mirror of
https://github.com/supabase/supabase.git
synced 2026-10-07 18:35:07 +03:00
chore/function-recent-errors
38771
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
|
dc16371a47 |
feat(studio): configure BigQuery table layout (#49535)
## What kind of change does this PR introduce? Feature. The last remaining piece of this PR's original scope, rebased onto current `master`. ## What is the current behavior? A BigQuery pipeline replicates every published table into a flat destination table. There is no way to say how those tables should be laid out in BigQuery, so partitioning and clustering have to be applied by hand after the fact, and are lost whenever a destination table is reset. The rest of this PR's original scope has since merged separately: #49841, #49842, #49843, #49844 and #49845. The branch now carries only the table layout work, and adds nothing to `packages/api-types`. It consumes the `table_options` contract that #49844 already brought in. ## What is the new behavior? The BigQuery destination form gains a "Table layout" section under Advanced settings. Every table in the selected publication appears as a row, and expanding one reveals optional partitioning and clustering: - Time-column partitioning, by hour, day, month or year - Integer-range partitioning, with start, end and interval - Ingestion-time partitioning - Clustering, up to four columns Rows are always present, so nothing implies a table can be excluded here. A collapsed row summarises what is applied: `Not configured`, `Daily by created_at`, `Integer range by id`, plus clustering counts. `Not configured` is dimmed a step further so configured rows are what the eye lands on in a long publication. Notes on behavior: - Partition columns are resolved from the published table's real columns, filtered to the types BigQuery accepts for each partition mode. - A row you expand but leave empty is dropped from the payload rather than saved half-configured. Choosing a partition mode without a column is a validation error, not a silent drop. - Clear returns a row to `Not configured` and keeps it in the list. Remove is reserved for stale configuration whose table has left the publication. - Updates send `null` to clear previously stored table options, since omitting the property leaves the stored value unchanged. - Layout applies when a destination table is first created or reset, matching the backend. Roughly 40% of the diff is tests. The bulk of the rest is four new files under `DestinationForm/BigQuery`, which are all one feature. The three commits are readable in order: types and payload builders, then the columns query, then the UI. ## To test Open a project's Database > Replication, then create or edit a BigQuery pipeline. 1. Expand Advanced settings. "Table layout" lists every table in the selected publication. 2. Expand a row, set Partition by to Time column and pick a column. Collapse. The row reads `Daily by <column>`. Hit Clear. The row returns to `Not configured` and stays in the list. 3. Set Partition by to Time column, leave Partition column empty, collapse, and Save. The row explains "Select a partition column" in red rather than saving and silently dropping it. 4. Expand a row and add clustering columns. The fifth is refused. 5. Switch publications and watch the loading state. The row list should not jump size when it resolves. 6. Narrow the sheet. The integer range Start, End and Interval fields should reflow rather than stay in three columns. <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * BigQuery replication destinations now support per-table layout settings. * Configure partitioning by time column, integer range, or ingestion time. * Configure up to four clustering columns per table. * View available columns and validation feedback while editing table layouts. * Table settings are preserved when editing destinations and cleaned up when publications change. * **Bug Fixes** * Improved handling of invalid, unavailable, or removed table and column configurations. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Co-authored-by: Danny White <3104761+dnywh@users.noreply.github.com> Co-authored-by: Joshen Lim <joshenlimek@gmail.com> |
||
|
Joshen Lim
|
957f84b17c |
Joshenlim/fe 4348 find an alternative to the jsr stdpath dependency (#50111)
## Context Drops the `@std/path` dependency which is used in `EdgeFunctions.utils` as `npm.jsr.io` was putting up a Cloudflare bot challenge on some connections which blocks `pnpm install`. Instead, opting to directly port the exact required methods as self-contained functions. Also added some unit tests to check that UI behaviour remains status quo. ## To test: Important to test that everything in the edge functions UI remains status quo - [ ] Open an existing edge function with a single root-level file - should load as expected <img width="310" height="176" alt="image" src="https://github.com/user-attachments/assets/4d724ab0-33bb-4093-a574-52984b2743fd" /> - [ ] Open (or create) an edge function with nested folders - confirm file paths in the editor are shown correctly - Can create nested folders by using `../` as such <img width="319" height="228" alt="image" src="https://github.com/user-attachments/assets/dd345b1f-7c45-47e9-975c-f2f2e53a0106" /> - [ ] Similarly, download the edge function as ZIP to verify that the nested folders are all correctly located - [ ] Open a function with `import_map.json` - confirm still detected as import map through the network tab GET request for the edge function code (Examples here with and without import map) <img width="333" height="245" alt="image" src="https://github.com/user-attachments/assets/90fc90e3-4a62-493b-9246-ed7e3b662e96" /> <img width="290" height="237" alt="image" src="https://github.com/user-attachments/assets/2f692362-c9d4-4394-bc5a-4f84ab5fb6f1" /> - [ ] Deploy a new function via the editor - [ ] Update an existing function via the editor (Test adding new files etc) <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit - **Bug Fixes** - Improved Edge Functions file path handling across supported application environments. - Nested entrypoints, URL-based entrypoints, root-level files, and unmatched paths are now handled consistently. - Generated files retain their content and receive sequential identifiers reliably. - Improved compatibility when processing and displaying files in different application environments. - **Tests** - Added coverage for entrypoint path formatting, relative paths, fallback behavior, unchanged paths, and identifier assignment. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Danny White
|
e351e8a1a4 |
test(studio): scope realtime broadcast helper (#50164)
## What kind of change does this PR introduce? Bug fix ## What is the current behavior? The Realtime Inspector has two buttons named “Broadcast a message” after joining a channel: one in the listening status toolbar and one in the empty grid state. The shared E2E helper matches both and fails with a Playwright strict mode violation. ## What is the new behavior? The helper scopes the action to the listening status region, where the persistent broadcast control lives. ## To test Run the Realtime Inspector E2E tests: ```bash pnpm --prefix e2e/studio run e2e -- features/realtime-inspector.spec.ts <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Tests** * Improved end-to-end test reliability when opening the broadcast message modal. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Danny White
|
bfe07a5937 |
fix(ui-patterns): stabilise multi-select trigger height (#50163)
## What kind of change does this PR introduce?
Bug fix and design-system documentation update.
## What is the current behaviour?
Multi-select examples inconsistently hide the default trigger icon,
which makes the component default difficult to understand from the
design-system page.
The empty trigger label and selected badges also use different content
heights. Selecting or removing the first item causes the default 34px
trigger to grow or shrink slightly.
## What is the new behaviour?
- Shows the default trigger icon in every existing multi-select example
- Adds a dedicated **Without trigger icon** example for
`showIcon={false}`
- Normalises the label and badge content rows to 20px so the default
trigger remains 34px when the first item is selected or removed
- Adds regression coverage for the default trigger height classes
This follows the 34px trigger sizing introduced in
[#48696](https://github.com/supabase/supabase/pull/48696).
## To test
1. Open [Design System > Fragment Components > Multi
Select](https://design-system-git-dnywh-fixmulti-select-icon-ex-772e13-supabase.vercel.app/design-system/docs/fragments/multi-select).
2. Confirm every existing example shows the default trigger icon.
3. Confirm the new **Without trigger icon** example is the only example
without it.
4. In the first example, select and remove the first fruit. The trigger
should remain the same height throughout.
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Added a multi-select example that hides the trigger icon.
* Added documentation and a preview for using multi-select without a
trigger icon.
* **Bug Fixes**
* Improved multi-select badge and label alignment.
* Preserved consistent trigger height when selecting the first item.
