` (last step's `inputTokens`) which
represents the actual conversation history size sent to the model
- New `conversationSize` column on thread entity with migration
### Workspace AI instructions
- Support for custom workspace-level AI instructions
---------
Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com>
Fix#17711
### Reproduce steps
1. import this xlsx file
[test-import.xlsx](https://github.com/user-attachments/files/25156642/test-import.xlsx)
2. import company worksheet at first and then import people worksheet
3. you will see an error <img width="324" height="101"
alt="Snipaste_2026-02-07_19-22-04"
src="https://github.com/user-attachments/assets/bf2703da-a57a-4795-805b-6ddcc689c621"
/>
And I found neither the frontend nor the backend applies lowercase
normalization to the query value. Although the standard UI input always
displays company links in lowercase, user might mixed the case in their
xlsx/csv bulk import. So I did a simple check in frontend.
### Additional findings:
1. Email has the same case sensitivity issue when opportunities import.
You can test same as in test-import.xlsx file (I created a opportunities
worksheet. It will have same issue: <img width="330" height="101"
alt="email error"
src="https://github.com/user-attachments/assets/db36b19b-2abe-4c61-a661-f8d1eb357a5e"
/>
2. API also has: I also tested via the GraphQL API Playground and
confirmed that createOnePerson with a mixed-case URL fails to find an
existing company. <img width="1419" height="513"
alt="Snipaste_2026-02-07_23-33-56"
src="https://github.com/user-attachments/assets/c189f1fd-9793-42d5-a1a9-616cffd2592e"
/>
### Approach:
1. Only change it in frontend, and I will add email check later; (my
current commit)
2. Or backend: Normalize values in computeUniqueConstraintCondition in
twenty-server/src/engine/twenty-orm/utils/compute-relation-connect-query-configs.util.ts
(which handles connect.where queries, and this will cover xlsx import
and api import: createone, createmany, updatemany, updateone. And much
better for future extension.
(Personally, I'd prefer the backend approach. But I'd love to hear your
thoughts on which approach you'd prefer, and whether my analysis is on
the right track.
---------
Co-authored-by: Etienne <45695613+etiennejouan@users.noreply.github.com>
## Summary
- Validates that the timezone parameter in group-by date expressions is
a recognized IANA timezone
- Adds SQL string literal escaping as a defense-in-depth measure for the
timezone value interpolated into SQL expressions
- Moves the `IANA_TIME_ZONES` constant to `twenty-shared` so it can be
reused across frontend and server packages
- Adds `INVALID_TIMEZONE` error code mapped to 400 Bad Request in both
GraphQL and REST API exception handlers
## Test plan
- [x] Unit tests for `validateIanaTimeZone` (valid IANA zones, fixed
offsets, rejects invalid strings)
- [x] Unit tests for `escapeSqlStringLiteral`
- [x] Integration tests for `getGroupByExpression` covering timezone
validation and granularity handling
Made with [Cursor](https://cursor.com)
---------
Co-authored-by: Cursor <cursoragent@cursor.com>
Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com>
On iterator creation, we get an error "Bad configuration". Looks like a
race condition with generation that happens before apollo cache gets
updated. Adding defensive checks
**What this PR does:**
- Migrates existing `Favorite` and `FavoriteFolder` entities to the new
`NavigationMenuItem` structure
- Preserves user-level vs workspace-level ownership
- Preserves folder structure, positions, and relationships
- Handles both view-based favorites (linked to views) and record-based
favorites (linked to records)
- Soft-deletes original favorites and folders after successful migration
- Enables the `IS_NAVIGATION_MENU_ITEM_ENABLED` feature flag
post-migration
- Skips migration if the feature flag is already enabled
- Idempotent: checks for existing navigation menu items to prevent
duplicates
**Migration flow:**
1. Migrate favorite folders first (creates folder mapping)
2. Migrate favorites
3. Soft-delete migrated favorites and folders
4. Enable feature flag
**What's next:**
- After all workspaces are migrated and the navigation menu item feature
is fully rolled out, we can:
- Remove the old `Favorite` and `FavoriteFolder` entities and related
code
- Remove the feature flag check and make navigation menu items the
default
- Clean up any deprecated favorites-related code paths in the frontend
---------
Co-authored-by: Etienne <45695613+etiennejouan@users.noreply.github.com>
- Add FILES field on attachment
- Adapt Attachment logic in front to use new resolver/controller
- Update files-field logic to infer applicationId from fieldMetadataId +
ask for fieldMetadataId in upload resolver
- Design update
To do in next PR :
- Adapt activity files logic
We generate Field widgets for relations on-the-fly, when a record page
layout is first requested by the user. When the user changed their data
model and then returned to a record page layout, the Field widgets
weren't updated. **This PR ensures Field widgets are recomputed when
relations change.**
Future subjects:
- Now that we will fetch the configuration from the backend and start
storing updates, we will have to think about how we deal with these
generated relation Field widgets.
