## What kind of change does this PR introduce?
Minor polish
## What is the current behavior?
We have a privacy policy notice that uses `Alert_Shadcn_` with bespoke
links and badge.
## What is the new behavior?
- Reuse `Admonition` and `InlineLink` components
- Use sentence case as per our copywriting guidelines
- Tighten copywriting
| Before | After |
| --- | --- |
| <img width="1024" height="563" alt="Projects Boring Org Supabase"
src="https://github.com/user-attachments/assets/43fd9d4f-21e2-461b-92e3-d75578ae2ddd"
/> | <img width="1024" height="563" alt="Projects Boring Org
Supabase-27BBA2B8-4F90-4FB5-852D-B92B38AF5055"
src="https://github.com/user-attachments/assets/c0c09aae-c110-46cb-86ad-f6df557e1a4f"
/> |
## Additional context
Legal should sign off on this too.
## Context
Adds a "source" param in the new Connect sheet to allow selecting read
replicas - this was present in the old Connect dialog
<img width="962" height="713" alt="image"
src="https://github.com/user-attachments/assets/3c805e41-e8df-431b-83d0-e5f227d68475"
/>
Selecting sources will update the connect steps at the bottom -
connection string, host etc
## I have read the
[CONTRIBUTING.md](https://github.com/supabase/supabase/blob/master/CONTRIBUTING.md)
file.
YES
## What kind of change does this PR introduce?
We have users reporting random four oh fours in the dashboard. We can't
reproduce locally, but this is pointing to an next js issue.
The reason is the users were getting appended a double /dashboard route.
it all points to 80852 in next js, double base path. We added app dir
recently as well
https://github.com/vercel/next.js/issues/80852
Adds `chatId` (the chat UUID) to the three assistant PostHog events so
we can count distinct chat sessions in PostHog. This naming is
consistent with how we name the field in Braintrust logs.
Also migrates from the deprecated `useSendEventMutation` hook to
`useTrack` — groups (project/org) are still sent automatically by
`useTrack`.
Closes AI-519
### Summary
This PR adds support for Isle of Man customers to use GB VAT tax IDs.
Isle of Man uses the UK VAT system, so Orb requires country: 'GB' for
validation - but the billing address country is IM.
- Adds an IM VAT entry to `TAX_IDS` with a new `taxCountryIso2` override
field, so the UI shows "IM VAT" while sending country: 'GB' to Orb
- Introduces `getEffectiveTaxCountry()` to resolve the correct country
code for Orb
- Introduces `resolveStoredTaxId()` to map stored Orb data back to the
correct UI entry, using billing address country to disambiguate IM vs GB
- Updates all three billing forms to use these utilities: Billing
Settings page, Add Payment Method modal and Update Billing Address modal
### Testing
#### Automated tests
I've added unit tests in this PR
#### Manual tests
You can use the following country/city/postal code combinations to test:
- Isle of Man, Douglas, IM1 2LE
- United Kingdom, Crumlin, BT29 4AA
**Billing Settings page**
- Select Isle of Man as billing country → tax ID dropdown shows "Isle of
Man - IM VAT"
- Select United Kingdom as billing country → tax ID dropdown shows
"United Kingdom - UK VAT" / "United Kingdom - GB VAT" (not "IM VAT")
- Select IM VAT, enter a valid GB VAT number, save → check network tab,
Orb receives country: 'GB'
- Reload page for an IM customer with existing tax ID → "IM VAT" is
pre-selected (not "UK VAT")
- Reload page for a GB customer with existing tax ID → "UK VAT" is
pre-selected (no regression)
- Change country from IM to GB → available tax IDs update, IM VAT no
longer shown
**Add Payment Method modal**
- Open modal for an IM customer with existing tax ID → "IM VAT" is
pre-selected
- Check "I'm purchasing as a business", select Isle of Man in Stripe
address → tax ID dropdown shows "IM VAT"
- Check "I'm purchasing as a business", select United Kingdom in Stripe
address → tax ID dropdown shows "UK VAT" / "GB VAT"
- Submit with IM VAT → check network tab, tax ID is saved with country:
'GB'
- Open modal for a GB customer → no regression, "UK VAT" shown as before
**Update Billing Address modal**
Appears for paid orgs missing a billing address
- Open modal for an IM customer with existing tax ID → "IM VAT" is
pre-selected
- Open modal for a GB customer with existing tax ID → "UK VAT" is
pre-selected
- Fill in IM address + IM VAT and save → saves correctly, modal closes
---------
Co-authored-by: Ignacio Dobronich <ignacio@dobronich.com>
## I have read the
[CONTRIBUTING.md](https://github.com/supabase/supabase/blob/master/CONTRIBUTING.md)
file.