* Updated examples to display the trigger icon by default.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
|
||
|
Danny White
|
6a4cbefbcf |
feat(ui): align combobox trigger with select styling (#49665)
## What kind of change does this PR introduce? Bug fix and shared UI primitive. ## What is the current behaviour? In **Database > Replication > Add destination**, the searchable **Publication** field uses a double chevron and button styling. It looks inconsistent beside the **Initial sync** select, which uses a single chevron and the standard select surface. The shared multi-select trigger uses the same inconsistent double chevron. These differences are easy to repeat because searchable comboboxes do not have a dedicated trigger that shares the select styling. ## What is the new behaviour? - Adds `ComboboxTrigger` with the same shared styling as `SelectTrigger` - Uses a fixed single down chevron so combobox and select triggers cannot drift - Migrates the Studio **Database > Replication > Add destination > Publication** field to `ComboboxTrigger` - Updates `MultiSelectorTrigger` to replace `ChevronsUpDown` with the same fixed `ChevronDown`, including **Connect > MCP > Feature groups** in Studio - Updates the canonical design-system combobox examples - Leaves `Button` unchanged, so adopting combobox styling is explicit | Before | After | | --- | --- | | <img width="1256" height="448" alt="CleanShot 2026-09-09 at 10 21 32@2x" src="https://github.com/user-attachments/assets/bb7a6056-0b4e-4cb4-8dce-e92fa4f5afe7" /> | <img width="1256" height="448" alt="CleanShot 2026-09-09 at 10 21 11@2x" src="https://github.com/user-attachments/assets/860a06b9-617d-4d94-ac98-944316e9fdc3" /> | | <img width="1210" height="520" alt="CleanShot 2026-09-09 at 10 58 28@2x" src="https://github.com/user-attachments/assets/e916d3a2-6a75-40eb-965a-56e0694ba447" /> | <img width="1210" height="520" alt="CleanShot 2026-09-09 at 10 57 57@2x" src="https://github.com/user-attachments/assets/c5f1b8ff-8083-47b8-8242-59116d085c00" /> | ## Left for incremental migration These existing Studio comboboxes still use bespoke button triggers and will move to `ComboboxTrigger` in focused follow-up changes: - **Database > Backups > Point in time > Timezone** - **Authentication > Third-party auth > AWS region** - **Account > Preferences > Timezone** - Shared schema and function selectors See https://github.com/supabase/supabase/pull/50157. The organisation and project switchers in the top navigation are intentionally out of scope. They are context switchers rather than form-field comboboxes. ## To test ### Studio callsites On the [Studio preview](https://studio-staging-git-dnywh-featcombobox-trigger-supabase.vercel.app): 1. Open **Database > Replication > Add destination**. 2. Compare **Publication** with **Initial sync**. Both should use a single down chevron and the same closed-field surface, border, sizing, and hover treatment. 3. Open **Publication**, search for an option, and select it. The popover should close and the selected publication should be shown. 4. Open **Connect** from the project header, select **MCP**, and check **Feature groups**. Its multi-select trigger should use a single down chevron. ### Design system 1. Check the [combobox examples](https://design-system-git-dnywh-featcombobox-trigger-supabase.vercel.app/design-system/docs/components/combobox). 2. Check the [multi-select examples](https://design-system-git-dnywh-featcombobox-trigger-supabase.vercel.app/design-system/docs/fragments/multi-select). 3. Both trigger types should use a single down chevron. |
||
|
K-Dog (Kevin)
|
605c6ef065 |
feat: include log ingestion in free plan usage summary (#49576)
<img width="368" height="288" alt="Screenshot 2026-08-26 at 12 24 52 PM" src="https://github.com/user-attachments/assets/ec39a35d-a7fb-4957-8fee-b657125d1c95" /> Gated behind feature flags as rollout is still pending <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Added conditional visibility for the Log Ingestion usage metric based on feature availability. * Updated usage displays and loading states to reflect the available metrics. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Julian Domke
|
21077a4071 |
fix(billing): allow for invoice_pdf to be null (#50138)
The OpenAPI schema guarantees the existence of `invoice.invoice_pdf`, but it can be null in certain cases. This has been uncovered by the recent schema validation initiative :) <img width="435" height="168" alt="Screenshot 2026-09-08 at 16 37 58" src="https://github.com/user-attachments/assets/e9c9a9a2-759f-4ec0-904f-64425b9a65e7" /> <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit - **Bug Fixes** - Invoice downloads now clearly indicate when a PDF is unavailable. - Downloading an invoice without a PDF displays an error notification instead of failing silently. - Invoice download controls are disabled when no PDF exists. - **Accessibility** - Added an accessible label to the receipt download button. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
|
c75e213ade |
[bot] Sync from supabase/troubleshooting (#49906)
This PR syncs the latest troubleshooting guides from the supabase/troubleshooting repository. --------- Co-authored-by: github-docs-bot <github-docs-bot@supabase.com> Co-authored-by: Nik Richers <nrichers@gmail.com> Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> |
||
|
ŁUKASZ KORBASIEWICZ
|
8121df5b0a |
docs: correct Auth rate limit details (#49772)
## 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? Docs update ## What is the current behavior? The Auth rate-limit table contains stale customization statuses and time windows, omits SMS and Web3 limits, and describes the anonymous sign-in burst incorrectly. ## What is the new behavior? - Aligns documented limits with the current Auth, Studio, and Management API behavior - Documents SMS, Web3, and sign-up/sign-in request limits - Corrects verification, token, MFA, email, and anonymous sign-in details - Updates shared rate-limit values and units used by the docs ## Additional context Validation: - Prettier check - Focused MDX lint - Shared-data TypeScript check - All 16 SharedData references resolve <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Updates** * Refined authentication rate limits with clearer per-minute and per-five-minute windows. * Added rate limits for SMS, password reset requests, and Web3 sign-ups and sign-ins. * Updated sign-in, sign-up, verification, token refresh, MFA, and anonymous sign-in limits, including customizable settings where supported. * Clarified email-sending limits and OTP behavior. * **Documentation** * Updated rate-limit reference tables and guidance on request bucket capacity and sustained traffic. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
kemal.earthandClaude Opus 5
|
41e6d489f8 |
feat(studio): mcp elicitation url mode ui (#49954)
## 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? Initial build of the URL mode interstitial. Has complete mock data to be able to view different states (this will be stripped out in the end). A starting point for us to use as an intercept. This is linked with @barryroodt ticket in the MCP project. https://github.com/supabase/mcp/pull/412 <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit - **New Features** - Added an authenticated MCP Secrets page for securely storing API secrets requested by AI tools. - Added provider-aware guidance for OpenAI, Anthropic, Resend, and Stripe keys, including validation warnings and dashboard links. - Added clear success, cancellation, expiration, error, and wrong-account states with account switching. - Added loading placeholders, secret visibility controls, overwrite warnings, and accessibility announcements. - **Bug Fixes** - Improved interstitial animations to respect reduced-motion preferences. - Preserved return destinations and related parameters during sign-in flows. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com> |
||
|
Coenen Benjamin
|
67e48f51b6 |
fix(warehouse): don't display ducklake metadata schema in replicated tables/schemas (#50123)
Don't show to users replicated schema or tables from the ducklake metadata schema to avoid infinite loop. (This is already forbidden at API level) Signed-off-by: Benjamin <5719034+bnjjj@users.noreply.github.com> |
||
|
Anthony Lio
|
45199443c8 |
fix(ui): admonition component parity (#49925)
## what is the current behavior? admonition icon <> text not optically aligned + rendered differently in docs and the design system _ docs showed admonition text at 15px/28px because the page's prose styles reached inside the component, while the same callout was 13px in the design system _ lists ## what is the new behavior? - the title offset is now conditional. a title and body copy have different line heights, so they need different nudges to sit level with the icon. - fixes list markers and the ordered-list chip alignment inside callouts. - removes `.admonition-content` css that nothing referenced - fixes 5 admonition titles that were not capitalized. | state | preview | | -------|------| | before | <img width="902" height="279" alt="image" src="https://github.com/user-attachments/assets/2fffb183-81e2-4eff-8f0d-8a07649390e8" /> | | after | <img width="902" height="279" alt="image" src="https://github.com/user-attachments/assets/22abe3ed-fd6d-49ac-aa37-4292bca5850a" /> | ## follow ups - better composition: title, description and actions are still props _ a compound api (`Admonition.Title`, `Admonition.Actions`) would remove the `childProps` escape hatch <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Documentation** * Improved capitalization of note and warning titles in the Metabase and Prisma guides for consistency. * Updated the contributing guide’s table of contents to exclude feedback headings. * Improved heading structure for the documentation feedback section. * **UI Improvements** * Refined admonition and alert typography, spacing, list formatting, and ordered-list alignment. * Improved content spacing when titles, descriptions, or icons are present. * Updated action links and buttons for more consistent sizing. * Adjusted alert content styling for a clearer presentation. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
|
f5d0bf5d98 |
fix(ui): let command dropdowns scroll on touch inside overlays (#50103)