## Before
https://github.com/user-attachments/assets/99d53b19-b231-435f-b14f-4473ba269ad2
## After
https://github.com/user-attachments/assets/357d956d-8b3b-448c-a983-82569a7dd0a0
---------
Co-authored-by: Félix Malfait <felix.malfait@gmail.com>
## Summary
Three fixes in this PR:
### 1. Migration commands now set `isActive = true` and use generic
'Target' label
When converting relation fields to `MORPH_RELATION` type, the migration
commands now:
- Set `isActive = true` to prevent inactive fields from being selected
as the representative morph field
- Update all field labels to generic **'Target'** instead of keeping
individual labels like 'Company', 'Person'
This ensures the UI shows a coherent label ('Target') alongside 'X
Objects' for the type.
**Files changed:**
- `1-17-migrate-note-target-to-morph-relations.command.ts`
- `1-17-migrate-task-target-to-morph-relations.command.ts`
### 2. Added system fields/relations toggles in Settings
The fields and relations tables in Settings > Data Model were filtering
out system fields with no way to view them. Added a "System fields" /
"System relations" toggle (visible in advanced mode) to allow viewing
these fields.
**Files changed:**
- `SettingsObjectFieldTable.tsx`
- `SettingsObjectRelationsTable.tsx`
This matches the existing behavior on the Objects table which already
has a "System objects" toggle.
This bug was cause by a combination of an expired token stored inside
SSE client on the front end, and a server restart or a cache flush, we
ended up with an SSE client that tries to reconnect infinitely with an
expired token.
The fix is to improve the connection retry handler in the frontend to
detect an expired token and recreate an SSE client.
## Overview
Fixes the `unknown fields opportunityId in objectMetadataItem
noteTarget` error when creating notes/tasks from record pages (like the
Notes/Tasks tab on an opportunity).
## Root Cause
The `useOpenCreateActivityDrawer` hook was not handling `MORPH_RELATION`
field types:
1. Code only checked `field.relation` (which is populated for `RELATION`
type fields)
2. When `field.relation` was undefined (because the field is
`MORPH_RELATION`), it fell back to the old naming convention (e.g.,
`opportunityId`)
3. But after the morph migration, the columns are named differently
(e.g., `targetOpportunityId`)
This caused the error on both new workspaces (which are created with
morph relations from the start) and existing workspaces that ran the
migration.
## Solution
Use the existing `findTargetFieldInfo()` utility which properly handles
both `RELATION` and `MORPH_RELATION` field types by:
- Checking `morphRelations` for morph fields and computing the correct
field name
- Checking `relation` for regular relation fields
- Returning the correct `joinColumnName` for each case
This utility is already used elsewhere in the codebase (e.g., in
junction field handling) and is the cleanest solution.
# Refactor workflow–logic function interaction
## Why
Workflow code steps and standalone logic functions shared the same build
layer and DB layer, which blurred two use cases: code steps belong to a
workflow version; standalone functions are deployable units. That made
workflow code steps harder to own and evolve.
## Goal
Treat code steps as **workflow-owned**: build and run them in workflow
context, and expose workflow-scoped APIs so the editor can load, test,
and save code step source without going through the generic
logic-function layer.
## Context
Previously, if for example a Person had a companyId but the company
relation was null due to RLS, the company column appeared empty. Now it
displays ForbiddenFieldDisplay to indicate the relation exists but is
inaccessible.
When RLS restricts access to a related record, the frontend now shows a
"Not shared" indicator (with lock icon) instead of an empty field.
<img width="1275" height="399" alt="Screenshot 2026-02-04 at 15 37 53"
src="https://github.com/user-attachments/assets/870306f8-f811-4dc0-b23b-a33e89043a37"
/>
## Remove logic function layer
Package.json and yarn.lock are now on the application entity, so the
logic function layer is no longer used except as a legacy source for the
1.17 backfill. This PR removes all layer usage outside of that migration
and keeps only the entity for backfill.
### Summary
- **Kept:** `LogicFunctionLayerEntity` and its table, only used by the
1.17 backfill command to read legacy layer data and backfill application
package files.
- **Removed:** All other layer logic: CRUD, cache, resolvers, services,
DTOs, and frontend types. Logic functions now depend only on the
application for package/dependency context.
### Why Dependencies instead of Source for package files
Package.json and yarn.lock are the application’s dependency set and are
stored under the application in **FileFolder.Dependencies**. The build
service and drivers now read them only from Dependencies; nothing is
written to Source for these files.