YES
## What kind of change does this PR introduce?
show the "enable" button if the `GET /auth/v1/admin/custom-providers`
returns 404 (without a json body). this is temporary as I want to enable
the rollout of this feature now, and there could be projects which
didn't get the latest auth server release(should be completed by next
week).
## What is the current behavior?
We only show a generic error message.
## What is the new behavior?
Show custom provider specific message and add CTA to enable the custom
providers (simply trigger a PATCH config request to trigger auth config
updates)
## Additional context
<img width="1454" height="390" alt="image-IPleA6ze@2x"
src="https://github.com/user-attachments/assets/b0cb4df8-2499-4749-900c-b78543a72800"
/>
This error is shown after 2 retries of the `GET
/auth/v1/admin/custom-providers`, would be nice to have if we show this
error message without retries.
## Context
Table Editor - hide "Is unique" configuration for a column that's a
primary key - unnecessary to configure a column as unique if it's
already a primary key (redundant index is created)
## Changes involved
- TableEditor side panel: No longer shows is unique option under
configuration popover if column is a PK
- ColumnEditor side panel: Disable is unique + is nullable option if
column is a PK (Added tooltip)
<img width="810" height="151" alt="image"
src="https://github.com/user-attachments/assets/0f877706-1a56-44b7-864e-ba30efbb67f7"
/>
Also chucked in some minor refactors:
- Refactored some `any` types
- Refactored Column to use the latest Checkbox component
## Problem
The edge function logs table at
`/observability/logs/edge-functions-logs` did not show which edge
function was called. Users had to open individual log entries to find
the function name, even though the information is available in the log
data.
## Fix
- Added `request.pathname` to the SQL SELECT query for
`function_edge_logs` so the pathname is returned with each row.
- Added `extractEdgeFunctionName` utility that parses the last path
segment from the pathname (e.g. `/functions/v1/hello-world-1` becomes
`hello-world-1`).
- Updated `FunctionsEdgeColumnRender` to display the function name
between the HTTP method and the log ID columns.
- Added unit tests for `extractEdgeFunctionName` covering normal paths,
null/undefined, non-string values, and pathnames without slashes.
<img width="1472" height="776" alt="CleanShot 2026-03-19 at 13 18 47@2x"
src="https://github.com/user-attachments/assets/e437f7df-7aab-4ea5-b421-95bab3923605"
/>
## How to test
- Navigate to Logs and Analytics, then Edge Functions logs.
- Confirm each row now shows the function name (e.g. `hello-world-1`)
between the HTTP method and the log ID.
- Confirm rows without a pathname (e.g. non-platform or self-hosted)
still render without errors.
- Confirm clicking a row still opens the log detail panel correctly.
---------
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
## Problem
We want `ui-patterns` to contain only visual components but
https://github.com/supabase/supabase/pull/43577 introduced a component
with logic about navigation
## Solution
Extract the logic part into a hook and use it in the component.
Next step will be to remove this component and use
`<DiscardChangesConfirmationDialog>` directly along the hook where
needed
## How to test
Check that the fixes from #43577 still work:
On staging, for each case:
- Authentication email templates
- Observability reports
- Edge functions creation
- Edge functions edition
Test:
- Modify the template/report/function
- Navigate away using either the sidebar link, the browser back button
or closing the tab
- Cancel navigation in the confirmation dialog
- Navigation should be prevented and you should not loose your changes
Test:
- Modify the template/report/function
- Navigate away using either the sidebar link, the browser back button
or closing the tab
- Confirm navigation in the confirmation dialog
- Navigation should not be prevented and you should have lost your
changes
## Problem
The database observability dashboard flattens critical spikes when
viewing the 7-day window. A spike clearly visible at the 3-hour view
(2-minute resolution) disappears completely at the 7-day view because
the Prometheus query step was 1 hour, averaging each spike into a
60-minute bucket and reducing its apparent magnitude to near zero.
