## Context
Taking a slightly different approach to [this
PR](https://github.com/supabase/supabase/pull/43370)
Original problem was that if you opened some folders while in a bucket
and then switched to a different bucket, the folder UI will persists
(folders from Bucket A will render when landing on Bucket B)
## Changes involved
- Shift `StorageExplorerStateContextProvider` into `[bucketId].tsx]`
instead of `ProjectContext`
- The valtio store here only applies for the storage explorer so having
it so high in the project's context was unnecessary
- This also just implies that the valtio store will automatically reset
whenever the bucket changes
- Simplify storage explorer valtio store by initializing the store with
the bucket
- We'll initialize the selected bucket with the store now (Same as
previous PR)
- Removes unnecessary `setSelectedBucket` method which required a
separate `useEffect` in `StorageExplorer.tsx`
## To test
- [ ] Verify that the original is resolved
- [ ] General smoke test of the storage explorer - i've also re-added
the e2e test that Gildas wrote up in his PR
## What kind of change does this PR introduce?
Resolves FDBK-40065
## What is the current behavior?
Using the _Run_ button on either inline or “full” SQL Editor removes
focus from the Monaco editor.
## What is the new behavior?
The text caret remains in the Monaco editor even after button press.
## To test
Try running a few queries both with keyboard (⌘ enter) and mouse click
on the _Run_ button:
```sql
-- 1) Create a mock table + seed data
drop table if exists public.test;
create table public.test (
id bigserial primary key,
name text not null,
created_at timestamptz not null default now()
);
insert into public.test (name)
values
('alpha'),
('beta'),
('gamma'),
('delta');
select * from public.test order by id;
```
```sql
-- 2) Deletion query that should trigger the warning modal (destructive op)
delete from public.test where id <= 2;
-- Verify remaining rows
select * from public.test order by id;
```
## Additional context
- [ ] This was FDBK-40065 that we should reply to
## What kind of change does this PR introduce?
Bug fix: custom icons now respect their source SVG’s stroke/fill and
stroke-width instead of always getting a global stroke.
## What is the current behavior?
Every custom icon gets `stroke="currentColor"` and `strokeWidth={2}`
from `createSupabaseIcon`, so fill-only logos get an extra stroke and
icons designed at `stroke-width="1"` (e.g. postgres, auth) render too
thick. Call sites and the design-system grid needed workarounds
(`strokeWidth={0}`, `FILL_ONLY_ICONS`, `STROKE_WIDTH_FROM_SOURCE`).
## What is the new behavior?
The icon build reads root SVG attributes (`fill`, `stroke`,
`stroke-width`, etc.) and passes them as per-icon defaults. Fill-only
icons (chatgpt, claude, axiom, last9) have `stroke="none"` in source and
render with no stroke; stroke icons keep their source stroke-width (e.g.
postgres at 1). No `strokeWidth={0}` or allowlists needed at call sites
or in the icon grid.
| Before | After |
| --- | --- |
| <img width="1826" height="552" alt="CleanShot 2026-03-05 at 10 53
31@2x-3D63CEF0-1132-44F5-A382-730346432F1E"
src="https://github.com/user-attachments/assets/9caf73fa-351b-4ea2-a420-b3339f934742"
/> | <img width="1814" height="552" alt="CleanShot 2026-03-05 at 10 53
40@2x-514FC8BD-D30F-4D23-B209-0ECD6D13A184"
src="https://github.com/user-attachments/assets/936ea5a5-dae0-413f-8545-b90a502cea69"
/> |
## Additional context
- Added `stroke="none"` to the four fill-only source SVGs;
sentry/grafana/otlp/datadog already had `stroke-width="0"`.
- `createSupabaseIcon` only applies `color`/`strokeWidth` when the
consumer passes them, so `svgDefaults` from the build are used
otherwise.
- Removed workarounds from `icons.tsx` and GuidesSidebar; updated
icons.mdx (defaults + fill-only guidance).
