mirror of
https://github.com/supabase/supabase.git
synced 2026-10-08 19:05:06 +03:00
docs/e2e-nav
471
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
 CharisandJoshen Lim
|
0b97e37ccf |
feat: notebook content schema (#48813)
Related to FE-4109. ## Summary - **API codegen workaround**: Platform API's `notebook` content type hasn't shipped to the OpenAPI spec yet, so `pnpm api:codegen` can't be run. Locally widened `ContentBase.type` to include `'notebook'` (marked with TODO for removal once spec publishes). - **Notebook schema & type system**: Introduced Zod schemas mirroring RFC-defined notebook shape (`schema_version: 1, cells: Cell[]`). Maintains wire/domain boundary (cell `sql` → `unchecked_sql` branded for security). Agent-writable schema for `create_notebook` tool omits cell IDs (backend-generated); future update operations will require them. All TypeScript types are `z.infer`'d from schemas (no hand-written parallel interfaces). - **IsoDateTimeString moved**: Extracted ISO datetime validator from `querySource.ts` to `lib/iso-datetime.ts` (data layer shouldn't import from components layer). Needed by notebook `time_range` fields. ## Test plan - [x] Unit tests: `notebook-schema.test.ts` (9 tests), `iso-datetime.test.ts` (3 tests), `querySource.test.ts` updated and passing (26 tests) - [x] Typecheck: no new errors - [x] Prettier: formatting clean <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit - **New Features** - Added support for validating and processing notebook content, including markdown, database, log cells, time ranges, and chart configurations. - Added compatibility for notebook content types in content handling. - Added reliable ISO date-time validation for notebook data and related features. - **Tests** - Expanded coverage for valid and invalid notebook structures, cell requirements, time ranges, chart settings, and date-time values. <!-- end of auto-generated comment: release notes by coderabbit.ai --> Co-authored-by: Joshen Lim <joshenlimek@gmail.com> |
||
|
|
cddb430310 |
feat(studio): scoped pat root branch (#48384)
## Description This is the Scoped PAT stacked PRs root branch ## How to test ### With the `scopedPAT` enabled (default on staging) Go to https://studio-staging-git-scopedpat-merge-token-lists-supabase.vercel.app/dashboard/account/tokens. - You shouldn't see two tabs anymore - If you had classic tokens, they should have the _Legacy_ badge - You can create scoped tokens - You have a way to copy newly created tokens before closing the form side panel ### With the `scopedPAT` disabled (use the devtool to override) - You shouldn't see two tabs anymore - If you had classic tokens, they should **not** have the _Legacy_ badge - You can create classic tokens - You have a way to copy newly created tokens above the list upon form submission <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Show classic and scoped access tokens together in one list, with classic tokens labeled “Legacy” when the scoped experience is enabled. * Add scoped access token creation with a two-step configure → review → success flow (when enabled). * Add a dismissible migration notice about scoped tokens with a link to API docs. * Show “View permissions” only for scoped tokens. * **Bug Fixes** * Token deletion now supports both classic and scoped tokens with the correct confirmation and success handling. * The scoped tokens page now redirects to the unified access tokens page. * **Accessibility** * Improved accessibility by adding a label to the token “more options” action. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Co-authored-by: Ali Waseem <waseema393@gmail.com> Co-authored-by: kemal.earth <606977+kemaldotearth@users.noreply.github.com> |
||
|
Charis
|
21511042a3 |
feat(studio): assistant logs context and reports guard (#48514)
## 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? Feature — final PR (9/9) of the SQL editor logs-source stack. **Base branch:** `charislam/sql-editor-inline-ai-clickhouse-dialect` (PR 8). Nothing here is user-visible: entry points stay behind `sqlEditorLogsSource` + `otelLegacyLogs`, and flag rollout happens after the whole stack merges. ## What is the current behavior? - The Assistant has no idea a SQL editor snippet targets the logs backend. Ask it about a logs snippet and it answers in Postgres, because the attached query is fenced as ` ```sql ` and nothing tells the model otherwise. - Because the `sql` fence is what `MessageMarkdown` treats as runnable Postgres, an attached ClickHouse query is rendered with a Run-against-Postgres affordance and branded with `untrustedSql`. - "Debug with Assistant" on a failed logs query produces a dialect-less prompt, so both the in-app assistant and the copyable version get debugged as Postgres. - A report referencing a `log_sql` snippet runs its ClickHouse SQL against the user's Postgres database and surfaces the resulting error. ## What is the new behavior? **Assistant panel.** The "Current Query" chip records which backend the attached query targets. That reaches the model two ways: each attachment is fenced with its own dialect (` ```clickhouse ` vs ` ```sql `), and a `containsLogsSnippets` flag rides on the user message as AI SDK `metadata`. The server reads the flag off the conversation and prepends the ClickHouse dialect rules plus the logs schema reference as a non-cached context message. Two design points worth calling out in review: - The flag lives on the **message**, not the request body, so Retry and the tool-approval continuation reproduce the context a message was originally asked in — neither of those passes a per-call body. - It's derived from **what's actually attached**, so detaching the chip drops the claim rather than leaving the two able to disagree. The `clickhouse` fence also keeps a logs query out of `MessageMarkdown`'s `sql` branch, so it's no longer offered as runnable Postgres or branded with `untrustedSql` — a boundary this stack's distinct brands exist to prevent crossing. **Debug flow.** `buildDebugChatArgs` attaches its query with a source for the same reason, and names the dialect in the prompt text so the copyable version stands on its own outside the app. **Reports.** A report only stores a snippet id, so whether it queries the logs backend is only knowable once the content loads. `ReportBlock` guards on the fetched type and renders a `LogsSnippetReportBlock` placeholder instead of executing. Double-guarded: no `sql` for a logs snippet (so it's out of the query key and `queryFn` short-circuits even on an explicit `refetch`) and `enabled` excludes it. **Incidental cleanups.** `buildAssistantContextMessages` extracted out of `generate-assistant-response`; a schema-access sentinel that was duplicated as a string literal across two files (and compared against) replaced with one exported constant; `SqlSnippet` deduplicated to a single declaration; `resolveSnippetSource` / `isLogsSource` shared instead of re-implemented per surface. **Tests.** 4 new/extended suites. Notable cases pinned: a message with no metadata must validate (`safeValidateUIMessages` applies `metadataSchema` to *every* message, so a required schema would 400 every existing conversation); only *user* messages count, so a model reply can't talk the server into a different dialect; a mixed-attachment message is flagged without overclaiming a single source; and `ReportBlock` registers no pg-meta mock for the logs cases, so an unhandled request failing the test *is* the assertion that logs SQL never reaches Postgres. Verified: `pnpm typecheck`, `lint:ratchet` (no regression), Prettier, and the full Studio suite (459 files / 4969 tests). ## Additional context <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit - **New Features** - Added support for recognizing log snippets in reports, with clear guidance to open them in the SQL editor or remove them. - AI Assistant now understands log snippets and provides ClickHouse-specific context, formatting, and troubleshooting guidance. - Snippets retain their source information when shared with the AI Assistant. - **Bug Fixes** - Prevented unsupported log snippets from being executed as regular database queries. - Improved source detection when opening snippets directly from links. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Charis
|
8b38e0d1ed |
feat(studio): ClickHouse dialect for logs snippet AI + rewrite to ClickHouse (#48501)