## Fix
Increase the granularity in the 7 day window to prevent spikes from
being flattened.
Data point counts: 7-day view goes from ~168 to ~336 points per series,
more consistent with the existing 12-hour view (~360 points).
## How to test
- Open the observability database dashboard for a project that has had a
recent CPU, RAM, or connection spike
- Select the "Last 3 hours" time window and confirm the spike is visible
- Switch to "Last 7 days" and confirm the same spike is now visible
(previously it would disappear)
- Confirm all other time windows (10m, 30m, 1h, 24h) still load and
display correctly
---------
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
## Context
If you've got a queued operation on a table, then switch tables, the
dashboard stalls for a few seconds before finally rendering the selected
table.
This was caused by an infinite loop in a useEffect calling
`reapplyOptimisticUpdates` due to `dataUpdatedAt` being in the
dependency array (which comes from react query)
Removing `dataUpdatedAt` would resolve the issue, but then it won't
solve the following comment in the code:
`// This ensures pending changes remain visible when switching tabs or
after data refresh`
## Changes involved
Opting to refactor the logic for rendering data in the Grid instead
- Avoid manipulating the data directly from the query client
- Manual changes to the query client are hard to track when debugging
- Ideally the data in the query client are exactly what is coming from
the API
- Instead just compute the row data if there's operations applied, and
render that in the grid
```
const baseRows = data?.rows ?? EMPTY_ARR
const rows = formatGridDataWithOperationValues({ operations, rows:
baseRows })
```
- Simplifies operations logic + data in react query remains as the
source of truth
- This also improves perceived performance, as previously we'd need to
invalidate the query client if we're removing any operation.
- A lot more apparent with the undo operation introduced
[here](https://github.com/supabase/supabase/pull/43957)
- Whereas now, we'll just revert back to whatever's in the query client
(and still do the invalidation behind the scenes) so things feel faster
Upgrades `ai` from v5 to v6 and all related packages.
**Package bumps:**
- `ai`: `5.0.52` → `^6.0.116`
- `@ai-sdk/openai`: `2.0.32` → `^3.0.41`
- `@ai-sdk/react`: `2.0.52` → `^3.0.118`
- `@ai-sdk/provider`: `^2.0.0` → `^3.0.8`
- `@ai-sdk/provider-utils`: `^3.0.0` → `^4.0.19`
- `@ai-sdk/amazon-bedrock`: `^3.0.0` → `^4.0.81`
- `@ai-sdk/mcp`: N/A → `^1.0.25`
- `openai`: bumped to `^4.104.0`
- `braintrust`: `3.0.x` → `^3.4.0`
**Breaking change migrations:**
- `generateObject` removed in v6 — migrated 5 API routes to
`generateText` with `Output.object({ schema })`, returning
`result.output`
- `convertToModelMessages` is now async — added `await`
- MCP import path changed: `experimental_createMCPClient` from `ai` →
`createMCPClient` from `@ai-sdk/mcp`
- `openai()` defaults to Responses API — added `store: false` to
provider options for ZDR org compatibility
**Streaming fix:**
Added `Content-Encoding: none` header to `pipeUIMessageStreamToResponse`
calls. Without it, proxy middleware buffers the entire SSE response
before flushing, causing the full reply to appear at once.
**Zero Data Retention fix:**
In recent AI SDK versions, `openai()` default to Responses API instead
of the legacy chat completions API. This produces a 404 from OpenAI with
message `"Items are not persisted for Zero Data Retention organizations.
Remove this item from your input and try again."` The Responses API is
OpenAI's [recommended
endpoint](https://developers.openai.com/api/docs/guides/migrate-to-responses).
This PR adds `store: false` as mentioned in
https://github.com/vercel/ai/issues/10060 to avoid incompatible
persistence attempts.