## What kind of change does this PR introduce? Bug fix. ## What is the current behavior? A command dropdown cannot be scrolled by touch when it sits inside a dialog or sheet. Radix wraps a modal dialog's overlay in a scroll lock that cancels wheel and touch events whose target is not inside the sheet. A dropdown portals to the body, so it falls outside that boundary and its scroll events get cancelled. Callers have been rediscovering this one at a time and fixing only half of it. `MultiSelectorList` stops wheel events reaching the document, and `SchemaSelector` and `FunctionSelector` expose a `stopScrollPropagation` prop that does the same. None of them handle touch, so the desktop symptom is fixed everywhere and the mobile one is fixed nowhere. ## What is the new behavior? `CommandList` keeps wheel and touch events off the document itself. That covers every command dropdown in the monorepo, in an overlay or not, with no call-site changes. A caller's own `onWheel` or `onTouchMove` still runs. The workarounds this replaces are removed: the handler in `MultiSelectorList`, and the `stopScrollPropagation` prop on `SchemaSelector` and `FunctionSelector` along with its four call sites. One behavior change worth naming: overscrolling past the end of a dropdown no longer scrolls the page behind it. That is what a dropdown should do, and it is what the four `stopScrollPropagation` call sites were already opting into. #50072 depends on this. It swaps two selects for comboboxes, and Radix Select brings its own scroll lock, so without this the swap would regress touch scrolling. ## To test - On the deploy preview, [open a project's Connect sheet](https://studio-staging-git-dnywh-fix-multi-select-scrol-63608b-supabase.vercel.app/dashboard/project/_) and pick the MCP tab. - Narrow the window to phone width and switch on touch emulation in devtools. - Open the features dropdown and drag the list. It should scroll, and the sheet behind it should stay put. - Repeat with a mouse wheel to confirm desktop scrolling still works. - [Open Authentication > Hooks](https://studio-staging-git-dnywh-fix-multi-select-scrol-63608b-supabase.vercel.app/dashboard/project/_/auth/hooks) > Create hook, open the schema picker, and confirm it still scrolls by wheel now that `stopScrollPropagation` is gone. - [Open the SQL editor](https://studio-staging-git-dnywh-fix-multi-select-scrol-63608b-supabase.vercel.app/dashboard/project/_/sql/new)'s schema picker on a page with no overlay and confirm the list scrolls normally. <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit - **Bug Fixes** - Improved wheel and touch scrolling behavior in command lists, selectors, and multi-select menus. - Reduced unwanted scroll-lock interference when using selectors inside overlays such as dialogs and sheets. - Preserved support for supplied scroll event callbacks. - **Accessibility** - Added clearer accessible labels to the Connect and mobile navigation menu buttons. - Updated the Connect button text behavior for icon-only and standard presentations. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Co-authored-by: Claude Opus 5 <noreply@anthropic.com> Co-authored-by: Joshen Lim <joshenlimek@gmail.com> |
||
|
Anthony Lio
|
84d3ce1bd2 |
fix(docs): keep code block line numbers sticky while scrolling (#49962)
## What kind of change does this PR introduce? bug fix + some ui polish on the docs code block ## What is the current behavior? line numbers scroll away with the code _ so you lose your place in any block wide enough to scroll _ the gutter is also see-through so scrolled code renders interleaved with the numbers ## What is the new behavior? - makes gutter sticky and stays pinned while the code scrolls (no rubber) - fixes gutter, row hover, and button backgrounds as opaque so nothing bleeds through - adds gutter right edge fades vs bg - adds right padding so line endings clear the buttons - adds press feedback on the shared in-house `Button` gated behind `motion-safe` | state | preview | | -------|------| | before | <img width="566" height="263" alt="image" src="https://github.com/user-attachments/assets/81778815-1f15-4ea5-a647-ed48418210e8" /> | | after | <img width="566" height="259" alt="image" src="https://github.com/user-attachments/assets/b7d770c5-4156-491f-93c8-b0d900b7dc85" /> | <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **UI Improvements** * Refined code block controls with consistent styling, active states, animated icons, and automatic copy-status reset. * Improved code block line-number presentation with hover effects, sticky gutters, spacing, shadows, and smoother horizontal scrolling. * Added smoother button transitions and a subtle pressed-state animation. * Respect reduced-motion preferences by disabling icon animations when requested. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Tarun Khandelwal
|
d7ae1c7eae |
chore: allow triggering of restarts for v3 projects (#49586)
## 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? Allowing v3 projects to be restarted. ## What is the current behavior? The current behaviour greys out the `Restart Project` ## What is the new behavior? It allows the restart the project: <img width="675" height="158" alt="Screenshot 2026-08-26 at 12 49 35 PM" src="https://github.com/user-attachments/assets/a80f64d3-9e4d-4b8e-bca2-ec0a6717496c" /> ## Additional context Tested with: https://github.com/supabase/platform/pull/37561 Part of: KUBE-1285 <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Bug Fixes** * Enabled server restarts for AWS (Revamped) projects. * Updated the restart control so it is no longer disabled or restricted for these projects. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Danny WhiteandJoshen Lim
|
4a5db160be |
feat(studio): extract shared zod number input helpers (#50102)
## What kind of change does this PR introduce? Refactor. Extracts zod number input preprocessing into a shared Studio helper and migrates existing call sites. ## What is the current behavior? Number input clearing logic for controlled `type="number"` fields is duplicated across Studio forms. The pipeline destination form had local helpers, while Auth and settings forms inline the same `z.preprocess` pattern. ## What is the new behavior? Adds `apps/studio/lib/forms/zod-number-input.ts` with: - `requiredNumberInputSchema` for required `z.number()` fields (`''` → `NaN`) - `optionalNumberInputSchema` for optional number fields (`''`/`null` → `undefined`) - `preprocessEmptyNumberInput` for `z.coerce.number()` schemas Migrates: - Pipeline destination form - Connection pooling settings - Auth provider validation (email and SMS OTP fields) - SMTP settings - MFA settings - Protection settings ## To test 1. Open **Authentication > Providers > Email**, clear **Email OTP expiration** or **Minimum password length**, and confirm the field stays empty and shows validation on save. 2. Open **Authentication > Providers > Phone**, select a provider, clear **SMS OTP Expiry**, and confirm validation still works. 3. Open **Authentication > Emails > SMTP settings**, clear **Port** or **Rate limit**, and confirm validation on save. 4. Open **Authentication > Multi-factor authentication**, clear **Max enrolled factors** or **Phone OTP length**, and confirm validation on save. 5. Open **Database > Replication**, start a new pipeline, select **BigQuery**, expand **Advanced settings**, clear a required numeric field, and confirm it stays empty until filled. 6. (Optional, hosted non-HA projects only) Open **Database > Settings**, scroll to **Connection pooling**, clear **Connection pool size**, and confirm the value does not snap back before save. <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Bug Fixes** * Improved handling of empty numeric fields across authentication, SMTP, database replication, and connection pooling settings. * Optional numeric settings can now be cleared without triggering unnecessary validation errors. * Required numeric fields continue to display appropriate validation errors when left incomplete. * Standardized validation behavior for OTP settings, password requirements, SMTP limits, replication configuration, and pool sizing. * Refined protection settings to focus on CAPTCHA and leaked-password options. * **Tests** * Added coverage for empty and optional numeric input behavior. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Co-authored-by: Joshen Lim <joshenlimek@gmail.com> |
||
|
Joshen Lim
|
db0e6b761b |
Joshenlim/fe 4304 bring database connections out of feature preview (#50107)
## Context
As per PR title - we're bringing Database Connections out of feature
preview and it'll live on the dashboard by default 🙂
Also deprecating the existing Ongoing queries panel which Database
Connections now supercedes.
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Database Connections is now available without feature-preview
activation.
* The SQL editor’s “View running queries” option now links directly to
Database Connections.
* **Bug Fixes**
* Query cancellation and session termination now refresh database
activity data.
* **Removed**
* Removed the in-editor ongoing queries panel and its termination
controls.
* Removed the Database Connections promotional banner, preview
messaging, settings, and related telemetry.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
|
||
|
Pedro RodriguesandClaude Sonnet 5
|
e5f2b29625 |
refactor(studio): remove USE_REMOTE_MCP gate, always use remote MCP server (#50089)
## Summary - Removes the `USE_REMOTE_MCP` env-var gate from the dashboard assistant: `getMcpTools` now always connects to the remote MCP server (the rollout from #47479 has been stable ~2 months and is enabled in prod). - Drops the var from `apps/studio/turbo.jsonc` and deletes the now-obsolete transport-selection tests. - The legacy in-process client (`createInProcessSupabaseMCPClient`) stays, re-scoped to the hermetic eval harness (`mock-tools.ts`, `evals/preflight.ts`); removing it is tracked by AI-897. ## Verification - `pnpm exec tsc --noEmit` in apps/studio: no errors in any changed file (one pre-existing unrelated error in `packages/ui-patterns/.../InstructionBlocks.tsx`). - `mcp-tools.test.ts` (7), `mock-tools.test.ts` (15), `tools/index.test.ts` (6), `supabase-mcp.test.ts` (11) all pass. ## Risk Low. Remote failure already degrades to non-MCP tools in `getTools`; rollback = revert this PR (or re-add the gate). ## Follow-up After this lands in prod, `USE_REMOTE_MCP` can be removed from the Vercel env vars — nothing in the repo reads it anymore. <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Improvements** * AI tools now consistently use the remote service when retrieving available tools. * If the remote service is unavailable, times out, or cannot authenticate, the assistant continues operating with the tools that remain available. * Evaluation and development behavior now more closely reflects the remote service experience. * **Maintenance** * Updated supporting documentation and automated coverage to reflect the streamlined tool connection behavior. <!-- end of auto-generated comment: release notes by coderabbit.ai --> Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com> |
||
|
CharisandJoshen Lim
|
2ac052752a |
fix(pg-meta): fix invalid SQL when setting column default to NULL (#50097)
## Summary
- Fixes invalid SQL generation when setting a column's default value to
NULL through Studio's Table Editor
- The bug occurred because raw `null` values bypassed literal
serialization in the SQL builder
- Added special case handling in both `create()` and `update()`
functions to use `literal(null)` regardless of format
## Details
Setting a column's default to NULL via the "Set as NULL" suggestion in
Studio generated invalid SQL: `ALTER TABLE ... ALTER COLUMN ... SET
DEFAULT ;` (missing the NULL keyword). This was caused by
`default_value` being JS `null` and bypassing the literal()
serialization function when `default_value_format === 'expression'`.
The fix ensures that when `default_value === null`, we always use
`literal(null)` to produce the correct `SET DEFAULT NULL` clause.
## Test plan
- New regression test in `packages/pg-meta/test/columns.test.ts`
("update column default value to NULL via expression format") verifies
the fix by:
- Creating a text column with a literal default
- Updating it using `default_value_format: 'expression', default_value:
null` (exact payload from Studio)
- Asserting the column's default value is null
- All 58 tests in `packages/pg-meta/test/columns.test.ts` pass
- TypeScript check (`tsc --noEmit`) passes
Fixes FE-4343
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
- **Bug Fixes**
- Fixed column creation and updates so explicitly setting a default
value to `NULL` correctly preserves the `DEFAULT NULL` clause.