## 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? Feature, plus a refactor of the shared logs-rewrite flow. PR 8 of the SQL editor query-source series. Stacked on #48457 — review that one first, and merge this after it. ## What is the current behavior? A `log_sql` snippet runs against the ClickHouse-backed analytics endpoint, but the SQL editor's AI still writes Postgres: inline edits get Postgres system prompts, and the result is run through `sql-formatter`, which mangles ClickHouse backticks and `log_attributes` map lookups. Legacy Logs Explorer saved queries open in the editor as `log_sql` snippets. Those are BigQuery dialect and error against the ClickHouse endpoint the editor runs them on, with no in-editor way out — only the Logs Explorer offered a rewrite. The completion route was also asymmetric. It assembled a schema/code/instruction message for Postgres but forwarded `prompt` verbatim for ClickHouse, so a client wanting ClickHouse had to hand-build the equivalent string. ## What is the new behavior? **Inline AI speaks ClickHouse for logs snippets.** `sqlSourceToDialect` maps a snippet's source to `postgres`/`clickhouse` and `buildCompletionRequestBody` threads it through. For ClickHouse, `useSqlEditorAi` strips code fences from the response and skips `formatSql`. Execution and dialect both follow the snippet type, so a snippet's valid dialect never flips. **Rewrite to ClickHouse in the editor.** A banner offers the rewrite for a logs snippet whose text trips `looksLikeLegacyLogsQuery`, and proposes the result through the editor's existing AI diff view rather than replacing the snippet, so it's accepted or discarded like any other AI edit. Gated on `otelLegacyLogs`: on a non-migrated org the BigQuery text is still correct, so rewriting it would break a working query. The offer is a state machine (`offered` / `rewriting` / `failed` / `noRewriteNeeded` / `dismissed`) with a declarative table of valid transitions, so the states are mutually exclusive by construction and dismissal is terminal. A failure keeps its message and offers a retry; a response identical to the input is reported rather than opening an empty diff. **One place assembles completion prompts.** The route now uses a single template for both dialects, branching only the schema section and — for `intent: 'rewrite'` — the instruction. `lib/ai/clickhouse-logs.ts` is the single home for ClickHouse-logs prompt content, replacing two independently maintained descriptions of the same table. Clients carry no prompt text. **The rewrite flow is shared with the Logs Explorer.** Both surfaces previously hand-rolled the same sequence and had drifted: only one detected a no-op rewrite, they sourced `log_attributes` keys differently, and the Explorer formatted errors with an `as Error` cast. Both now use `useLegacyLogsRewrite` and the same state-driven banner, so the Explorer picks up no-op detection and typed error extraction. **Attribute keys are fetched on submit, not while typing.** The detected source would otherwise feed a reactive query key, making every edit that changed it cost another network call. `useLogsAttributeKeys` is imperative and goes through `queryClient.fetchQuery`, so a source already cached — including by the Explorer header and query panel, which subscribe reactively — is reused. This also closes a gap where inline edits never received keys at all, unlike full rewrites. `getErrorMessage` gains an optional typed fallback and no longer stringifies a bare object into `'[object Object]'`; every existing caller already hand-rolled a fallback, except `QueueSettings`, which interpolated the raw result and now passes one. Nothing here is user-visible until the `sqlEditorLogsSource` flag is enabled. Tests: dialect selection and request-body shape, the ClickHouse prompt content (including that the schema section does not restate the dialect rules), the reducer's valid and invalid transitions, `shouldOfferLegacyLogsRewrite`, on-submit key discovery with cache reuse, and `getErrorMessage`. ## Additional context <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Added an Assistant banner to help rewrite legacy BigQuery-style logs queries into ClickHouse SQL. * SQL assistance now adapts to the selected query type, including relevant log attribute context. * Rewrite suggestions can be reviewed as editor diffs before being applied. * **Bug Fixes** * Improved rewrite failure handling, retry options, dismissal behavior, and “no rewrite needed” messaging. * Error notifications now provide a clearer fallback message when details are unavailable. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Joshen Lim
|
a0cec24f49 |
Remove references to fly (#48648)
## Context As per PR title - should not have any visual nor functional change <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit - **Changes** - Standardized infrastructure, region, and database configuration around AWS-based environments. - Removed Fly.io-specific region and provider options from project creation, instance sizing, and infrastructure settings. - Enabled disk validation, spend-cap eligibility, backup restoration, and extension setup consistently across supported projects. - Updated billing and region displays to use the applicable AWS configuration. - **Bug Fixes** - Corrected project-specific restrictions that could incorrectly hide configuration and billing controls. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Danny White
|
73c9dbfa52 |
fix(studio): standardise custom icon weight (#48478)
## What kind of change does this PR introduce? Bug fix and internal tooling update. Resolves FE-3472. ## What is the current behavior? Custom Studio icons use inconsistent source stroke widths, and some child-level styling prevents component props from overriding them. Mixed custom and Lucide icon sets can therefore appear uneven. ## What is the new behavior? Custom stroke icons use a root-level `stroke-width="1.5"`; fill-only logos use `stroke="none"`. The build validates that contract and regenerated components preserve existing exports and props. Studio applies the same `1.5` weight across Reports categories and uses one shared destination icon mapping in the replication selector, destination rows and diagram. | Before | After | | --- | --- | | <img width="418" height="516" alt="56398" src="https://github.com/user-attachments/assets/6afa7042-e6be-40e7-9911-af2f61238c9d" /> | <img width="390" height="550" alt="CleanShot 2026-07-30 at 17 12 37@2x" src="https://github.com/user-attachments/assets/870f49cf-c8fa-40db-8be8-2eb5f264ff4a" /> | | <img width="510" height="734" alt="CleanShot 2026-07-30 at 17 19 28@2x" src="https://github.com/user-attachments/assets/a5b2c088-dcd2-4907-976b-5820794d06e3" /> | <img width="554" height="742" alt="CleanShot 2026-07-30 at 17 16 06@2x" src="https://github.com/user-attachments/assets/ed3a77c4-5d94-4ca7-b9e4-1403b725a981" /> | ## Testing At 100% zoom, compare custom and Lucide icon weight in: - Reports: **Add your first chart** and **Add block** - Database > Replication: the destination selector, destination rows and replication diagram - Command menu (`⌘K`): **Search Database Tables**, **Search RLS Policies**, **Search Edge Functions** and **Search Storage** - Authentication > Users: right-click a user row and compare the context-menu icons - Database > Schema Visualizer: open a table node overflow menu - A paused project: **Export your data > Download backups** <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Added consistent destination icons across replication panels, rows, and diagrams. * Updated instance health and metric icons for clearer identification. * Standardized icon stroke weight and reduced default icon stroke thickness. * **Documentation** * Clarified custom icon requirements, default properties, and validation guidance. * **Bug Fixes** * Improved consistency of icon rendering across replication destinations and reports. * **Tests** * Added coverage for icon SVG validation and replication destination icon rendering. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Joshen Lim
|
2c53ca4a79 |
Support listing and reading custom reports from Assistant (#48530)