## 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?
fix: small issue when a filter was active and does not found error
popped up, we didn't allow users to reset the value. This handles the
edge cases where a filter is applied and the column is deleted
Bug fix
## What is the current behavior?
The `affected_regions` generated by the AI sometimes have inconsistent
casing, causing validation to fail.
## What is the new behavior?
The system now accepts `affected_regions` that match case-insensitively,
allowing for variations in casing.
JSON column viewer looks fine on dark mode, but barely visible on light
mode, especially when the value is of string type.
There are no colors that can satisfy the 4.5:1 contrast ratio (minimum
requirement for web and digital accessibility) on both white and black
background, so I had to pick different classes for ight mode (dark mode
is fine, I checked the colors against the dark background and the
contrast is good).
Even though the new colors may not look that contrasting to each other,
I prioritized accessibility, and made sure they all complied with the
4.5:1 contrast ratio.
Before:
<img width="499" height="179" alt="112"
src="https://github.com/user-attachments/assets/7f0821cf-6127-4246-a2c2-a78a5d31083a"
/>
After:
<img width="499" height="211" alt="222"
src="https://github.com/user-attachments/assets/0b5b2c35-9211-4a25-a9f5-eef4e5c8c8f3"
/>
Fixes#40633
---------
Co-authored-by: Ali Waseem <waseema393@gmail.com>
### Changes
- Replaces the isPaid plan-based check on the "Prevent use of leaked
passwords" (PASSWORD_HIBP_ENABLED) setting with a proper entitlement
check using the `password_hibp` entitlement key
- Adds a new `useHasEntitlementAccess` hook that returns a reusable
checker function for any entitlement key, backed by the same cached
entitlements query
### Testing
- Head to `/project/_/auth/providers?provider=Email` with an Org on the
Free Plan
- Assert that the "Prevent use of leaked passwords" toggle is disabled.
- Head to `/project/_/auth/providers?provider=Email` with an Org on the
Pro Plan
- Assert that the "Prevent use of leaked passwords" toggle is enabled
and can be toggled and saved.
<img width="612" height="496" alt="image"
src="https://github.com/user-attachments/assets/fc1ccc79-016c-4265-96ac-bdb458d2a8de"
/>
Adds a feature flagged exposed tables config to postgrest settings:
<img width="1158" height="589" alt="Screenshot 2026-03-02 at 17 04 13"
src="https://github.com/user-attachments/assets/8fa8ab81-0bfa-4781-83bc-80ac52e180f8"
/>
To test:
- make sure the existing (feature flag off) settings work as expected
- make sure exposing and removing schemas works in the new mode
- make sure exposing and removing tables works in the new mode
- ideally try with a lot of tables to check search and infinite scroll
work as expected
- try exposing a schema without any tables (like `graphql_public`) and
make sure it doesn't get removed randomly when editing tables
- try with a table with custom permissions (for example `REVOKE SELECT
ON public.posts FROM anon;`) and make sure the user is informed with the
tooltip
---------
Co-authored-by: Nick Babadzhanian <33933459+pgnickb@users.noreply.github.com>
Allows users to select logs and copy their contents for debugging, or
pass them on to assistant with one click.
## To test
- go to logs or log explorer
- select some logs
- try copying as json, markdown, or sending it to assistant.
### Changes
- Replace plan-based checks (`LOG_RETENTION`, `availableIn`) with
`useCheckEntitlements('log.retention_days')` in the chart interval
dropdown
- `ChartIntervalDropdown` now calls the entitlements hook internally
instead of receiving plan props
- Remove `LOG_RETENTION` constant and `availableIn` from
`CHART_INTERVALS`
- Update consumer components (`ProjectUsage`, `ProjectUsageSection`,
`ObservabilityOverview`) to remove plan prop passing
### Testing
- Head to `/project/_/observability` with a Free plan org.