- Ensured setting a column default to `NULL` through expression-based
editing is reflected accurately when column details are retrieved.
- **Style**
- Improved the column editor’s suggestion menu layout and spacing for
clearer readability.
- **Tests**
- Added coverage for updating a column default to `NULL`.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
---------
Co-authored-by: Joshen Lim <joshenlimek@gmail.com>
|
||
|
Gildas Garcia
|
bfb0737d14 |
Fix to ensure labels, descriptions and validation errors are correctly linked to their inputs (#50080)
## Problem `FormItemLayout` does not correctly binds inputs descriptions and validation messages to their inputs. This is because the input ids are generated and not correctly propagated to the `FormMessage` and `FormDescription` components. Besides, we still pass `name` or `id` directly to the inputs or `FormItemLayout` in some places. ## Solution - Fix `FormItemLayout` to correctly binds inputs descriptions and validation messages to their inputs - Fix incorrect usages - Fix Design System documentation ## How to test The issue is visible in production: - Open https://supabase.com/design-system/docs/ui-patterns/forms - Open the devtool and check the labels `for`, the description `id` and the input `id` or `aria-describedby` attributes. You'll see they often don't match Do the same on staging: - Open https://design-system-git-fix-a11y-form-input-descriptions-supabase.vercel.app/design-system/docs/ui-patterns/forms - Open the devtool and check the labels `for`, the description `id` and the input `id` or `aria-describedby` attributes. They now match Dashboard fixes: - https://studio-staging-git-fix-a11y-form-input-descriptions-supabase.vercel.app/dashboard/account/tokens: _Expires in_ select button is now correctly linked to its label - https://studio-staging-git-fix-a11y-form-input-descriptions-supabase.vercel.app/dashboard/account/me: the switches are now correctly linked to their label - In Database/Indexes: the select buttons when creating an index are now correctly linked to their label - All other changes are the same things |
||
|
Danny White
|
66a748cdd1 |
fix(ui): restore accordion trigger layout (#50114)
## What kind of change does this PR introduce? Bug fix. Follow-up to #49660. ## What is the current behavior? `AccordionTrigger` adds horizontal padding at the shared primitive level. This shifts content at call sites that already own their spacing. In the Studio Advanced settings sections, padding around the whole accordion also constrains the trigger and its hit area instead of letting it fill the row. ## What is the new behavior? The shared trigger no longer adds horizontal padding and its header explicitly fills its container. Both Studio Advanced settings sections now apply their padding inside the full-width trigger and content, keeping the text aligned while extending the focus outline and hit area across the row. The inset focus treatment from #49660 is unchanged. | Before | After | | --- | --- | | <img width="1334" height="286" alt="CleanShot 2026-09-08 at 16 18 57@2x" src="https://github.com/user-attachments/assets/3a057b45-1eb4-4928-ae69-2d636b0c4678" /> | <img width="1302" height="278" alt="CleanShot 2026-09-08 at 16 18 12@2x" src="https://github.com/user-attachments/assets/0454493e-4fbc-4ad2-a9fd-1b9dc1d486dc" /> | ## To test ### Studio replication destination 1. Open the [Studio staging preview](https://studio-staging-git-dnywh-fixaccordion-trigger-width-supabase.vercel.app) and select a project with Pipelines enabled. 2. Go to **Database > Replication**. Under **Destinations**, click **Add destination**. 3. Tab to **Advanced settings** and press Enter to expand it. 4. Confirm the text aligns with the form content, the focus outline spans the full row between the sheet edges, and clicking near the far right of the row toggles it. ### Studio access-token permissions 1. In the [Studio staging preview](https://studio-staging-git-dnywh-fixaccordion-trigger-width-supabase.vercel.app), go to **Account > Access Tokens**. 2. Click **Generate new token** and scroll to **Permissions**. 3. Tab to any permission category, then click near the far right of its row. 4. Confirm the focus outline fills the category row, the label keeps its existing inset, and the full row toggles the category. ### Pricing FAQs 1. Open the [Pricing preview](https://zone-www-dot-com-git-dnywh-fixaccordion-trigger-width-supabase.vercel.app/pricing) and scroll to **Frequently asked questions**. 2. Tab to any question, then click near the far right of its row. 3. Confirm the question has not gained extra horizontal inset and the full row remains interactive. ### Design-system Accordion 1. Open the [design-system Accordion preview](https://design-system-git-dnywh-fixaccordion-trigger-width-supabase.vercel.app/docs/components/accordion). 2. Tab through the three accordion triggers and toggle each one. 3. Confirm each focus outline fills its row, with no extra horizontal padding around the label. |
||
|
|
fb28b70fe7 |
Update PostgresConf SJC 2026 contest page content (#50112)
## 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? Content update to a `/go` landing page. ## What is the current behavior? The PostgresConf San Jose 2026 contest page (`supabase.com/go/postgresconf-sjc-2026/contest`) shows a "Learn about Multigres" CTA linking to multigres.com, a "Conference Talk: Thursday, April 23, 2026 2:30pm PDT" line, and a contest entry deadline of Monday, May 4, 2026. ## What is the new behavior? - Removed the outdated "Conference Talk" date/time line - Changed the CTA to "Watch the webinar," linking to the recording (https://www.youtube.com/watch?v=ahhQ0n1SHiQ) - Updated the contest entry deadline to Monday, September 14, 2026 at 12:00 PM PST ## Additional context Content-only change to `apps/www/_go/events/postgresconf-sjc-2026/contest.tsx`. <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit - **Content Updates** - Updated the event page to link to the webinar recording with a “Watch the webinar” label. - Changed the contest deadline to September 14, 2026, at 12:00 PM PST. - Removed the talk description from the event page. <!-- end of auto-generated comment: release notes by coderabbit.ai --> Co-authored-by: Wendie Cheung <wendie.cheung@supabase.io> Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com> |
||
|
Joshen Lim
|
ac0d4563d6 |
Add GH token for fetchAgentSkills to mitigate rate limits (#50106)
## Context As per PR title - just a nice to have as our `www` preview builds occasionally fail, likely due to rate limits on the GH API Env var `AGENT_SKILLS_GITHUB_TOKEN` has been added to the `www` app on Vercel <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Chores** * Added optional GitHub authentication support for retrieving agent skills. * Preserved existing request behavior when no authentication token is provided. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Danny White
|
3418975b70 |
feat(studio): configure publication partition handling (#49845)
## What kind of change does this PR introduce? Small Pipelines feature. Stacked on #49844. ## What is the current behavior? New publications always publish changes from partitioned tables through their parent table. The creation sheet does not expose the v2 API option that controls this behavior. ## What is the new behavior? Adds a default-on Publish partitions as the parent table checkbox to the publication sheet. Turning it off submits publish_via_partition_root as false so each partition can appear as a separate destination table. ## To test 1. Open the pipeline creation sheet and choose to create a publication. 2. Confirm Publish partitions as the parent table is selected by default. 3. Create a publication and confirm publish_via_partition_root is true. 4. Repeat with the option cleared and confirm publish_via_partition_root is false. 5. Confirm the selected tables and publication name are unchanged by the option. <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit - **New Features** - Added a “Publish partitions as the parent table” option when creating publications. - The option is enabled by default and can be turned off before submitting the publication. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Danny WhiteandClaude Opus 5
|
d0dd81ba62 |
fix(studio): point the Warehouse table picker at the v2 publication API (#50105)
## What kind of change does this PR introduce? Bug fix. Master is currently red, so this needs to land before anything else can go green. ## What is the current behavior? `WarehouseSchemaTablePicker` still imports `@/data/replication/publications-query`, which #49844 removed. Studio does not typecheck on master, which takes typecheck, knip, unit tests, E2E, the Studio Docker build and both Studio Vercel deployments down with it. ``` WarehouseSchemaTablePicker.tsx(20,49): error TS2307: Cannot find module '@/data/replication/publications-query' WarehouseSchemaTablePicker.tsx(87,8): error TS7006: Parameter 'publication' implicitly has an 'any' type ``` ## What is the new behavior? The picker reads the `supabase_warehouse` publication through `useReplicationPublicationQuery`, matching how `TableCopySelection` and `DestinationForm` were migrated. It only ever needed that one publication, and it needs the publication's tables, which the v2 list endpoint no longer returns. Behavior is unchanged. A missing publication still resolves to an empty selection during first-time setup, and a failed lookup still only blocks when editing an already-enabled Warehouse. ## To test - Confirm CI goes green here. - On the Studio deploy preview, open a project's Connect sheet and pick the Warehouse mode. - With Warehouse not yet enabled, the schema and table list renders with nothing checked. - With Warehouse already enabled, reopen the picker and confirm the previously replicated tables come back checked. <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit - **Bug Fixes** - Improved warehouse schema table selection by loading the Warehouse publication directly. - Table selections now initialize correctly from the returned publication. - Preserved loading and error handling for publication retrieval. <!-- end of auto-generated comment: release notes by coderabbit.ai --> Co-authored-by: Claude Opus 5 <noreply@anthropic.com> |
||