## Context This is pre-requisite work for adding support to managing custom reports from the Assistant. Planning to break this into a number of PRs, briefly - Adding read support for custom reports - Adding write support for custom reports - Adding run support for custom reports - Should be able to infer data from the results then This PR starts with adding support for listing and reading custom reports from the Assistant ## Other changes involved - Updates setting up of the home page report to have better title and description - Swaps the variant of the ToggleGroup in the SQL block for custom reports as the default variant blends into the background color of the PopoverContent ## To test - [ ] Assistant should be able to list custom reports + read its contents <img width="428" height="755" alt="image" src="https://github.com/user-attachments/assets/6a15b660-c0ee-4a06-984c-87eff3943eec" /> <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit - **New Features** - Added AI-assisted tools to list reports and retrieve report details, including chart counts, layouts, configurations, and SQL-backed chart information. - Added clearer empty-state messaging when no snippets are available. - **Improvements** - New homepage reports now use the name “Homepage Report” and include a descriptive project-home summary. - Updated query controls with refreshed visual styling. - Improved content requests to support additional request context. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Sean Oliver
|
e88a3723e1 |
feat(studio): add PostHog session replay with masked-by-default policy (#48515)
## 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? Telemetry feature. ## What is the current behavior? - Session replay is off, and nothing in the code keeps it off. - `packages/common/posthog-client.ts` sets no recording config at all. - So PostHog's project setting alone decides, for every app sharing that project. - Studio, www and docs share one project. - Studio shows customer data almost everywhere: SQL editor, table rows, connection strings, API keys. - posthog-js masks inputs by default. It does not mask rendered text. - [GROWTH-1055](https://linear.app/supabase/issue/GROWTH-1055) ## What is the new behavior? - `posthogClient.init()` takes a masking config, and disables recording when it gets none. - Studio passes one behind `NEXT_PUBLIC_POSTHOG_SESSION_REPLAY`. - Every other app passes nothing, so it never loads the recorder. - Studio masks all text and all inputs. - `data-ph-capture="true"` opts one element's text back in. Unused so far. - Canvas is blocked, because it records as images that text masking cannot reach. - Query strings and fragments are stripped from recorded URLs, where auth callbacks carry tokens. - Request and response bodies are never recorded. - Console logs are never recorded, since masking only reaches DOM text. - Masking is set in code, so PostHog's settings cannot loosen it. - Consent gating is unchanged. Nothing records before a user accepts. ## Additional context - Recording needs three things: this env var, the PostHog project toggle, and user consent. - All three are off or unset, so merging this changes nothing at runtime. - `NEXT_PUBLIC_POSTHOG_SESSION_REPLAY` goes into Vercel on Preview scope first, to test on a preview build. - Production scope comes later, once we are ready to record there. - `NEXT_PUBLIC_*` is inlined at build time, so each scope needs a rebuild afterwards. - Text inside HTML attributes (`title`, `alt`, `href`) is still recorded as-is. - posthog-js exposes no hook for masking attributes, so covering it needs `ph-no-capture` per component. - Staging has no server-side masking config, so that is where this gets verified. - Plan: enable recording on staging, verify masked text on a preview, then decide on production. - Network timing stays on for the dashboard performance work. Payloads stay off. - Tests cover both masking functions and the config values. ## Screenshots https://github.com/user-attachments/assets/aa064a04-f977-4453-a3da-2fe0cdcead08 <img width="889" height="651" alt="CleanShot 2026-07-31 at 10 13 43" src="https://github.com/user-attachments/assets/f1d07946-fd68-42b2-89f1-d201bc605638" /> <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit ## Summary by CodeRabbit * **New Features** * Added privacy-focused session replay for Studio. * Text and form inputs are masked by default, with explicit opt-in capture. * Network recordings remove query strings and fragments. * Headers, request bodies, canvas data, and console logs are excluded. * **Bug Fixes** * Improved whitespace and capture-attribute handling during masking. * Session replay remains disabled without a masking policy or explicit enablement. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Danny WhiteandJoshen Lim
|
69570a357d |
fix(studio): route vercel deploy-button params to create despite marketplace source (#48258)
## What kind of change does this PR introduce? Bug fix for the Vercel Deploy Button → Studio handoff. ## What is the current behavior? Vercel sometimes opens our install popup with `source=marketplace` while still sending Deploy Button params (`currentProjectId`, `external-id`). We trust `source` alone, so users are routed to choose-project (connect) instead of create — which is why create never gets reached in the Deploy Button flow. ## What is the new behavior? - When both Deploy Button signals (`currentProjectId` + `externalId`) are present, route to create even if Vercel sent `source=marketplace` / `external` - Hide Skip (and related empty-state copy) on choose-project when those signals are present, so Deploy Button users can't continue without linking ## Additional context Stacked on #48230. Test plan: - [ ] Unit tests for `resolveVercelInstallSource` / `hasVercelDeployButtonSignals` pass - [ ] Deploy Button flow with mislabeled `source=marketplace` + both params → lands on create after org install/continue - [ ] Genuine marketplace install (no `currentProjectId`/`external-id`) → still lands on choose-project with Skip available - [ ] If choose-project is opened with both Deploy Button params, Skip is hidden <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit - **New Features** - Improved Vercel installation handling for Deploy Button workflows, ensuring the correct setup path is selected. - Added clearer project-connection guidance when no projects are available (including conditional skip copy). - **Bug Fixes** - Prevented Deploy Button installations from incorrectly offering a skip option. - Preserved the skip-and-connect-later guidance for other Vercel installation flows. - Improved recognition of Deploy Button installations even when the reported Vercel source differs. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Co-authored-by: Joshen Lim <joshenlimek@gmail.com> |
||
|
Crispy
|
8c092185ae |
feat: paused project restore window copy and backup downloads (#48279)
Shows the restore deadline as a date on the paused-project screens, and offers backup downloads while a paused project is still restorable (previously only after the restore window ended). Depends on a backend change — keep as draft until that is live. <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Extended the paused-project restore window from 90 days to up to 1 year. * Added clearer, downloadable options for database backups and storage objects while a project is paused. * Paused-project screens now show an “available until” date when applicable (and updated resume guidance). * **Documentation** * Updated platform and troubleshooting guides to reflect the new 1-year restore window and post-window recovery limitations. * **Bug Fixes** * Standardized restore-window wording across the pause confirmation and paused-state UI. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Joshen LimandDanny White
|
7f42765070 |
Joshen/fe 3983 no way to create a new project in vercel integration when (#48230)
## Context For the Vercel integration flow (e.g "Deploy with Vercel" button on GH) If an organization has no projects, there currently isn't a way to create a project and connect it in the same session - users can only hit "Skip". This addresses that by directing users to the /deploy-button/new-project route in this scenario <img width="505" height="539" alt="image" src="https://github.com/user-attachments/assets/6cc85030-42c7-4e58-b4b3-cb8ac0f5da9e" /> ## Other changes involved - Also separates `ProjectLinker` into smaller components - preference for avoiding declaration of components within a component ## To test I'm not sure if this can be tested on staging to be honest, but otherwise we can give it a go on production after the changes are through, as this doesn't change any existing logic to the usual "Connect project" flow I did try clicking the "Deploy with Vercel" button on a repo, and just changing the URL to the staging URL at the Supabase step - seems to work <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit ## Summary * **UI Improvements** * Streamlined the Vercel/GitHub project-linking step while keeping the same create/connect/skip flow, including the searchable project picker, branding/status indicators, and the feature-flagged “create new project” option. * On the Vercel choose-project step, the default selection now reflects the current project context. * **Bug Fixes / Tests** * Improved Vercel install routing query handling to preserve deploy-button configuration when present, with updated automated test coverage. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Co-authored-by: Danny White <3104761+dnywh@users.noreply.github.com> |
||