**References:**
- https://ai-sdk.dev/docs/migration-guides/migration-guide-6-0
-
https://ai-sdk.dev/docs/troubleshooting/streaming-not-working-when-proxied
- https://github.com/vercel/ai/issues/10060
Closes AI-514
Related AI-509
## Context
Shifting more dashboard queries into pg-meta so that we centralize all
manually written queries in one place
Having them in packages/pg-meta also allows us to write tests for them
## To test
Just needs a smoke test on
- Table Editor
- Fetching entities
- Viewing definition
- SQL Editor
- View ongoing queries
- Abort queries
- Integrations
- Queues
- Database
- Migrations
-Triggers (Updating)
## I have read the
[CONTRIBUTING.md](https://github.com/supabase/supabase/blob/master/CONTRIBUTING.md)
file.
YES
## What kind of change does this PR introduce?
Refactor
## What is the current behavior?
The `/incident-banner` endpoint is implemented using the Pages Router.
## What is the new behavior?
The `/incident-banner` endpoint is moved to the App Router, enabling
caching of the upstream fetch. This does not turn on the querying from
the frontend yet, making that a separate PR so we can revert easily if
needed.
## Additional context
Fixes#42089
Supersedes #43352 and
https://github.com/supabase/postgres-meta/pull/1049
## Problem
When creating a table in the Dashboard, if a column-specific error
occurs (e.g., invalid enum value or broken foreign key), the Dashboard
displays an error toast, but the side panel remains open. In the
background, the table is actually created (partially), leading to a
_42P07: relation already exists_ error if the user tries to click _Save_
again.
## Solution
We actually had nested transactions which is not supported by Postgres.
- [x] Allow generating SQL without transactions from `@supabase/pg-meta`
- [x] Fix the table creation
- [x] Fix some accessibility issues
- [x] Add tests
## How to test
1. Click _New table_ in the Table editor
2. Add a column with an _int8_ type and set its default value to
`bazinga`
3. Click _Save_
You should see an error.
4. Fix the default value by setting it to `10`
5. Click _Save_
Results:
- You should see a success message about the table creation
- The table should have the column with the correct default value
<img width="1543" height="981" alt="image"
src="https://github.com/user-attachments/assets/8e6c37c5-15c7-4bc4-be87-477e36a62a65"
/>
Switches from two to one column layout for details page and removes an
inconsistent background colour in header. Moves critical detail
information from page into header (url and deploy dates/counts)
## What kind of change does this PR introduce?
Bug fix. Resolves FE-2802. Fixes#43882.
## What is the current behavior?
Second-level foreign key side panels in Table Editor can be dismissed
while dirty without a discard confirmation. That makes accidental
dismissal easy and is inconsistent with the shared dirty-form dismissal
pattern from FE-2802.
## What is the new behavior?
- Adds `DiscardChangesConfirmationDialog` handling to
`ForeignKeySelector`
- Guards all selector close paths behind a local dirty-state check
- Normalizes foreign key state so derived type metadata does not create
false dirty prompts
- Clarifies in `modality.mdx` that route-driven/page-unload flows should
use a navigation guard rather than the dialog/sheet close guard
| Dirty State | Dialog |
| --- | --- |
| <img width="1502" height="899" alt="test Table Editor AWS Healthy
Toolshed Supabase-D1EA9D5F-46EB-4DB5-830D-7CEF36A55B7E"
src="https://github.com/user-attachments/assets/3d866850-d062-4fb7-9949-2fd713e6088f"
/> | <img width="1502" height="899" alt="test Table Editor AWS Healthy
Toolshed Supabase"
src="https://github.com/user-attachments/assets/1396361d-1d44-4206-862f-3379ea22be30"
/> |
## Additional context
Includes a focused unit test for foreign key dirty-state normalization.
## What kind of change does this PR introduce?
Bug fix.
## What is the current behavior?
Webhook endpoint validation is inconsistent across Studio forms. The
webhook sheet accepts incomplete hostnames like `https://webhook`, event
type validation is not surfaced clearly, and HTTP endpoint validation
differs between webhooks, log drains, cron jobs, and database hooks.
## What is the new behavior?
- Tightens webhook endpoint URL validation and rejects incomplete
hostnames while still allowing localhost and IP-based endpoints.
- Surfaces the Event types validation through the standard form error
styling and highlights the existing accordion item border when invalid.