|
Danny WhiteandJoshen Lim
|
8dd97d75aa |
refactor(studio): use v2 replication publication APIs (#49844)
## What kind of change does this PR introduce? Studio data-layer migration. ## What is the current behavior? Studio loads complete publication details through the original bulk endpoint and creates publications by executing SQL against the source database. Publication and source-table data use names where stable table IDs are available. ## What is the new behavior? Uses the v2 publication-name, publication-detail, publication mutation, and source-table endpoints. The existing creation sheet continues to behave the same, including publishing partition changes through the parent table by default. Initial-sync selection and Analytics Bucket associations now consume the selected publication detail. Generated platform API types and their required nullability updates are included. The generated Platform contract accounts for roughly 10,000 changed lines in this PR. ## Dependency Depends on the v2 source table, table column, and publication endpoints from [supabase/platform#37505](https://github.com/supabase/platform/pull/37505), which are deployed to production. ## To test 1. Open the pipeline creation sheet and select an existing publication. 2. Create a publication with mixed-case schema and table names, then confirm the table names are shown while stable IDs are submitted. 3. Exercise all four initial-sync policies, including selecting individual tables. 4. Reopen the publication and table selectors and confirm they refresh without replacing populated options. 5. Edit and delete a publication. 6. Open an Analytics Bucket associated with a pipeline and confirm its publication tables resolve correctly. 7. Confirm unlimited WAL retention renders as Unlimited on pipeline status. <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit - **New Features** - Improved replication publication setup with on-demand table loading, refresh controls, clearer table labels, and streamlined publication selection. - Publication creation and updates now use the latest replication API and table-based configurations. - Added clearer handling for tables removed from publications, including stale-selection warnings. - **Bug Fixes** - Prevented table selections from carrying over when switching publications. - Improved replication status displays when lag or WAL metrics are unavailable. - Updated replication deletion and table management for the latest API behavior. - **Tests** - Expanded coverage for publication creation, table selection, stale tables, loading states, and replication metrics. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Co-authored-by: Joshen Lim <joshenlimek@gmail.com> |
||
|
Danny WhiteandJoshen Lim
|
282a2a22cf |
fix(studio): allow pipeline number inputs to be cleared (#50078)
## What kind of change does this PR introduce? Bug fix for the pipeline destination form. ## What is the current behavior? Clearing a numeric field under Advanced settings restores its previous value immediately, so users cannot replace the value normally. ## What is the new behavior? Numeric fields use an empty string as their controlled blank value while editing, so users can clear an existing value before entering its replacement. Existing field-specific validation messages are preserved for invalid or empty required values. Maximum staleness remains optional and can be left blank. Adds component and schema coverage for all five numeric Advanced settings fields, including the BigQuery-only fields. ## To test 1. Open Database > Replication and start a new pipeline. 2. Select BigQuery and expand Advanced settings. 3. Clear each numeric field and confirm it can be edited without reverting to its previous value. Required fields may show their existing validation message while empty. 4. Enter a replacement value such as `5` and confirm the field accepts it. 5. Leave a required numeric field blank and confirm the existing field-specific validation message appears. <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit - **Bug Fixes** - Improved validation for replication destination settings, with clearer errors when required numeric fields are left empty. - Validation now updates immediately when numeric values are cleared or corrected. - BigQuery connection pool size must be greater than zero. - Empty maximum staleness values are accepted and handled correctly. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Co-authored-by: Joshen Lim <joshenlimek@gmail.com> |
||
|
Marouane SoudaandCharis Lam
|
0e35cbf4a4 |
Security definer materialized view (#40800)
Fixes #40799 Now, the correct error will be shown for materialized views accessible to `anon` and `authenticated` roles via Data API, instead of the unrelated "Security Definer view" error. For convenience, users can immediately fix the issue by running the correct SQL query to revoke select from `anon` and `authenticated`, just by clicking on "confirm" on the confirmation modal I just created. https://github.com/user-attachments/assets/f3ce9353-4ad0-4063-bf33-0b403f5fa87a Before <img width="958" height="440" alt="materialized" src="https://github.com/user-attachments/assets/89047c91-da35-4b9f-b7e3-82e877bcf2c6" /> Edit: Thanks to the review by @saltcod, I now realise that revoking access from `anon` and `authenticated` users might not be the optimal solution since it would break many projects relying on the access to materialized views. After pondering on the possible solutions, I figured there isn't an easy one, so I did away with the autofix button, and instead created a dialog explaining three possible options for the user, with a sample query under each one for convenience. <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Added clearer warnings when materialized views are accessible through the API. * Added a dialog explaining how to review and revoke API access for materialized views. * Added guidance and code examples for restricting access to API roles. * Added a link to the Security Advisor for additional information. * **Improvements** * Updated materialized view tooltips to accurately describe API accessibility. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Co-authored-by: Charis Lam <26616127+charislam@users.noreply.github.com> |
||
|
Nik RichersandNik Richers
|
c2ebfebb39 |
docs: restore Troubleshooting nav link under Resources dropdown (#50028)
## 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 restores the "Troubleshooting" quick link to the docs top navigation. It was dropped from the "Manage" dropdown as a side effect of a broader information-architecture rework, and this PR re-adds it under the "Resources" dropdown instead, since the underlying troubleshooting guide page was never removed, only its top-nav shortcut. ## What is the current behavior? The "Troubleshooting" entry was removed from the "Manage" dropdown in the top nav during a prior IA restructuring commit. The `/guides/troubleshooting` page itself still exists and is still linked from within the Monitoring & Debugging sidebar, but there is no longer a top-level nav shortcut to it. Note that we might need to make some additional updates as our observability docs changes wrap up. ## What is the new behavior? - Adds a "Troubleshooting" item (with its existing icon) to the end of the "Resources" dropdown, after Glossary, Changelog, Status, and Contributing. - Links to the existing `/guides/troubleshooting` route. - No changes to the "Manage" dropdown or any other nav group. ## Additional context **Verification:** | Check | Result | | --- | --- | | `pnpm typecheck --filter=docs` | pass | | `pnpm lint --filter=docs` | pre-existing 1138 warnings/errors unrelated to this file; edited file has zero lint issues (lint is not hard-gated for docs) | | Manual check: dev server, opened "Resources" dropdown | Troubleshooting appears with icon, links to `/guides/troubleshooting` (200) | | Manual check: "Manage" dropdown unchanged | confirmed | | Manual check: mobile nav (shares same config) | confirmed same entry renders | ### Proof: Screenshots of the "Resources" dropdown, before and after this change: | [Before (production)](https://supabase.com/docs/guides/getting-started) | [After (PR preview)](https://docs-git-docs-restore-troubleshooting-nav-resources-supabase.vercel.app/docs/guides/getting-started) | | --- | --- | |  |  | ### Test plan - [ ] Open the "Resources" dropdown in the top nav and confirm "Troubleshooting" appears after "Contributing" - [ ] Click "Troubleshooting" and confirm it navigates to `/guides/troubleshooting` - [ ] Confirm the "Manage" dropdown is unchanged - [ ] Check the mobile nav shows the same entry under "Resources" Co-authored-by: Nik Richers <nik@validmind.ai> |
||