|
Ivan Vasilov
|
efed6b5c6a |
chore: Server monaco editor from Cloudflare (#47973)
How to test: - Editor should load and work in the SQL Editor - Search for JS bundles `https://cdnjs.cloudflare.com/*` in the preview to verify that's it's loaded from a CDN - Check in the local build whether the Monaco editor is loaded from the base URL (have to do it locally, SQL Editor doesn't work on the self-hosted Vercel deployment) <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Performance** * Improved delivery of Studio assets through CDN-based URLs in supported environments. * Updated the Monaco editor to load assets from the configured CDN when available, with a local fallback. * **Reliability** * Added support for explicitly enabling or disabling CDN asset delivery across environments. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Ali Waseem
|
da7a10be6b |
chore: simplify CPU messaging for compute sizes (#48109)
## Summary - Simplify CPU messaging on the Compute and Disk docs page and in Studio's compute size UI to keep it generic rather than architecture-specific. ## Test plan - [x] Unit tests pass - [x] Typecheck passes - [x] Lint passes <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Updates** * Simplified compute size labels across the UI by removing cloud-provider architecture details from CPU text. * Standardized CPU descriptions to show core counts and whether resources are shared or dedicated. * Updated the “Compute Size” pricing/specs table in the compute & disk guide to use generic CPU labels while keeping pricing, memory, and database size guidance the same. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Vaibhav
|
05d8dd356c | fix: selfhost edge URLs (#47861) | ||
|
Charis
|
4df3c86e93 |
feat: add admin studio link (#47976)
Add link to Admin Studio in top bar for internal use only. Resolves FE-3864 <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Added a conditional Admin Studio shortcut to the desktop project header. * The shortcut opens the relevant project in Admin Studio and is available only when configured and enabled. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Katerina Skroumpelou
|
069051b032 |
test(studio): cover Supabase API key formats in project client tests (#47960)
Adds a contract test that runs `createProjectSupabaseClient()` against the real `@supabase/supabase-js` for each Supabase API key format (temporary, publishable, secret, legacy JWT), asserting that client construction succeeds. Also replaces the placeholder key fixtures in the existing unit tests with realistically shaped ones. Previously the tests mocked the SDK entirely and used keys that don't resemble real formats, so an SDK version that rejects a valid key at construction (as `@supabase/supabase-js` 2.110.4 and 2.110.5 did, reverted in #47945 and fixed in supabase/supabase-js#2526) passed CI unnoticed; with this test, such a regression fails on the dependency bump PR itself. <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Tests** * Added coverage confirming project clients work with temporary, publishable, secret, and legacy API key formats. * Updated test scenarios to use realistic temporary API key values. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Pamela Chia
|
1b1b1ea0e0 |
chore: remove dead telemetry code (#47950)
## Summary Removes two pieces of dead telemetry code found while root-causing the docs pageview re-fire investigation (GROWTH-997, closed with no fix needed). Pure deletion, 30 lines, no behavior change. ## Changes - Delete `apps/www/app/ConsentWrapper.tsx`: an unwired third `PageTelemetry` mount. www already mounts `PageTelemetry` in `pages/_app.tsx` (Pages Router) and `app/providers.tsx` (App Router); nothing imports this wrapper. - Remove the exported `POSTHOG_URL` constant from `apps/studio/lib/constants/index.ts`: zero consumers. The CSP allowlist in `apps/studio/csp.ts` defines and uses its own local `POSTHOG_URL`, which stays. ## Testing Deadness verified before deletion, on current master: - [x] Repo-wide grep for `ConsentWrapper`: only self-references inside the deleted file - [x] Repo-wide grep for `POSTHOG_URL`: remaining references are the `csp.ts` local const only ## Linear - fixes GROWTH-1002 <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Bug Fixes** * Removed an obsolete consent-related page wrapper to streamline page behavior. * Updated application configuration handling without changing existing payment or usage settings. * **Refactor** * Simplified internal configuration and page composition while preserving the existing user experience. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Pamela ChiaandAlaister Young
|
9858562b8b |
fix(telemetry): dedupe funnel toast error events (#47802)
## Summary Since #47293, an API failure on a signup / org-creation / project-creation form emitted `dashboard_error_created` twice: `useTrackFunnelError` fired the origin-tagged event and the global `ToastErrorTracker` independently fired the legacy untagged `source:'toast'` event for the same toast, each behind its own 10% sampling draw. I verified the twin rate empirically at 8-11% of origin-tagged funnel toasts, exactly the floor for two independent 10% draws, meaning the twin co-fires for effectively every funnel error ([Hex thread](https://app.hex.tech/supabase/thread/019f3bc1-3a5c-7200-9122-8e3439bfbe8c)). Any consumer counting funnel errors without an `origin IS NOT NULL` filter saw ~2x inflation. The fix makes `ToastErrorTracker` the sole emitter of `source:'toast'` events, so the duplicate is unrepresentable rather than suppressed. Funnel call sites pass the id returned by `toast.error()` into `trackFunnelError`, which registers the funnel properties against that toast id instead of firing its own event – the tracker then emits a single `dashboard_error_created` enriched with `origin` / `errorCategory` / `errorReason` / `errorCode` for registered toasts, and the plain untagged event otherwise. The `'toast'` overload of `trackFunnelError` requires the toast id, so a missed pairing is a compile error rather than a silent double count. Registration is unconditional and there's only one sampling draw, so suppression can't lose a sampling race. `'form'`-sourced funnel events are unchanged. ## Changes - `lib/toast-errors.tsx`: toast-id → funnel-properties registry (`registerFunnelErrorToast`); `ToastErrorTracker` emits one (optionally enriched) event per error toast under a single 10% draw, deleting entries once consumed - `lib/telemetry/use-track-funnel-error.ts`: overloaded signature – `'toast'` requires the id returned by `toast.error()` (type-enforced), `'form'` keeps direct emission with its own sampling - Update the 7 funnel `toast.error` call sites in `NewOrgForm`, `SignUpForm`, and `pages/new/[slug]` to pass the toast id - Component tests for the tracker (previously uncovered), including an end-to-end test through `useTrackFunnelError` - Code hygiene (also flagged by CodeRabbit): all four `dashboard_error_created` emitters (toast, form, `AlertError`, `ErrorMatcher`) independently encoded the 10% draw – downstream analysis assumes a uniform sampling multiplier across sources, so one site drifting would silently skew comparisons. The rate and the draw now live in one place (`isDashboardErrorSampled()` in `lib/telemetry/error-sampling.ts`). No behavior change. - Mount `ToastErrorTracker` in the TanStack root (`routes/__root.tsx`), mirroring `pages/_app.tsx`. The TanStack tree mounted `Toaster` but never the tracker, so untagged toast error telemetry has never fired in that flavour – and with the tracker now the sole emitter, the missing mount would have silently dropped funnel toast events there too. Side effect once the TanStack flavour ships: untagged `source:'toast'` volume from it goes from zero to normal. ## Testing Component-tested (`apps/studio/lib/toast-errors.test.tsx`): - [x] Unregistered error toast fires exactly one untagged `dashboard_error_created {source:'toast'}` - [x] Registered funnel toast fires exactly one event, enriched with `origin`/`errorCategory`/`errorReason`/`errorCode` - [x] `useTrackFunnelError` with a toast id routes through the tracker as a single enriched event - [x] Non-error toasts ignored; the 10% sampling gate still applies Full Studio unit suite passes (392 files / 4371 tests), plus typecheck and lint. Also verified end-to-end in a local browser (TanStack flavour, sample rate temporarily forced to 1): a