- **Extracts a shared HTTP endpoint URL validator and reuses it in
webhooks, log drains, cron jobs, and database hooks.**
- Adds focused regression tests for the webhook sheet and the
shared/consumer validation paths.
| After |
| --- |
| <img width="1728" height="997" alt="Webhooks Settings AWS Healthy
Toolshed Supabase-CB0D999C-D0BF-47AA-A10F-342A2E328DF9"
src="https://github.com/user-attachments/assets/bcbe4876-f9a7-497a-b288-460087a65546"
/> |
## To test
Form behaviour (in particular URL validation) on:
- Webhook endpoint
- Log drains
- Cron jobs
- Database hooks
## Problem
PR #43704 added the `tableEditorApiAccessToggle` flag to revoke default
privileges on project creation, but there's no PostHog instrumentation
to measure the rollout or understand how users respond to the new
behavior.
## Changes
**1. Flag state on project creation**
(`project_creation_simple_version_submitted`)
Adds `tableEditorApiAccessToggleEnabled` property — reads the raw
PostHog flag via `usePHFlag` rather than going through
`useDataApiGrantTogglesEnabled`, which coerces `undefined` (flag still
loading) to `false` and would silently poison the data. When the flag
hasn't loaded, the property is omitted from the event entirely.
**2. Per-table toggle tracking** (`table_api_access_toggle_clicked`)
New event that fires when a user flips the Data API access switch in the
table editor side panel (new tables only — existing tables link out to
settings). Tracks `newState: 'enabled' | 'disabled'` and `schemaName` so
we can distinguish public-schema toggles from custom schemas.
Together these let us answer: what % of projects have the new defaults,
and how do those users respond (do they opt tables back in, and in which
schemas)?
Part of the Secure by Default initiative (#43704)
## Context
Similar to how we added a banner for the new table filter bar - mainly
to improve visibility for this feature preview as we'd like some
feedback from usage
<img width="313" height="314" alt="image"
src="https://github.com/user-attachments/assets/4cea6383-483a-42cf-b4a5-bb8c3918c00d"
/>
## Other changes
- Also adding some typing to the `addBanner` and `dismissBanner`
- Removed banner for table editor filter bar
## What kind of change does this PR introduce?
Chore that resolves DEPR-399.
## What is the current behavior?
The cron jobs table in Studio uses a fixed-width `Name` column, which
can truncate long job titles and make them hard to read on
`/project/:ref/integrations/cron/jobs`.
## What is the new behavior?
The cron jobs table now allows resizing the informational columns,
including `Name`, `Schedule`, `Last run`, `Next run`, and `Command`,
while keeping the `Active` and `Actions` columns fixed-width.
This also gives the `Name` column a sensible minimum width so it can be
resized without collapsing.
## Additional context
Adds a small regression test around the cron jobs column configuration
to verify that:
- informational columns are resizable
- utility columns remain fixed
- the `Name` column has a non-zero minimum width
## I have read the
[CONTRIBUTING.md](https://github.com/supabase/supabase/blob/master/CONTRIBUTING.md)
file.
YES
## What kind of change does this PR introduce?
- refactor navigation utils to pass in is_platform
- write unit tests for configuration returns
- default is always passed from constant IS_PLATFORM
## Context
Adds a source filter for the query performance advisor so you can filter
out queries from the dashboard, or not from the dashboard
<img width="309" height="217" alt="image"
src="https://github.com/user-attachments/assets/c1fab9af-e57e-482f-afdb-d77a6600edb3"
/>
For transparency how this works:
- Queries fired via the dashboard through the /query endpoint get
enriched with metadata from the API to include a comment like `--
source: dashboard`
- That's mainly how this filter works atm, to check if this comment
exists if the source "Dashboard" is selected, and the inverse if the
source "Non dashboard" is selected
## I have read the
[CONTRIBUTING.md](https://github.com/supabase/supabase/blob/master/CONTRIBUTING.md)
file.
YES
## What kind of change does this PR introduce?
Refactor
## What is the current behavior?
`/incident-status` is handled via Pages Router.
## What is the new behavior?
`/incident-status` is handled via App Router, enabling use of Vercel
Data Cache to cache the upstream fetch.