|
Jeremias Menichelli
|
10df98f2d5 |
feat(kb): Create homepage and pinned topics and guides (#49999)
## 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? Adding the ability to pin guides or specific topics to homepage. _To test:_ - Enter the preview of the project - Go to the homepage - Make sure pinned topics, pinned guides and the rest of the links work correctly. _Screenshot_ <img width="894" height="3772" alt="Screenshot 2026-09-04 at 12-12-37 Home Supabase Knowledge Base" src="https://github.com/user-attachments/assets/a1b33e1b-bcbf-43db-b453-64fc1a0c9a00" /> <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit - **New Features** - Added homepage sections for pinned topics, featured guides, all guides, and all topics. - Added reusable cards with titles, descriptions, topic badges, and optional links. - Added support for GitHub-style Markdown alerts, including notes, tips, warnings, and cautions. - Added the ability to mark guides and topics as pinned for homepage featuring. - Added a guide demonstrating supported alert styles. - **Style** - Improved header, hero, and card borders for a more consistent visual treatment. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Jordi Enric
|
be030229bf |
feat(studio): add workers to unified logs FE-4281 (#49682)
## Problem Unified Logs does not expose Workers logs, so users cannot search Workers ingress, runtime, or build events alongside other services. ## Fix Add a Workers log type that classifies all three Workers OTEL streams. Gate the option and any persisted Workers filters with the existing Workers feature flag. ## How to test - Enable the Workers feature flag and open Unified Logs. - Select Workers from the Log Type filter. - Expected result: Unified Logs shows ingress, runtime, and build events with the Workers icon. - Disable the Workers feature flag and load a URL containing `log_type:eq:workers`. - Expected result: the Workers option and filter are removed, and Workers logs are not queried. - Run `./node_modules/.bin/vitest --run components/interfaces/UnifiedLogs/UnifiedLogs.queries.test.ts components/interfaces/UnifiedLogs/UnifiedLogs.utils.test.ts data/workers/worker-logs-query.test.ts` from `apps/studio`. <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Added Workers as a selectable log type in Unified Logs. * Unified Logs now combines worker ingress, guest, and API streams under the Workers category. * Added a dedicated Workers icon and worker log filtering. * **Improvements** * Worker filters and URL parameters respect feature availability. * Worker details show relevant metadata while omitting unavailable HTTP fields. * Improved handling of worker log levels, statuses, and raw data. * Added stronger validation for unified log data. * **Tests** * Added coverage for worker routing, filtering, feature visibility, parsing, and metadata redaction. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Alaister YoungandAlaister Young
|
ec1029dff0 |
chore: migrate from clsx + tailwind-merge to shadcn-ui/cn (#49938)
Migrates the repo off `clsx` + `tailwind-merge` to [shadcn-ui/cn](https://github.com/shadcn-ui/cn). Every app and package already gets `cn` from `packages/ui`, so the swap happens in that one helper and flows through to Studio, docs, www, and the rest. **Changed:** - `packages/ui` `cn` helper now uses `createCn` from `cn/config`, keeping the custom `card`/`content` spacing scale so `p-card` still overrides `p-4`. It has an explicit signature and re-exports `ClassValue`. - The four www Launch Week files that imported the `ClassValue` type from `clsx` now import it from `ui`. - `blocks/vue` local `lib/utils.ts` re-exports `cn` from the package. - Comments/README that referenced tailwind-merge. **Removed:** - Direct `clsx` and `tailwind-merge` deps from `ui`, `ui-patterns`, `www`, and `blocks/vue`. `ui-patterns` and `www` declared them without importing. **Added:** - `packages/ui/src/lib/utils/cn.test.ts` covering clsx-style joining, conflict resolution, the custom spacing scale, and variant handling. Not migrated: the standalone apps under `examples/`. They're outside the workspace and mostly on Tailwind v3, which `cn` doesn't support. Lockfile note: after merging master, the lockfile diff is only the intended swap (`clsx` and `tailwind-merge` out, `cn@0.2.5` in). `tailwind-merge` stays in the lockfile as a transitive dep of a third-party package. Release-age note: this sat in draft with a temporary `minimumReleaseAgeExclude` entry for `cn` while `cn` was inside the workspace's 3-day `minimumReleaseAge` window. That window has closed, so the exclusion is gone and nothing bypasses the release-age gate. ## To test - `pnpm install --frozen-lockfile` succeeds with no `minimumReleaseAgeExclude` entry for `cn`. - `pnpm --filter ui test` – new `cn.test.ts` passes, including `cn('p-4', 'p-card')` → `p-card`. - Typecheck passes for studio, ui, ui-patterns, vue-blocks. www typecheck panics under tsgo on master already (pre-existing, unrelated); it passes with the JS `tsc` binary. - Spot-check Studio locally: class overrides still win in the usual places (e.g. `CodeEditor` height, `Button` variants with a custom `className`). https://claude.ai/code/session_01MkAt16tsPRDTm9oB5Jr8Ub <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Improvements** * Standardized Tailwind class merging across shared UI utilities while preserving conditional classes, custom spacing classes, and variant behavior. * Updated related components and examples to use the standardized class-merging utility. * **Tests** * Added coverage for conditional class handling, conflicting utility resolution, custom spacing classes, and variant separation. * **Documentation** * Updated usage guidance to reflect the standardized Tailwind class-merging approach. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Co-authored-by: Alaister Young <10985857+alaister@users.noreply.github.com> |
||
|
Francesco SansalvadoreandClaude Sonnet 5
|
b0601f9ab7 |
feat(storage): add object versioning feature flag and feature preview (#49202)
| # | Branch | Base |
| - | ------ | ---- |
| 1 | `feat/storage-versioning-private-alpha` ◀ | `master` |
| 2 | `feat/storage-versioning/002-bucket-form-fields` | 1 |
| 3 | `feat/storage-versioning/003-bucket-modals` | 2 |
| 4 | `feat/storage-versioning/004-object-versions-data` | 3 |
| 5 | `feat/storage-versioning/005-file-preview-versions` | 4 |
| 6 | `feat/storage-versioning/006-billing-storage-retention` | 5 |
### PR 01
Adding feature flag for versioning.
- `UI_PREVIEW_STORAGE_VERSIONING` local storage key, allowlisted
- Feature preview entry gated on
`useFlag('storageVersioningPrivateAlpha')`
- `useIsStorageVersioningEnabled()`, following the existing
`useIsSqlEditorManualSaveEnabled` shape
<img width="910" height="604" alt="Screenshot 2026-08-19 at 11 33 45"
src="https://github.com/user-attachments/assets/5c51c3fa-6100-48e5-914d-63a9948a5024"
/>
Note: will improve the feature preview with an image once we approach
release.
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
- **New Features**
- Added a Storage Versioning feature preview in the dashboard.
- Preview access is controlled by feature flags and is disabled by
default.
- Added descriptive information about Storage Versioning and its Private
Alpha availability.
- Added support for remembering the Storage Versioning preview setting.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
---------
Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com>
|
||
|
Pedro Rodrigues
|
16046a0139 |
docs: add omp (oh-my-pi) to supported clients in MCP and plugin guides (#50045)
## What this does
Adds **omp** (oh-my-pi) across the Supabase AI-tools docs.
- **Plugin docs** (`AgentPluginsPanel`) — omp client + Anthropic
marketplace install: `omp plugin marketplace add
anthropics/claude-plugins-official`, then `omp plugin install
supabase@claude-plugins-official`.
- **MCP docs** (`McpUrlBuilder`) — omp under "AI Agent CLI":
`.omp/mcp.json` / `~/.omp/agent/mcp.json` (JSON), `/mcp add` guided
setup, project vs user scope.
- **"Pick your agent" grid** — add the omp logo and tagline ("A coding
agent with the IDE wired in.").
No CLI command mapping entry — omp has no `omp mcp …` subcommand, so the
page documents file-based setup (`/mcp add` wizard or direct JSON edit).
Matches other file-configured clients like kiro.
## Testing
Verified against omp 18.1.11: written `.omp/mcp.json` parsed by runtime,
`/mcp list` shows server connected; `omp plugin marketplace add/install
supabase@claude-plugins-official` works; `omp plugin list` confirms
installed.
## Preview
### Agent Plugin page
<img width="1128" height="706" alt="image"
src="https://github.com/user-attachments/assets/1f06895b-2d45-4b64-b9ab-c1670201c370"
/>
### MCP page
<img width="1128" height="706" alt="image"
src="https://github.com/user-attachments/assets/30a9a823-0e78-4c3b-aeeb-3a6a7aa40ed6"
/>
### AI Tools main page
<img width="1128" height="706" alt="image"
src="https://github.com/user-attachments/assets/064a3234-9f4d-4f0a-9598-fbb2a465266f"
/>
Closes AI-1184
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
- **New Features**
- Added OMP as a supported AI coding agent across documentation and
integrations.
- Added OMP installation guidance, including marketplace setup, scope
options, session reloads, and authentication.
- Added support for configuring the Supabase MCP server through OMP.