failed signup produced exactly one `dashboard_error_created` with `{source:'toast', origin:'signup', errorCategory:'api', errorReason:'email_already_registered', errorCode:403}` and no untagged twin (two independent trials); an unregistered error toast produced exactly one plain `{source:'toast'}`; a client-side validation failure produced exactly one `{source:'form', origin:'signup', errorCategory:'validation', errorReason:'email_invalid'}`; success toasts produced nothing. Post-deploy I'll re-run the twin-rate query from the Hex thread; the untagged-twin rate on funnel pages should decay to ~0 as stale bundles reload over 2-3 days. ## Notes - Origin-tagged funnel toast events now ride the tracker's single 10% draw instead of their own independent draw – statistically identical volume, but the event fires on the tracker's next effect rather than synchronously at the call site (irrelevant for PostHog) - Registration must happen in the same synchronous block as `toast.error()` (documented on the `TrackFunnelError` type) – all current call sites comply - The invalid Postgres version toast in `pages/new/[slug].tsx` (~line 416) needs no special-casing: unregistered toasts keep the plain untagged event, so its telemetry is preserved - Heads-up for `dashboard_error_created` consumers: overall untagged `source:'toast'` volume will dip slightly after this deploys, since funnel-page twins disappear. A volume monitor seeing that drop is this fix landing, not a tracking regression (same class as the intended GROWTH-893 sampling-unification drop). ## Linear - fixes GROWTH-965 <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Bug Fixes** * Enhanced error telemetry for organization creation, sign-up, payment, and project-creation flows by associating failures with toast identifiers and enriched funnel context. * Standardized dashboard error sampling logic across error handling components for consistency. * **Tests** * Added comprehensive test coverage for toast error tracking, including funnel registration, deduplication, filtering, and sampling behavior. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Co-authored-by: Alaister Young <10985857+alaister@users.noreply.github.com> |
||
|
Joshen Lim
|
dc23320e43 |
Add sentry capture exception to apiWrapper (#47804)
## Context As per PR title - also adjusts the imports for files consuming `apiWrapper` to remove the default export for `apiWrapper` Have tested locally by throwing an error in one of the API routes - verified that the event shows up on Sentry <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Bug Fixes** * API errors are now captured in Sentry before returning server error responses, improving production visibility while keeping endpoint behavior the same. * **Tests** * Added coverage to confirm rejected handler executions are reported to Sentry and return the expected HTTP 500 JSON payload. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Cemal Kılıç
|
3667601895 |
feat(studio): add sign in with ChatGPT (#47772)
## 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? Feature ## Summary Introduce a "Sign in with ChatGPT" option gated by the new `dashboard_auth:sign_in_with_chatgpt` feature flag and a manual localStorage rollout switch (`SIGN_IN_CHATGPT_ENABLED`), since the feature is still WIP. <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Added support for signing in with ChatGPT alongside GitHub. * ChatGPT sign-in now depends on both a feature flag and an additional rollout setting. * Updated provider availability so the app can show the correct sign-in options. * **Bug Fixes** * Improved validation and coverage to ensure sign-in options appear only when fully enabled. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Alaister YoungandAlaister Young
|
74bc0a8e27 |
fix(studio): initialize Sentry on the TanStack build (captures were silent no-ops) (#47666)
Stacked on #47657 (base is `alaister/tanstack-migration-fixes`; retarget to `master` once that merges). The TanStack runtime never ran `Sentry.init` — `instrumentation-client.ts` is a Next-convention file nothing imports under TanStack Start, so every `Sentry.captureException` on that build (including the `routes/__root.tsx` error-boundary / `routerErrorComponent` reports) was a silent no-op. - **Shared config source**: the entire client config moves verbatim from `instrumentation-client.ts` into `lib/sentry-client-options.ts` (`buildSentryClientOptions`). Both runtimes build from it, so Next and TanStack can't drift — the builds differ only in two explicit knobs. - **TanStack init**: `sentry.tanstack.ts` initializes `@sentry/react` from `getRouter()` (TanStack Start's real client bootstrap — the earliest point with the router instance), wiring `tanstackRouterBrowserTracingIntegration(router)`. Window-guarded + idempotent; `router.tsx` is TanStack-only so the Next build is untouched. (Named without `.client.` — Start's import-protection fails the build for `*.client.*` in the server graph.) - **Third-party error filter is intentionally Next-only**: without the bundler-injected `applicationKey` metadata (only `withSentryConfig` provides it), the SDK tags *every* event `third_party_code: true` and `beforeSend` would drop them all — recreating the silent no-op with a DSN set. Follow-up: add `@sentry/vite-plugin` moduleMetadata, then enable. - **DSN-less builds stay crash-free**: `vite.config.ts` inlines `undefined` for unset `NEXT_PUBLIC_SENTRY_DSN`/`NEXT_PUBLIC_SENTRY_ENVIRONMENT` (a literal `process.env.*` in the bundle is the exact `process is not defined` class #47657 fixed). No-DSN → disabled client, plus the existing `IS_PLATFORM`/consent gates. - Tests: `instrumentation-client.test.ts` moved to `lib/sentry-client-options.test.ts` with all 36 assertions kept, plus integration-gating and Next/TanStack parity tests. `tsc` clean; full `vite build --mode test` passes. Follow-up (separate): server-side Sentry for the Start handler (`server.ts` entry + `@sentry/node`-style init). ## To test - **Locally (no DSN set)**: load the TanStack build — no Sentry network requests, no console errors, and crucially no `ReferenceError: process is not defined` (the define fallback). Forcing an error must not POST to any `/envelope` endpoint. - **On a preview/deploy (DSN set, telemetry consent accepted)**: throw a test error (e.g. crash a route component) → a POST to `o…ingest.sentry.io/api/…/envelope/` fires, and the event lands in Sentry with a `codeSampleRate` tag and **no** `third_party_code` tag. Navigation spans named after TanStack routes appear when the 2% pageload trace samples in. - **Next build regression check**: the Next dev/preview still reports errors exactly as before (`instrumentation-client.ts` now builds its options from the same shared source). --- ### Review feedback: Sentry `/envelope` never fires on TanStack (Joshen) Root-caused: `@sentry/core`'s `Client.sendSession` silently drops the session when the client has no `release`. The Next build gets a release injected by `withSentryConfig` (the Vercel commit SHA); the Vite build runs no Sentry bundler plugin, so it had no release → session envelopes were discarded before transport → zero `/envelope` traffic (errors/transactions are separate). Fix: inject `release: NEXT_PUBLIC_VERCEL_GIT_COMMIT_SHA` on the TanStack build (vite.config re-exposes `VERCEL_GIT_COMMIT_SHA` under the `NEXT_PUBLIC_` name, same SHA the Next release resolves to). Also switched `integrations` to the function form so defaults are preserved by contract (not just by current SDK behavior). 45 unit tests green. **To test (deploys only — the SHA is unset locally, so this can't be reproduced on a local dev build):** on this PR's Vercel preview with a DSN + telemetry consent, load any page and watch the Network tab for a POST to `…ingest.sentry.io/…/envelope/` — a session envelope should now fire on load, matching the Next build. <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Improved client-side error and performance monitoring for the Studio app across both router setups. * Added support for passing release/version information into monitoring data. * **Bug Fixes** * Reduced noisy error reporting by better filtering common browser, extension, cancellation, and load-related issues. * Prevented browser bundles from referencing missing environment values at runtime. * Made monitoring initialization safer in server-rendered and client-only environments. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Co-authored-by: Alaister Young <10985857+alaister@users.noreply.github.com> |
||
|
Joshen Lim
|
4901f081e5 |
Migrate remaining requests to pg-meta API to use query endpoint (#47758)