- Click on the time selector "Last 7 days" is disabled, default is "Last
60 minutes", tooltip shows retention limit and upgrade link
<img width="827" height="151" alt="image"
src="https://github.com/user-attachments/assets/20988398-8f0a-46b6-be3d-af4e9d530f81"
/>
- With a Pro Org and above the "Last 7 days" option should be enabled
## 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?
Small docs update to troubleshooting guide. Adds circuit breaker error
to list of related errors that troubleshooting guide addresses.
The "Management of read replicas has moved" banner in Settings >
Infrastructure linked to /database/replication, which returns a 404-like
page for users who don't have the database:replication feature enabled.
The notice is now only shown when the feature is enabled.
## Problem
On Safari, it's impossible to resize a column.
## Solution
This is actually a bug in `react-data-grid` that has been fixed in more
recent versions but we can't use them.
Patch the package
## How to test
- Open the any table in the _Table editor_
- Resize a column
## What kind of change does this PR introduce?
Form handling improvement.
## What is the current behavior?
https://github.com/supabase/supabase/pull/43201/ standardised our
discard changes behaviour with a shared hook and
`DiscardChangesConfirmationDialog` component. But many forms and sheets
still:
1. Don’t have any Discard-confirm close behaviour, making it too easy to
make accidental discards
2. Use a more complicated, manually-created `CloseConfirmationModal`
approach
## What is the new behavior?
- Replaced all instances of `#2` above that had `CloseConfirmationModal`
with `DiscardChangesConfirmationDialog` and its hook
- Improved design system documentation around dirty form dismissal
| Before | After |
| --- | --- |
| <img width="987" height="569" alt="Mercor Apexroles Foo
Supabase-9A40EC7C-F335-4B26-B567-450FC0845463"
src="https://github.com/user-attachments/assets/363bed82-34d2-4cc8-9164-6d18cfdbdbbc"
/> | <img width="987" height="569" alt="Mercor Apexroles Foo
Supabase-F427F1FA-DECC-4194-B663-A9E5A6F285A1"
src="https://github.com/user-attachments/assets/d49fafdc-a5c2-46df-9b67-ec42bacbe716"
/> |
## To test
Try editing values these sheets in staging, then blurring the sheet or
pressing `esc`:
- CreateQueueSheet.tsx
- CronJobsTab.tsx
- CronJobPage.tsx
- EditWrapperSheet.tsx
- OverviewTab.tsx
- WrappersTab.tsx
- CreateFunction/index.tsx
- EditHookPanel.tsx
- TriggerSheet.tsx
- SidePanelEditor.tsx
- EditSecretSheet.tsx
- PolicyEditorModal/index.tsx
- PolicyEditorPanel/index.tsx
## Still to come
- [ ] Incrementally take on `#1`: implement
`DiscardChangesConfirmationDialog` and its hook in sheets or dialog
forms that have no dirty form dismissal handling
---------
Co-authored-by: Joshen Lim <joshenlimek@gmail.com>
## What kind of change does this PR introduce?
- UI fixes and cleanup
## What is the current behavior?
- We use the _Anthropic_ logo for _Claude Code_ in the MCP connector
panel
- We have a bunch of redundant dark mode logos (visually the same as
their light mode siblings)
| Before | After |
| --- | --- |
| <img width="987" height="373" alt="Krosno Toolshed
Supabase-03CD5E6F-53E6-4282-9087-ACB9F67B6DDA"
src="https://github.com/user-attachments/assets/ca7b5d9c-2a2b-4520-8bd4-8cf26c759c73"
/> | <img width="987" height="373" alt="Krosno Toolshed
Supabase-53F3849B-2D86-494B-BE25-8D23EA050243"
src="https://github.com/user-attachments/assets/7bd2e9a6-82ae-48a1-9ef1-591b7a33632e"
/> |
## What is the new behavior?
- We use the _Claude_ logo for _Claude Code_
- Removed duplicative logos
- Made note of which MCP servers have a different logo for dark mode
## What kind of change does this PR introduce?
UI fix
## What is the current behavior?
Radio button circles are off-center in Safari.