- Added OMP branding, icon, tagline, repository, and documentation
links.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
|
||
|
Jeremias Menichelli
|
7e19cbe140 | feat(kb): Add topics pages support (#49959) | ||
|
Jeremias Menichelli
|
0ab0bb8efd | feat(kb): Apply supabase theme to code blocks (#49957) | ||
|
Coenen BenjaminandCharis Lam
|
11289328e5 |
add support for warehouse connection string (#49914)
Add support for connection string for warehouse. This PR gives the ability to enable warehouse on a project and also get the connection string to connect to. > This project is only available in staging for now and gated behind a feature flag <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit - **New Features** - Added a Warehouse connection option to the Connect dialog. - Select schemas and tables to replicate, with setup progress, error recovery, and retry support. - View copyable Warehouse connection details, credentials guidance, command-line instructions, and DuckLake setup scripts. - Warehouse availability is controlled by feature configuration. - **Tests** - Added coverage for Warehouse table selection, setup script generation, URL parsing, and connection configuration utilities. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Signed-off-by: Benjamin <5719034+bnjjj@users.noreply.github.com> Co-authored-by: Charis Lam <26616127+charislam@users.noreply.github.com> |
||
|
|
c086fe0d3f |
fix(studio): stop duplicating access_token_creation_sheet_dismissed on Done (#50077)
<!-- ccr-slack-attribution --> _Requested by **Pam Chia** · [Slack thread](https://supabase.slack.com/archives/C076KTY11DF/p1788743741768969?thread_ts=1788743741.768969&cid=C076KTY11DF)_ ## 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? Bug fix (telemetry). ## What is the current behavior? `access_token_creation_sheet_dismissed` (added in #49965) fires on every close of the "Generate token" sheet in `/account/tokens`, including a successful completion. The "Token created" step closes the sheet through a "safe" path (clicking **Done**), but `handleOpenChange` tracked the dismissed event on that path too: a 1:1 duplicate of the already-tracked `access_token_done_button_clicked` event, with `step: 'success'`. The event's `tokenType` property was also never meaningful: it is derived from a variable that is only set after a token is actually created and never reset, so on a first-attempt abandonment it is always `'none'` by construction, and on a later abandonment in the same session it carries the *previous* token's type. The sheet also force-closes when the permissions map fails to load. That close was indistinguishable from a user abandonment. Linear: [GROWTH-1196](https://linear.app/supabase/issue/GROWTH-1196/fix-access-token-creation-sheet-dismissed-duplicate-on-done) ## What is the new behavior? - The event no longer fires on **Done**. The token-created step already blocks Escape, outside click, and Cancel, so the event now only reflects the sheet closing before a token exists. - Dropped `tokenType` (never described the abandoned attempt, see above) and `step` (a constant `'form'` once Done stops firing it; it was also typed Numeric project-wide in PostHog, so its string values read as NULL in HogQL). - New properties, read from the live form at close time through a small `useImperativeHandle` ref on `NewScopedTokenForm` (`useForm` ownership stays inside the form component): - `resourceAccess` (`project` | `organization` | `account`): the in-flight scope selection. `account` is the classic-token path, so this carries what `tokenType` was meant to. It is the default `project` when the form is untouched, so filter on `isFormTouched` before reading it as intent. - `formStep` (`form` | `review`): which screen the user was on. The sheet-level `step` never captured this. - `isFormTouched`: whether any field was changed from its default. - `trigger` (`user` | `permissions_load_error`): the forced close on a failed permissions load now fires with its own trigger, so an endpoint regression shows up in the funnel instead of silently lowering `access_token_created`. - Fixed a double-fire on the load-error path: the form's error effect depended on the `onCancel` callback identity and re-ran on the sheet's close re-render (double toast, and a double event). It now reads the callback through a ref and depends only on `isError`, matching the existing `isReducedMotionPreferredRef` pattern in the same file. ## Additional context `NewScopedTokenSheet.test.tsx` asserts: Done does not emit the dismissed event; Cancel and Escape emit it with `trigger: 'user'` and the in-progress `resourceAccess` and `isFormTouched`; dismissing from the review screen reports `formStep: 'review'`; a 500 from the permissions endpoint emits exactly one event with `trigger: 'permissions_load_error'` and closes the sheet. `step` being typed Numeric in PostHog affects every event that sends a string `step`. That is a PostHog data-management fix handled separately, not in this PR. 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_01XW73umv73LrrKxFwwymSaH --------- Co-authored-by: Claude <noreply@anthropic.com> Co-authored-by: Pamela Chia <pamelachiamayyee@gmail.com> |
||
|
Danny WhiteandJoshen Lim
|
a351a36e9b |
refactor(studio): centralise replication payload builders (#49842)
## What kind of change does this PR introduce? Studio data-layer refactor. ## What is the current behavior? Pipeline creation, editing, and validation build similar destination and pipeline payloads separately. The duplicated mappings rely on type assertions and can drift between actions. ## What is the new behavior? Uses shared typed builders for create, update, and validation payloads across the existing destinations. Update payloads continue to omit blank secrets, while create payloads preserve their current values. This PR does not add table partitioning configuration. ## To test This is a data-layer refactor. No visible behaviour should change. 1. Open **Database > Replication** and click **Start a new pipeline**. 2. Select **BigQuery**, or any other enabled destination. 3. Edit a few non-secret fields and expand **Advanced settings**. 4. Confirm the form remains usable and no runtime errors appear. Create, update, validation, and secret-handling behaviour is covered by the focused tests and CI. Deploy previews and fresh local projects do not have the existing destinations or credentials needed to exercise those paths manually. <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Bug Fixes** * Improved replication destination configuration handling during creation, updates, and validation. * Applied consistent configuration mapping across supported destination types. * Ensured blank secret values are omitted during updates while retained when creating destinations. * Standardized table synchronization defaults when no specific setting is provided. * **Tests** * Added coverage for BigQuery configuration mapping and secret handling. * Updated DuckLake tests for destination updates. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Co-authored-by: Joshen Lim <joshenlimek@gmail.com> |
||
|
Danny White
|
9f5b5ea6a7 |
fix(ui): standardise disclosure focus rings (#49660)
## What kind of change does this PR introduce? Bug fix and design-system accessibility update. ## What is the current behavior? Accordion and collapsible triggers do not consistently use the shared focus treatment. Accordion focus can animate from the browser outline, while call sites need to manage radius and focus geometry themselves. ## What is the new behavior? Bare accordion and collapsible triggers use the rounded `focus-inset` treatment by default, positioned 2px inside the control. Interactive children composed with `asChild`, such as a standard `Button`, retain their own focus treatment. Accordion triggers include a small default horizontal hit-area padding (`px-2`) and retain an explicit outer-ring option for standalone controls. The focus indicator is not animated. Focused tests cover the default and opt-out accordion variants. | Before | After | | --- | --- | | <img width="1792" height="384" alt="CleanShot 2026-09-07 at 15 40 23@2x" src="https://github.com/user-attachments/assets/8c8fd99a-3669-465a-a7b2-d272ec92dec7" /> | <img width="1784" height="498" alt="CleanShot 2026-09-07 at 15 39 52@2x" src="https://github.com/user-attachments/assets/7a0b7de1-a4bc-48cf-b9fc-3421618806b9" /> | | <img width="1564" height="360" alt="CleanShot 2026-09-07 at 15 35 11@2x" src="https://github.com/user-attachments/assets/34945180-e671-4dc2-981a-4c683d08f221" /> | <img width="1552" height="378" alt="CleanShot 2026-09-07 at 15 36 19@2x" src="https://github.com/user-attachments/assets/09f2c030-9ffb-4fa5-bf63-5e172326e8ef" /> | ## Visual impact Most call sites should look the same at rest. The exceptions are accordion triggers that did not already set their own horizontal padding: they pick up the new default `px-2`, `rounded-md`, and lose the old blanket `transition-all` on the trigger (chevron rotation is unchanged). | Resting layout change | Why | | --- | --- | | Accordion triggers without an explicit `px-*` | Inherit new default `px-2` and `rounded-md` | | Accordion triggers with their own `px-*` | Unchanged (call-site padding wins) | | Collapsible triggers | Focus ring only; spacing comes from the child element | | Direct `focus-inset` users (table rows, etc.) | Focus ring only | ## To test Use Tab to focus each disclosure trigger. Confirm the focus outline is rounded, sits 2px inside the trigger, and appears immediately without a dark flash. Previews: [design-system](https://design-system-git-dnywh-fixaccordion-focus-rings-supabase.vercel.app) · [studio-staging](https://studio-staging-git-dnywh-fixaccordion-focus-rings-supabase.vercel.app) · [docs](https://docs-git-dnywh-fixaccordion-focus-rings-supabase.vercel.app) · [www](https://zone-www-dot-com-git-dnywh-fixaccordion-focus-rings-supabase.vercel.app) ### Design system - [Accordion](https://design-system-git-dnywh-fixaccordion-focus-rings-supabase.vercel.app/design-system/docs/components/accordion) - [Collapsible](https://design-system-git-dnywh-fixaccordion-focus-rings-supabase.vercel.app/design-system/docs/components/collapsible) ### Studio (focus-only unless noted) - [Account → Access tokens → **New token**](https://studio-staging-git-dnywh-fixaccordion-focus-rings-supabase.vercel.app/dashboard/account/tokens): permission category accordions (already `px-4`) - [Database → Roles](https://studio-staging-git-dnywh-fixaccordion-focus-rings-supabase.vercel.app/dashboard/project/_/database/roles): role row collapsibles - [Observability → Query performance](https://studio-staging-git-dnywh-fixaccordion-focus-rings-supabase.vercel.app/dashboard/project/_/observability/query-performance): index suggestion accordions (already `px-4`) - [Settings → JWT (legacy)](https://studio-staging-git-dnywh-fixaccordion-focus-rings-supabase.vercel.app/dashboard/project/_/settings/jwt/legacy): migration help collapsible ### Docs - Docs mobile nav (narrow viewport): accordion sections **(resting layout: +px-2)** ### WWW - [Pricing → FAQs](https://zone-www-dot-com-git-dnywh-fixaccordion-focus-rings-supabase.vercel.app/pricing): FAQ accordions **(resting layout: +px-2)** - WWW mobile nav (narrow viewport): menu accordions **(resting layout: +px-2)** ### Not worth opening manually Sheets, modals, error fallbacks, and replication/cron/integration advanced-settings accordions inherit the focus fix but need specific flows or failure states to reach. Error-state troubleshooting accordions (`TroubleshootingSections`) also gain **+px-2** when visible. <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit ## New Features - Accordion triggers now use an inset focus style by default, with an option to use the standard ring style. - Accordion and collapsible trigger styling can be customized more flexibly. ## Accessibility - Focus indicators appear consistently without flashing browser defaults. - Focus transitions are limited to visual properties that preserve immediate keyboard focus visibility. ## Documentation - Updated accessibility guidance explains inset focus behavior and transition considerations. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Saxon FletcherandCursor
|