## Context Migrates the remaining API requests to the pg-meta endpoint to use the query endpoint directly with the SQL from the pg-meta package. This touches the following: - policies - publications - triggers - views - materialized views - types ## To test Just need to verify that we're still fetching the data correctly on these pages - Database policies - Database publications - Database triggers - Database tables (views + materialized views) - Database types <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Bug Fixes** * Improved and stabilized loading of database metadata (views, triggers, RLS policies, publications, materialized views, and enum types), including more reliable schema-scoped filtering. * Updated policy loading behavior and related UI queries to consistently use schema arrays, improving cache correctness and consistency. * **Tests** * Updated end-to-end test synchronization to wait for the correct metadata responses using more specific request identifiers. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Carel de WaalandAlaister Young
|
0acc0eb8b3 |
feat: Support Form - Sync AI assistant conversation to Front (#46778)
# Sync AI assistant conversation to Front ## What & why When a user submits a support ticket, an AI assistant chat opens so they get help immediately while waiting for a human agent. This PR mirrors every turn of that chat into the Front conversation the support form already created, so the support team sees the full context and Front automations (routing, emails, CSAT) can act on it. Studio holds no Front credentials — it calls the platform endpoints (see the platform PR) to do the syncing. The assistant card is gated behind the `supportAssistantFollowUp` ConfigCat flag. ## How it works 1. **Submit** — `SupportFormV3` generates a stable `threadRef` (via the `uuid` package — `crypto.randomUUID()` is `undefined` in insecure contexts like non-localhost HTTP and would throw, silently aborting the submit) and sends it on `/platform/feedback/send`. The response returns the Front `conversationId`. Both are stored on `SubmittedSupportRequest`. 2. **Open chat** — `SupportAssistantSuccessCardContent` opens a chat seeded with `supportMetadata` (`threadRef`, `frontConversationId`, subject, category, severity, …). The first message is a `<support>…</support>` XML block. 3. **First user message** — the chat is tagged `isSupportChat = true`; the `onFinish` hook fires `syncSupportChatToFront`. 4. **Subsequent turns** — each `onFinish` slices the unsynced delta, strips the XML metadata block from the seed message, and posts to the platform messages endpoint. 5. **Escalation / resolve** — the `escalate_to_human` / `resolve_support_conversation` tools (and manual **Escalate**/**Resolve** buttons in the assistant input) flip lifecycle status via `setSupportLifecycleStatus` → `syncSupportLifecycleToFront`, which calls the escalation/resolve endpoints. Front rules act on `ai_support_status`. The assistant only resolves after the user explicitly confirms the issue is fixed. ## Key design decisions - **`threadRef` as the shared key** — one UUID travels as `threadRef` on submit and as `chatId` on every sync, so all messages thread into a single Front conversation. - **`conversationId` from the form response** — passed to all sync/lifecycle calls so the platform skips lazy derivation and PATCHes custom fields directly. - **Delta-only sync** — `lastSyncedMessageCount` tracks what's been sent; the boundary is snapshotted before the async call to avoid skipping messages that arrive mid-flight. - **Server-side de-dup** — stable `external_id` (`chatId:msg.id`) means retries don't duplicate in Front. - **Fire-and-forget** — sync failures log to Sentry, never break the chat; `isSyncing` resets on rehydration so the next `onFinish` retries the same delta. Message and lifecycle syncs use separate guards (`isSyncing` / `isLifecycleSyncing`) so an in-flight message sync can't drop an escalate/resolve. - **Lifecycle queued until the conversation exists** — if a lifecycle transition is requested before the initial message sync has returned a `frontConversationId`, it's stored as `pendingLifecycleStatus` and flushed once the id is assigned, rather than dropped. - **Tools return immediately** — the lifecycle tools return a stub to the AI SDK; the real Front call happens in `onFinish`, keeping async I/O out of the tool execute path. - **XML seed stripped before sync** — only the user's actual `<message>` is sent to Front (or dropped entirely if the form already created the conversation). ## Changes | Area | File(s) | | --- | --- | | Support form state | `SupportForm.state.ts` — `threadRef` / `frontConversationId` on `SubmittedSupportRequest` | | Support form submit | `support-ticket-send.ts` — sends `threadRef`, reads `conversationId` | | Support form UI | `SupportFormV3.tsx` — generates `threadRef`, stores `conversationId` | | AI assistant state | `ai-assistant-state.tsx` — `SupportChatMetadata`, `setSupportLifecycleStatus`, `onFinish` wiring, tool handling | | Message sync | `state/ai-chat-front-sync.ts` — delta tracking, message filtering, initial vs. incremental | | API data layer | `data/feedback/ai-chat-front-sync.ts` — typed platform-client wrappers for the three conversation endpoints | | Support tools | `lib/ai/tools/support-tools.ts` — `escalate_to_human`, `resolve_support_conversation` | | Tool integration | `lib/ai/tool-filter.ts`, `tools/index.ts`, `generate-assistant-response.ts` | | Success card | `SupportAssistantSuccessCardContent.tsx` — tags chat on first engagement | | Assistant panel UI | `AIAssistant.tsx` — Escalate/Resolve buttons, disabled input on closed chats, support placeholders | <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit ## Summary by CodeRabbit - **New Features** - Support chats now include “Escalate to human” and “Resolve” actions. - Support submissions can be associated with a stable Front thread via a generated `threadRef`, preserving linkage across follow-ups. - AI assistant responses and input hints adapt when support mode is active. - **Bug Fixes** - Improved support chat state management and lifecycle handling to keep conversation metadata and message history synchronized more reliably with Front. - **Chores** - Added/updated coverage to reflect the new support-chat state and syncing behavior. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Co-authored-by: Alaister Young <10985857+alaister@users.noreply.github.com> |
||
|
|
18431efb25 |
fix(studio): TanStack post-merge fixes — Monaco loader, fonts, CSP (from #46424) (#47657)
Post-merge fixes for the TanStack Start migration (#46424) — things that broke on the TanStack build as master evolved under the migration branches. Kept on their own branch off master rather than piling onto the E2E-matrix PR (#47119); all land on master and cascade up to S6 + the big PR. Common theme: a master PR changed something the Next pipeline handles via `next/font` / `pages/_app.tsx` / `next.config.ts`, but the hand-rolled TanStack equivalent (`routes/__root.tsx`, `styles/fonts.css`, `vercel.ts`) wasn't updated to match — invisible on the Next deploy, broken only on TanStack. --- ## 1. Monaco loader path (#47182) #47182 re-nested the served Monaco assets from a flat `public/monaco-editor/` layout into `public/monaco-editor/vs/` and updated `pages/_app.tsx`, but `routes/__root.tsx` still pointed `loader.config` at the old path, so `loader.js` 404'd and **no Monaco editor mounted anywhere in the TanStack build**. Now mirrors the Next config (`${origin}${BASE_PATH}/monaco-editor/vs`, window-guarded for SSR). Was failing the whole `tanstack` E2E shard on #47119. ## 2. Inter + Manrope fonts (#47306) #47306 renamed Tailwind's sans var `--font-custom` → `--font-sans` and added `--font-heading` (Manrope), set via `next/font` on Next. `fonts.css` still only set the now-ignored `--font-custom`, so the body fell back to the theme's system chain (`Circular, custom-font, Helvetica…`) at weight 450 — that's the "Inter weights look wrong". Manrope was missing entirely. - Wire `--font-sans` (Inter) + `--font-heading` (Manrope) to match `next/font`. - **Vendor all three families** (Inter, Manrope, Source Code Pro) via `@font-face` so nothing depends on the Google Fonts CDN — matches `next/font` self-hosting, and (see below) `font-src` doesn't allow `fonts.gstatic.com` anyway. Verified in-browser: computed `body` → `Inter`, headings → `Manrope`, all loading from local `/assets/*.woff2`. ## 3. Security headers / CSP (next.config.ts `headers()`) The Next build sets X-Frame-Options / X-Content-Type-Options / HSTS / **Content-Security-Policy** / Referrer-Policy via `next.config.ts`. The TanStack build never carried these over — `vercel.ts` only set cache-control, so **the deployed TanStack dashboard shipped with no CSP at all**. The TanStack deploy serves a static shell (no server to attach headers), so they go in the Vercel config: - `security-headers.ts` — shared source of truth, reuses `getCSP()`, env-gated exactly like next.config. - `vercel.ts` — apply to every response (all base-path prefixes): full `getCSP()` + HSTS on platform. - `scripts/serve.js` — the non-platform set (`frame-ancestors 'none'`) for the self-hosted server. **Tested the policy in a real browser** (temporarily enforced it on the TanStack build via /test-supabase-local): everything passed except one real gap — `font-src` was missing `data:`, so GraphiQL's bundled Monaco codicon font and Stripe's payment-element fonts (both data: URIs) were blocked (37 violations on a cold load). Added `data:` to `font-src` in `csp.ts` → violations drop to zero, SQL editor Monaco renders clean. That gap affects the Next build too. --- ## 4. `node:path` import crashing `/project/[ref]/merge` Found by a full-site click-through of the TanStack build (all product areas, ongoing — see below). `useEdgeFunctionsDiff.ts` + `EdgeFunctionsDiffPanel.tsx` did `import { basename } from 'path'` in client code. Webpack (Next) polyfills `path` in the browser; Vite externalizes it, so the whole `/merge` route crashed with "Module \"path\" has been externalized for browser compatibility". Replaced the two `basename` call sites with a string helper. Verified in-browser: `/merge` renders. ## 5. URL shape — Next-style search-param semantics + shim fixes The dashboard produced malformed URLs vs the Next build (strange query params, trailing slashes, `##` hashes). Root cause + audit verified empirically against `@tanstack/react-router@1.170.10`; all fixed with unit tests and browser-verified: - **`createRouter` used TanStack's default JSON search codec** — `?flag=true` became `?flag=%22true%22` via links, repeated `?filter=…&filter=…` collapsed into a JSON array (breaking multi-filter/sort table-editor URLs and the account-page round-trip, which double-encoded), and search values arrived as numbers/booleans where the app expects strings. New `lib/router-search-params.ts` (Next-style: strings in, strings out, repeated keys → string[]) wired into the router. - **Link shim** (`compat/next/link.tsx`): `URL.hash` includes the leading `#` while TanStack's `hash` prop adds its own → every `href="…#section"` navigated to `##section` (hash-scroll broke); `Object.fromEntries(searchParams)` dropped repeated query params. Both fixed. - **Trailing slash injected before the query** on every `?`-only relative navigation (`/auth/providers/?provider=…`): fixed in the compat router (prefix current pathname) and via a custom nuqs adapter (`lib/nuqs-tanstack-adapter.tsx`) replacing the stock tanstack-router adapter, whose `navigate({ to: '?…' })` writes hit the same TanStack behavior (123 files use nuqs). - **Pathname-less `router.push({ query })` leaked path params** — Next re-consumes `ref`/`id` from `query` into the path pattern; the shim didn't, yielding `/editor/17597?schema=public&ref=<ref>&id=17597&filter=…` from table-editor filter/sort, linter panels, and advisor shortcuts. The shim now defaults the pathname to the current route pattern and backfills omitted params. - **Redirects dropped query + hash** (Next's `redirects()` preserves them): `__root.tsx` `matchRedirect` and `routes/index.tsx` now carry incoming params/hash through (consumed rule params excluded, destination's own params win). `/?next=new-project&projectName=zzz` → `/new/new-project?projectName=zzz`; `/sql/quickstarts?template=x#frag` → `/sql/examples?template=x#frag`. Browser-verified post-fix: advisors `?preset=WARN`, providers `?provider=Google`, `?schema=auth` — all clean (no `/?`, no leaks); repeated `filter` params survive hydration; `=true` unquoted; single `#`. ## 6. TanStack `navigate` corrupting query values (Logs Explorer SQL newline loss) TanStack router-core treats a query string embedded in `navigate({ to })` as part of the *path*: `decodePath` percent-decodes it and `sanitizePathSegment` strips control characters, silently deleting every `%0A`. Logs Explorer's SQL (`s` param) lost its newlines on Run/reload — `order by timestamp desc` / `limit 5` glued into `desclimit 5`, which then failed the LIMIT lint. Pre-existing on the TanStack build (the stock nuqs adapter had the same shape); Next unaffected. Fixed by never embedding query strings in `to`: the nuqs adapter and the compat `router.push`/`replace`/`prefetch` (plus the `next/navigation` shim) now pass search as an object through the app codec (`splitInternalUrl` hoisted to `lib/internal-url.ts`). Guard test drives a real `createRouter` with multi-line SQL through both producers. Browser-verified: newlines survive the full Run → reload → re-Run cycle. ## 7. Integration overview markdown never loaded (all integrations) `MarkdownContent` used a template-literal dynamic import (``import(`@/static-data/integrations/${id}/overview.md`)``) — webpack builds a context module for that, Vite can't analyze it, so every integration detail page threw `Failed to resolve module specifier` and rendered no overview text. Fixed with an explicit lazy registry of literal imports (`static-data/integrations/overviews.ts`, drift-guarded by a test) plus an `mdRawLoader()` Vite plugin mirroring next.config's turbopack raw-loader rule. Both runtimes keep working; md stays out of the main bundle. ## 8. GraphiQL editor never mounted (`exports is not defined`) Our `umdAmdShortCircuit()` Vite plugin (which disarms Monaco's global AMD loader for deps like papaparse) rewrote `typeof define === 'function' && define.amd` to `false` inside `monaco-editor`'s bundled copy of marked — whose UMD relies on its own *local* `define` shim — so the whole optimized monaco chunk failed to evaluate and GraphiQL's editor pane stayed blank. The check now only short-circuits when `define` is the global AMD loader. Browser-verified: all four GraphiQL Monaco panes mount, queries execute. (Known follow-up: GraphiQL's Monaco workers fall back to the main thread under Vite — functional, worker wiring is Next-specific `setup-workers/webpack`.) ## 9. `@sentry/nextjs` bundling Next internals — built TanStack bundle crashed (caught by E2E) The E2E suite against the **built** TanStack bundle (not the dev server) found lazy chunks like `table-editor-*.js` dead on arrival: `@sentry/nextjs` (imported by ~25 client files) drags in `next/dist/shared/lib/constants`, whose module scope evaluates `process?.features?.typescript` — optional chaining doesn't guard an undeclared `process` in the browser, so the whole chunk failed at load with `ReferenceError: process is not defined`. Dev shims `process`, which is why weeks of dev-server testing never saw it. Fixed by aliasing `@sentry/nextjs` → `compat/sentry-nextjs.ts` (re-exports `@sentry/react`, same deduped 10.59.0, plus explicit stand-ins for the three Next-only APIs) in the Vite build only. Verified: fresh build has zero Next-internals markers in any chunk; table editor loads clean; full E2E suite run against the built bundle. Note for the stack: `alaister/tanstack-start` / the E2E-matrix branch already carried a different fix for the same crash (a `next/constants` shim) that never made it to master — the cherry-pick onto those branches keeps **both** (the shim covers any other transitive importer; the alias keeps Next internals out of the client bundle entirely). **Follow-up found while fixing:** Sentry is never *initialized* in the TanStack runtime — `instrumentation-client.ts` / `sentry.server.config.ts` are Next-convention files nothing imports under TanStack, so `captureException` calls are silent no-ops. Needs an `@sentry/react` init (+ `tanstackRouterBrowserTracingIntegration`) wired into the TanStack client entry as its own PR. ## 10. GraphiQL Monaco workers + edge-function Deno typings (Vite-only gaps) - **GraphiQL's Monaco workers ran on the main thread** under Vite ("Could not create web worker(s)…" — `setup-workers/webpack`'s `new URL(...)