## Additional context
Adding the first App Router route handler triggered `next typegen` (run
as `pretypecheck`) to generate `.next/dev/types/validator.ts`, which
imports all route files and expanded the type-checked graph. This
surfaced pre-existing `null`-safety errors in:
- `components/grid/SupabaseGrid.utils.ts` — `useSearchParams()` result
- `components/layouts/ProjectLayout/UpgradingState/index.tsx` —
`useSearchParams()` result
- `pages/project/[ref]/sql/quickstarts.tsx` — `useParams()` result
- `pages/project/[ref]/sql/templates.tsx` — `useParams()` result
These are fixed with optional chaining. The `tsconfig.json` change
(adding `.next/dev/types/**/*.ts` to `include`) is auto-generated by
Next.js and committed as correct behavior.
This PR:
* Adds an upgrade flow to the stripe sync engine, allowing users to
upgrade to the latest version when it becomes available.
* When a new version of sync engine becomes available, users will see an
upgrade button instead of install button.
* Bumps `supabase-management-js` to version 2.0.2 and
`stripe-experiment-sync` to version 1.0.27.
* Uses `parseSchemaComment` and related logic from the
`stripe-experiment-sync` package in order to avoid writing duplicate
code in supabase ui.
* Allows installation/uninstallation to timeout after 5 minutes to avoid
these operations from getting stuck in case an error occurs in their
processing. This allows users to retry the operation, as opposed to the
older behaviour where the users always see a spinner on the
install/uninstall button and couldn't do anything.
* Remove the SSL enforcement admonition as it is no longer required.
Sync engine can now be installed with or without SSL enforcement
enabled.
---------
Co-authored-by: Joshen Lim <joshenlimek@gmail.com>
## Context
Resolves FE-2794
Just addressing some friction points i ran into when setting up MFA for
my account under account preferences
## Changes involved
- Hitting enter at each step of setting up MFA will submit and proceed
(Previously didn't, had to click buttons)
- When logging in with MFA, automatically submit after entering 6 digits
- Refactored to use react hook form for `AddNewFactorModal` in
`FirstStep` and `SecondStep` + simplified the logic a little
When the `tableEditorApiAccessToggle` feature flag is enabled, project
creation now appends SQL to revoke default privileges for `anon`,
`authenticated`, and `service_role` on the `public` schema. This runs
after the base image init script's default grants. This is temporary
while we're still using a feature flag. Eventually it'll be moved into
the base image.
Applies to both the main project creation flow and the Vercel deploy
button flow.
Part of the "Secure by Default" initiative – new projects created under
this flag won't automatically expose tables/functions/sequences to the
Data API via default privileges. Users can still opt in at a table
level.
## Notes
Reusing the existing `useDataApiGrantTogglesEnabled()` flag here rather
than creating a new one – it's the same feature surface area and avoids
unnecessary flag proliferation.
## To test
1. **With flag enabled:**
- Enable the `tableEditorApiAccessToggle` flag in PostHog for your user
- Create a new project via the dashboard
- Create a new table
- Confirm in `/project/_/integrations/data_api/settings` that the new
table is not exposed by default
2. **With flag disabled:**
- Disable the flag (or use a different user without it)
- Create a new project
- Verify default privileges are intact and tables are accessible via the
Data API as usual
3. **With RLS event trigger enabled too:**
- Enable both the feature flag and the "enable RLS event trigger"
checkbox during project creation
- Verify both SQL statements run correctly on the new project
---------
Co-authored-by: Joshen Lim <joshenlimek@gmail.com>
Co-authored-by: Alaister Young <10985857+alaister@users.noreply.github.com>
## What kind of change does this PR introduce?
Feature / UX update.
## What is the current behavior?
Platform webhook deliveries cannot be retried from the deliveries table
or the delivery details sheet.
## What is the new behavior?
- Adds a retry action to individual platform webhook deliveries in the
endpoint deliveries table
- Adds a retry action to the shared delivery details sheet footer
- Limits retry to non-success deliveries
- Applies the table action hit area guidance from the design system
Successful deliveries should also not expose retry in this mock flow.
## Additional context
- Addresses DEPR-340.
- In the current mock implementation, queueing updates the existing
delivery in place: it sets the status to `pending`, clears
`responseCode`, and refreshes `attemptAt` to the current timestamp.
- The mock does not create a separate queued job or append a new
delivery record yet; the same delivery is resurfaced at the top of the
list because deliveries are sorted by latest attempt time.
Joshen Lim