c365549ab8 |
feat(studio): resizable query/results split in the Explorer query editor (#49984)
## What's changed Stack 1/2 (next: https://github.com/supabase/supabase/pull/49985). - `QueryEditor` (`viewport` variant, e.g. Explorer query tabs and assistant query cells) now renders the SQL editor and results in a vertical `ResizablePanelGroup` instead of a fixed `h-[45%]` editor. The split is persisted under `LOCAL_STORAGE_KEYS.EXPLORER_QUERY_SPLIT_SIZE`. The `embedded` variant (notebooks) is unchanged. - Editor and results JSX are extracted into `querySql` / `queryResults` so the two layouts share one definition. - `isRunDisabled` now hides the toolbar run button instead of rendering it disabled (editor shortcuts are still disabled). `AssistantQueryCell` only sets it while an approval is pending, so the run button comes back once the tool call has resolved. - `QueryRunButton`: "Run selected" → "Run selected SQL", plain `DropdownMenuItem` instead of `DropdownMenuItemTooltip`. - `QueryResultError` no longer paints its own table-header background. ## How to test 1. Explorer → open a query tab. Drag the handle between the editor and results; reload — the split size is restored. 2. Toggle "Hide query" / "Show query" — results fill the tab when the editor is hidden. 3. Open a notebook — cells still render with the fixed-height editor (no resizable handle). 4. In the AI Assistant, ask for a query that needs approval. While the approval footer is shown there is no run button in the cell toolbar; after "Run query" / "Skip", the run button appears and works. 5. `pnpm --filter studio exec vitest --run components/interfaces/Explorer` passes. <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit - **User Interface** - Query results now appear in a resizable vertical split view, allowing users to adjust the space allocated to the editor and results. - Updated query result styling provides a cleaner background presentation. - Query result panels are better centered when appropriate. - **Query Execution** - The menu option is now labeled **“Run selected SQL.”** - The run button is hidden when query execution is unavailable. - **AI Assistant** - Query execution is disabled only during the relevant confirmation states. - Assistant query panels now use a wider, full-width layout. - Debugging a query now updates the assistant’s initial input. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Co-authored-by: Cursor <cursoragent@cursor.com> |
||
|
Joshen Lim
|
79964102d3 |
Bump reviewdog action misspell (#50071)
## Context Our `avoid-typos` GHA which uses `reviewdog/action-misspell` is currently failing with a 404 while fetching some files to build the Docker container as per [here](https://github.com/supabase/supabase/actions/runs/34074954209/job/101604888827) for example. ```E: Failed to fetch http://deb.debian.org/debian-security/.../perl-modules-5.32... 404 Not Found``` Happening as v1.26.3's Dockerfile builds on `debian:bullseye-slim`, which is now EOL as of 31st Aug 2026 ([ref](https://www.debian.org/News/2026/20260831)). `bullseye-security` apt mirror has dropped the pinned perl package version, so the action's Docker image can no longer build [v1.28.0](https://github.com/reviewdog/action-misspell/releases/tag/v1.28.0) (Published on 6th Sept) switches the base image to `debian:bookworm-slim`, which resolves this ## Changes involved: - Bumps reviewdog/action-misspell from v1.26.3 to v1.28.0 in `avoid-typos.yml`. <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Chores** * Corrected the pinned revision used by automated spelling checks. * No changes to application functionality or end-user behavior. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Saxon FletcherandCursor
|
04747937a2 |
fix(ui-patterns): align badges and input in the tiny multi-select trigger (#49986)
## What's changed **Before** <img width="467" height="353" alt="image" src="https://github.com/user-attachments/assets/4587d51a-8bbe-4da4-a3dd-dcaac6b17993" /> **After** <img width="541" height="360" alt="image" src="https://github.com/user-attachments/assets/de66b229-54b0-4443-9c49-6f2ff0dd93aa" /> Independent of the Explorer/assistant stack. `MultiSelectorTrigger` with `size="tiny"` had badges and the inline input overflowing the 26px control. The trigger now stretches its children (`items-stretch`), badges are `h-full py-0 leading-none` with a tighter `gap-0.5`, the input/label drop their extra padding and line-height, and the chevron self-centers. Other sizes are unchanged (`isTiny` guards every new class). Also drops a redundant `text-sm` from `MultiSelectorInput`, which `MultiSelectorInputVariants` already sets per size. ## How to test 1. Design system (`pnpm dev:design-system`) or anywhere Studio uses `<MultiSelector size="tiny">` (e.g. Logs filters): with 1–3 values selected the badges sit inside the 26px control with no clipping; the placeholder and chevron are vertically centered. 2. `size="small"` / default still look the same as on `master`. 3. `pnpm --filter ui-patterns exec vitest --run src/multi-select` passes. <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit - **Style** - Improved the compact multi-select control layout, including spacing, alignment, selected-value badges, input sizing, and dropdown indicator positioning. - Adjusted placeholder spacing when the tiny control has no selected values. - **Tests** - Updated coverage to verify compact spacing and sizing when the tiny multi-select control displays a selected value. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Co-authored-by: Cursor <cursoragent@cursor.com> |
||
|
Andrey A.
|
bd43802d7b | docs(self-hosted): add a section about key rotation to docker guide (#50015) | ||
|
Andrey A.
|
8a1adbbd14 | docs(self-hosted): use opaque api keys in self-hosted how-to guides (#50000) | ||
|
Barry Roodt
|
5a673668cb |
chore(studio): update self-hosted MCP server to 0.12.0 (#50007)
## 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? Dependency update. ## What is the current behavior? `apps/studio` depends on `@supabase/mcp-server-supabase` `^0.11.0`, which pulls in `@supabase/mcp-utils` `0.7.0` transitively. ## What is the new behavior? - Bump `@supabase/mcp-server-supabase` to `^0.12.0`. The lockfile moves it to `0.12.0` and its `@supabase/mcp-utils` dep to `0.8.0` (still indirect). Nothing else in the lockfile changes. - Peer deps are unchanged (`@modelcontextprotocol/server ^2.0.0`, `zod ^3.25.0 || ^4.0.0`). No studio code change needed. 0.12.0 adds an optional `costConfirmation` server option for `create_project` / `create_branch`; the self-hosted route doesn't set it, and self-hosted never registers those tools in the first place. The exported tool set is the same 33 schemas, so the tool-name guard in `lib/ai/tools/mcp-tools.ts` still passes. `get_advisors` now groups lints inside its result, which studio forwards to the model without parsing. Release notes: [mcp-server-supabase v0.12.0](https://github.com/supabase/mcp/releases/tag/mcp-server-supabase-v0.12.0) and [mcp-utils v0.8.0](https://github.com/supabase/mcp/releases/tag/mcp-utils-v0.8.0). ## Additional context [AI-1178](https://linear.app/supabase/issue/AI-1178/2b-update-self-hosted-remote-mcp-server) Testing: - `pnpm install --frozen-lockfile` passes. - Studio `pnpm typecheck` is clean. - MCP-related vitest files: 13 files, 108 tests passed. - In-memory smoke of `createSupabaseMcpServer` with the self-hosted route's options reports `serverInfo.version` `0.12.0` and 11 tools. <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Chores** * Updated the Supabase MCP integration dependency to version 0.12.0. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Sean Oliver
|
8c281e53d9 |
fix(studio): drop resource-warnings staleTime to 5 minutes (#49975)
## 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? Bug fix. ## What is the current behavior? The disk space exhaustion banner can stick around for roughly two hours after the user has actually fixed the problem, which is long enough that people act on stale information and change the wrong thing. Two caches stack: - Studio caches resource warnings for an hour: `staleTime: 1000 * 60 * 60` in `useResourceWarningsQuery` - The resource warnings endpoint independently only reports warnings raised in the last 60 minutes, and has no resolved signal, so a warning stops being reported by aging out rather than by being cleared Worst case is that backend hour plus however much of the client hour the user happened to be sitting on. Details in GROWTH-1190, split out of GROWTH-966. ## What is the new behavior? Client `staleTime` drops to 5 minutes, which removes the client-side hour. Worst case goes from about two hours to about one. No change to `refetchOnWindowFocus`, which already defaults to true in `apps/studio/data/query-client.ts`. ## Additional context This deliberately only fixes the client half. The backend resolved signal, and the question of which banner should show for which disk, both stay on GROWTH-966. Testing: unverified in the browser. It is a numeric literal on an existing field, so there is nothing new to typecheck, but I have not reproduced the stale banner locally. <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Bug Fixes** * Resource warning information now refreshes more frequently, helping ensure displayed warnings are more up to date. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Gildas Garcia
|
50a6ebbe6d |
Scoped PAT: improve error handling when showing a token details (#49997)
We already have proper error handling on: - the token list query - token creation/deletion with toasts We had custom error handling on the token permissions sheet. Replaced it with an `AlertError`: <img width="798" height="371" alt="image" src="https://github.com/user-attachments/assets/24145308-b1cd-491f-8f54-0c628dd185ce" /> Question: should we do something about the sheet header when the token couldn't be loaded? <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Bug Fixes** * Improved the access token loading error state with a clearer, consistent error display and “Please try again” guidance. * Added a fallback label of “Unknown” when an access token name is unavailable. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Chase Cresgy
|
ef596f3e12 |
Update humans.txt (#49291)
removal to reflect current employees. ## 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? docs update <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit - **Documentation** - Updated the published team roster to remove two former team members. - The team information now reflects the current roster. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Chase Cresgy
|
1a8ef58f96 |
Offboarding Update for humans.txt (#50010)
Removed Eyal Ehrlich from the humans.txt 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? docs update <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Documentation** * Updated the team roster in the public information file by removing one former team member entry. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