## What is the new behavior?
Radio button circles are centered in Safari. Extraneous stroke on the
Lucide Circle icon also removed.
| Before | After |
| --- | --- |
| <img width="1502" height="864" alt="General Supabase"
src="https://github.com/user-attachments/assets/7fea0cc5-d62f-4c29-b341-494e66b91c38"
/> | <img width="1502" height="864" alt="General Supabase"
src="https://github.com/user-attachments/assets/296cbb72-ced2-44f7-8c73-d2c758a66507"
/> |
It’s hard to see the before-after via screenshots. Go to org settings on
`master` in Safari to see the difference.
## Additional context
We already applied this fix in `radio-group-stacked`.
### Changes
- Remove maybeShowUpgradePrompt (plan-based) in favor of
maybeShowUpgradePromptIfNotEntitled (entitlement-based)
- Replaced the `maybeShowUpgradePrompt` with a `hasAccessToHelper`
callback to default the date selection to a safe default so pages don't
load with an out-of-range date.
### Testing
- With an Org on the Free Plan, head to
`/project/_/observability/api-overview?its=2026-03-04T15:06:50.704Z&ite=2026-03-20T15:06:50.704Z`.
Notice that the end date in the URL is 16 days after the start date.
- Assert that the actual range defaults to the entitled range of 1 day.
<img width="467" height="483" alt="image"
src="https://github.com/user-attachments/assets/e17c6957-3ece-4956-87aa-30cf62c84acd"
/>
## 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?
- Allow selecting the column, it respects different types and updates
the values correctly. If the values are different it gets wiped due to
mismatches
- Updated E2E tests and unit tests
Hey, I can't help that my name is alphabetically the first — in school
this meant I always had to give presentations first. Not fun!
## 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?
On-boarding addition of my name (Aaron Brooks) to humans.txt
## What is the current behavior?
Aaron Brooks is not part of humans.txt
## What is the new behavior?
Aaron Brooks is part of humans.txt
## 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?
This adds a component showing a list of AI curated related threads to
the detail thread view.
## What is the current behavior?
This is not available.
## What is the new behavior?
A new component on the thread view.
## Additional context
Add any other context or screenshots.
---------
Co-authored-by: Danny White <3104761+dnywh@users.noreply.github.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
Adds the new `dedicated_pooler` entitlement and it's respective
entitlement checks in the following components: `ConnectTabContent`,
`ConnectStepsSection`, `LogsSidebarMenuV2` and `ConnectionPooling`
## Testing
### ConnectTabContent and ConnectStepsSection
- Head to `project/_?showConnect=true` with an Org on the Pro Plan or
above
- Assert that the `Dedicated Pooler` option is available.
<img width="1007" height="641" alt="image"
src="https://github.com/user-attachments/assets/b4891544-b84b-4745-9e25-4cbc7c76686c"
/>
### IPv4SidePanel
- Head to `/project/_/database/settings` with an Org on the Pro Plan and
above and assert that the Dedicated Pooler is available
<img width="889" height="562" alt="image"
src="https://github.com/user-attachments/assets/3150cb3d-18e9-4b34-bc9b-2589d0a33c5f"
/>
### Dedicated Pooler Logs
- Head to `/project/_/logs/dedicated-pooler-logs` with an Org on the Pro
Plan and assert that you are in the `dedicated-poolers-logs` page.
- Head to `/project/_/logs/dedicated-pooler-logs` with an Org on the
Free Plan and assert that you are redirected to `/logs/pooler-logs`
### Logs Sidebar
- Head to `/project/jdvjfujajfyywbaaakje/logs/explorer` with an Org on
the Free Plan
- Assert that only the "Poolers" collection is shown
- Head to `/project/jdvjfujajfyywbaaakje/logs/explorer` with an Org on
the Pro Plan
- Assert that the Dedicated and Shared Poolers collections are shown
<img width="251" height="774" alt="image"
src="https://github.com/user-attachments/assets/747bceee-d5e7-4a8f-911d-6b02cdb115eb"
/>
### Changes
- Replace hardcoded plan-to-days mapping (`getAvailableInForDays`) with
entitlement-driven checks.