This PR migrates `noteTarget` and `taskTarget` to morph relations behind
separate feature flags, following the Attachment/TimelineActivity
pattern.
It introduces the `IS_NOTE_TARGET_MIGRATED` and
`IS_TASK_TARGET_MIGRATED` flags, updates standard field metadata and
indexes to use morph relations, and adds two **1.17 workspace
migrations** that:
- rename `noteTarget.*Id` / `taskTarget.*Id` columns to `target*Id`
- convert the corresponding field metadata to `MORPH_RELATION` with a
shared `morphId`
On the frontend, note/task target read and write paths switch to
`target*Id` when the respective flag is enabled. Deleted targets are
filtered on reload to prevent reappearing relations.
# Backfill application package files for custom and standard apps
- Backfill `package.json` / `yarn.lock` (and related fields) for
existing workspaces; new standard/custom apps get default dependency
files.
- Default package files under
`application/constants/default-package-files/`; util with hardcoded
checksums (comment on how to regenerate).
- New **FileFolder.Dependencies** for app dependency files;
**writeFile_v2** accepts optional `queryRunner` for transactional
writes.
Usages:
- **Upgrade command** `upgrade:1-17:backfill-application-package-files`:
standard/custom apps → default files; other apps → from logic function
layer.
- **Workspace creation**: create workspace with
`workspaceCustomApplicationId` first, then create application (enables
same-transaction insert). Migration makes workspace/application/file FKs
deferrable.
- dev seeder
## Root Cause
The `DateTimePicker` component lazy-loads `react-datepicker` using
React's `lazy()` with a `Suspense` fallback that shows skeleton loaders:
```typescript
const ReactDatePicker = lazy<ComponentType<DatePickerPropsType>>(() =>
import('react-datepicker').then(...)
);
// In render:
<Suspense fallback={<SkeletonLoader />}>
<ReactDatePicker ... />
</Suspense>
```
On slower CI runners (GitHub Actions vs depot.dev), the lazy load takes
longer, causing tests to timeout while still showing skeletons instead
of the actual date picker content.
## Fix
Increased `findByText` timeout from the default 1000ms to 10000ms for
tests that wait for the date picker to load:
- `DateTimeFieldInput.stories.tsx` - 4 stories fixed
- `InternalDatePicker.stories.tsx` - 2 stories fixed
## Why This Wasn't Flaky Before
depot.dev runners have faster I/O and more consistent performance, so
the lazy load completed quickly. GitHub Actions runners have more
variable performance, causing the load to sometimes exceed the 1000ms
default timeout.
## Summary
Follow-up to #17656 - the `RelationToOneFieldDisplay` performance test
is still failing on GitHub Actions runners.
**Failure:** 0.313ms actual vs 0.3ms threshold
**Fix:** Increased threshold from 0.3ms → 0.4ms to provide more headroom
for runner variability.
- fixedOverflowWidgets fixes the suggestions being overflow outside the
editor
- on focus, stop event propagation to avoid catching spaces as document
level events
## Summary
After migrating from depot.dev runners to GitHub Actions runners, three
performance tests are failing due to slower single-threaded performance
on the new infrastructure (even with 16 cores).
## Problem
The performance tests use React's `<Profiler>` API to measure component
render times. depot.dev runners have better single-core performance and
lower memory latency compared to GitHub Actions runners, even larger
ones.
Failed tests:
| Test | Threshold | Actual | Overage |
|------|-----------|--------|---------|
| `DateTimeFieldDisplay` | 0.2ms | 0.281ms | +41% |
| `ChipFieldDisplay` | 0.2ms | 0.222ms | +11% |
| `RelationToOneFieldDisplay` | 0.22ms | 0.237ms | +8% |
## Solution
Increased thresholds to account for GitHub Actions runner performance:
- `DateTimeFieldDisplay`: 0.2ms → 0.35ms
- `ChipFieldDisplay`: 0.2ms → 0.3ms
- `RelationToOneFieldDisplay`: 0.22ms → 0.3ms
## Future Considerations
For a more robust long-term solution, we could implement baseline
comparison that:
1. Saves performance results on main branch builds
2. Compares PR results against baseline with tolerance (±25%)
3. Self-calibrates to actual runner performance
This would catch real regressions without being sensitive to
infrastructure differences.
# Introduction
Following https://github.com/twentyhq/twenty/pull/17632 and
https://github.com/twentyhq/twenty/pull/17572
This PR deprecates the agent, skill, field metadata and role
`standardId` in favor of the `universalIdentifier` usage
## Note
- Removed previous standard ids declaration modules
- Twenty-sdk now re-exports the `STANDARD_OBJECTS` universalIdentifier
hashmap constant
- deleted some sync-metadata deadcode too ( mainly types )