- Show a lock icon on date picker helpers that exceed the user's
entitled log retention, instead of a plan name badge.
- Remove availableIn from DatetimeHelper type — the entitlement API is
now the single source of truth.
### Testing
- Head to `/project/_/logs` on a free plan project and verify lock icons
appear on helpers beyond 1 day
- Verify clicking a locked helper still triggers the upgrade prompt
<img width="511" height="483" alt="image"
src="https://github.com/user-attachments/assets/41dce8c0-205a-4d82-9c07-12f4bb6b0f4d"
/>
I'm relatively new to supabase, but in my current understanding the SQL
selects are inconsistent with the given example schema as the column is
named `room_topic` there.
## 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?
Updates documentation.
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
## Release Notes
* **Documentation**
* Corrected authorization policy examples in realtime guides to use
accurate column references for topic-based access control in Broadcast
and Presence features.
<sub>✏️ Tip: You can customize this high-level summary in your review
settings.</sub>
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
Co-authored-by: Chris Chinchilla <chris.ward@supabase.io>
- Filter out never-run queries early: previously the queries didn't
check if a statement had ever actually been executed, so PostgreSQL
would process and return those useless rows. Adding WHERE calls > 0
skips
them upfront.
- Replace window functions with a pre-computed total: two queries were
using a SQL window function to calculate the "% of total time" column,
which forces PostgreSQL to load every single row into memory before
returning results. Replaced with a small CTE that computes the total
once separately, so PostgreSQL can apply the row limit without buffering
everything first. This was the main cause of page crashes on databases
with a large query history.
- slowQueriesCount bug fix: added missing as statements alias
- queryMetrics WHERE clause fix: ${where} was producing WHERE ... WHERE
..., now uses .replace(/^WHERE/, 'AND')
## 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?
Update Log Drains feature page
## What is the current behavior?
The Log Drains feature page (`/features/log-drains`) reflects outdated
information:
- Status shows Public Alpha
- Description only mentions Datadog and custom HTTP endpoints
- Docs link points to the original blog post (`/blog/log-drains`)
- No mention of Pro plan availability
## What is the new behavior?
- Status updated to GA
- Description updated to reflect Pro plan availability (previously
Team/Enterprise only)
- All supported destinations listed: Datadog, Grafana Loki, Sentry, AWS
S3, Axiom, and generic HTTP
- Pricing included: $60/drain/project + $0.20/M events + $0.09/GB egress
- Docs link updated to `/docs/guides/telemetry/log-drains`
## Additional context
Add any other context or screenshots.
Co-authored-by: Ana Mogul <ana1337x@users.noreply.github.com>
## Problem
When testing things in SQL Editor, you often have to impersonate the
same user many times. It's cumbersome to look for them. We have a
_Recents_ list but its order is not practical.
## Solution
Invert the order of the _Recents_ list so that the last impersonated
user is the first one
## How to test
- Create multiple users
- Open the SQL Editor
- Impersonate different users
- Check the order of the list
## 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?
- For the new filters experience, we are missing null changes for dates
and also no support for isNull
- Updated E2E tests to account for new null cases
## Problem
When navigating between buckets after having selected a folder, the
storage view shows the previous folder content.
## Solution
Set the selected bucket before navigating to it, preventing a race
condition with Valtio updates and the effect that loads the bucket
content.
## How to test
1. Create 2 buckets
2. In one of them, create a folder and upload some files
3. Go back to the bucket list
4. Open the other bucket
Results:
- It should not show the previous bucket content anymore
- It should not show an empty 2nd column either
---------
Co-authored-by: Joshen Lim <joshenlimek@gmail.com>
This pull request makes a minor adjustment to the temporary API key
validation logic. The key is now considered invalid if it has less than
30 seconds remaining before expiry, instead of the previous 20 seconds.
This change helps avoid edge cases where a key might expire during use.
Joshen Lim