` form isn't rewritten by Vite). A `graphiqlViteWorkers()` plugin resolves the import to graphiql's own `setup-workers/vite` variant for client builds (SSR untouched, Next untouched); the setup-workers chain is `optimizeDeps.exclude`d because the Rolldown optimizer can't load `?worker` ids. - **Edge-function editors silently lost their Deno typings** — `AIEditor` loaded `public/deno/*.d.ts` via `/* @vite-ignore */` imports that always failed at runtime under Vite. The `.md` raw loader is generalized into `rawTextLoader` (exact-path allowlist for the two typings files, served as virtual string modules so the dep scanner never parses `.d.ts` syntax), and the imports are now static-analyzable literals that both bundlers handle (turbopack's raw-loader rules match them on the Next side). ## Split out for reviewability App-level fixes that reproduce on the Next build too (DOM-nesting hydration errors, the ghost deleted-snippet nav, the recurring pg-meta `migrations` 400) moved to their own PR: #47667. Sentry initialization for the TanStack runtime (captures were silent no-ops) is #47666, stacked on this PR. ## Full-site test campaign Drove every dashboard product area on the local TanStack build (Playwright, human-style) hunting migration regressions: redirects/404/catch-alls, org, account, project home/branches/merge, table editor CRUD, SQL editor (Monaco/run/save/templates/AI), all database pages, all auth pages, storage CRUD, edge functions + realtime, logs/observability, advisors, settings, integrations hub incl. nested routes, global UI (palette/connect/switchers/theme/fonts), and a cross-cutting sweep (document titles, back/forward chain, hard-refresh hydration on deep URLs, trailing-slash active state). Every failure found is fixed above and re-verified in-browser; remaining console quirks were cross-checked against the deployed Next build and are pre-existing (tracked separately). ## To test Most fixes are already browser-verified + covered by unit tests and the self-hosted E2E suite; the last two landed after the final browser pass and still need an in-browser check: 1. **GraphiQL Monaco workers** — restart the dev server (clear `apps/studio/node_modules/.vite` once first — the optimizer cache may hold a stale prebundle of the worker chain). Open `/project/<ref>/integrations/graphiql/graphiql` with the console open: the `Could not create web worker(s). Falling back to loading web worker code in main thread` warning must be gone, and DevTools → Sources → Threads shows the three workers (json, editor, graphql). Autocomplete in the query editor stays responsive. 2. **Edge-function Deno typings** — `/project/<ref>/functions/new`: no "Failed to load … typings" console error, and typing `Deno.` in the editor offers typed completions (e.g. `Deno.env`). Spot-checks for the rest (all previously verified): - `/project/<ref>/merge` renders (no "Module path" crash). - Multi-line SQL in Logs Explorer survives Run → reload (no `desclimit` gluing, no LIMIT-lint false failure); `s` param keeps `%0A`. - `/auth/providers` → open a provider → `?provider=…` with no trailing slash before `?`; table-editor filter/sort URLs carry no leaked `ref`/`id` params; `/?next=new-project&projectName=x` lands on `/new/new-project?projectName=x`. - Integration detail pages (cron/queues/vault/data_api) show their overview prose; GraphiQL query editor mounts. - Built bundle (`MODE=test vite build` + `start:tanstack`): table editor loads with no `process is not defined`. - `curl -sI` any page on a platform deploy: `X-Content-Type-Options: nosniff` (was the invalid `no-sniff`). <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Centralized integration overview markdown loading with registry-based lookup. * Improved Monaco loading/asset path handling for smoother editor startup. * **Bug Fixes** * Next-style navigation/search handling now preserves pathname, hash, repeated query keys, and special characters (including newlines). * Redirects now reliably carry over query and hash with correct precedence. * **Security/Configuration** * Updated CSP font sourcing and unified security headers delivery across environments; conditional HSTS behavior. * Refreshed font CSS variables and font-face definitions to match the theme. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --- ### Review feedback: non-prod favicon (Joshen) The TanStack `__root.tsx` hardcoded the prod favicon; local + hosted staging now use the white staging favicon (`/favicon/staging`), matching what `pages/_app.tsx` passes to `MetaFaviconsPagesRouter` for non-prod. Rather than pull the pages-router component into the TanStack head, it reuses the same synchronous `NEXT_PUBLIC_ENVIRONMENT` signal the file already uses for `IS_DEV_TOOLBAR_ENABLED` (the `head()` route option isn't a React component, so it can't run `_app`'s async CLI check — but the env signal covers the reported local/staging case). --------- Co-authored-by: Alaister Young <10985857+alaister@users.noreply.github.com> Co-authored-by: Joshen Lim <joshenlimek@gmail.com> |
||
|
Pedro RodriguesandClaude Opus 4.8
|
c4c213ce3d |
feat(studio): switch dashboard assistant to remote MCP server (#47479)
## 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? Feature / refactor. ## What is the current behavior? The dashboard assistant runs `@supabase/mcp-server-supabase` in-process over an in-memory transport (`lib/ai/supabase-mcp.ts`). ## What is the new behavior? The assistant connects to the **remote MCP server** over HTTP (`@ai-sdk/mcp`), forwarding the dashboard session token as a bearer. URL comes from `NEXT_PUBLIC_MCP_URL` with a local-dev fallback; platform-only, and Nimbus works via the same env var. * **Tool model unchanged:** UI-controlled `execute_sql` (with `needsApproval`) and `deploy_edge_function` still come from Studio; the allowlist (`TOOL_CATEGORY_MAP`) remains the gate keeping the remote's write tools away from the assistant (`read_only` is defense-in-depth). * **Attribution:** sends `x-source-name: supabase-studio` (+ `x-source-version`) → logged as `source_name`/`client_name`. * **Connection lifecycle:** the HTTP client is closed via the request's `AbortSignal` (tools execute later during streaming); `signal` is required on `getTools`/`getMcpTools`. * **Resilience:** a remote-MCP failure degrades to the remaining tools instead of failing the assistant. * **Drift protection:** relied-upon tools are typed against `keyof typeof supabaseMcpToolSchemas`, so a package bump that renames/removes one fails `pnpm typecheck`; a runtime check also warns if the deployed server returns fewer tools. * Adds unit tests for the above. ## Additional context * Verified end-to-end against a local remote MCP server with a dashboard token: `initialize` 200, tools listed, a tool executed, client closed cleanly. * The remote MCP (mgmt-api) already accepts dashboard session tokens (GoTrue-JWT auth path) — no backend change needed. `NEXT_PUBLIC_MCP_URL` must point at each env's `/mcp`. * `@supabase/mcp-server-supabase` is kept — still used by the self-hosted `/api/mcp` routes. Closes [AI-137](https://linear.app/supabase/issue/AI-137/switch-dashboard-assistant-to-remote-mcp) ## Rollout * **Rollout:** merges with `USE_REMOTE_MCP` off (in-process); flip it to `true` per environment (staging → prod → Nimbus) once each one's prerequisites land. * **Rollback:** unset `USE_REMOTE_MCP` and redeploy to fall back to the in-process client — no revert needed. ## Summary by CodeRabbit * **Bug Fixes** * Improved AI request handling so tool loading and generation clean up properly when a request is cancelled or the browser connection closes. * Added safer fallback behavior when remote tool loading fails, so AI features can continue with available tools instead of stopping entirely. * Updated remote tool access to use the current project reference and preserve the correct access headers. <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * AI tools now connect more reliably to remote services and stop cleanly when requests end or are canceled. * Tool loading is more resilient, continuing with available tools if remote access is unavailable. * **Bug Fixes** * Improved cleanup to prevent lingering connections during SQL generation and policy workflows. * Added safer handling for remote tool changes and invalid responses. * **Tests** * Expanded automated coverage for remote tool setup, cancellation, and fallback behavior. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Alaister YoungandAlaister Young
|
46b31eb53a |
[FE-3379] feat(studio): warn when db passwords need percent-encoding (#47564)
Users who set a database password with special characters (\`@\`, \`#\`, \`%\`, \`+\`, etc.) get no warning that it must be percent-encoded when used in a connection URL, which leads to confusing connection failures ([FE-3379](https://linear.app/supabase/issue/FE-3379)). <img width="700" height="200" alt="Screenshot 2026-07-03 at 6 26 43 PM" src="https://github.com/user-attachments/assets/48608d65-8057-4abe-96fc-c0ede3550951" /> <img width="1002" height="395" alt="Screenshot 2026-07-03 at 6 27 14 PM" src="https://github.com/user-attachments/assets/1366b985-7d80-4e7d-97f0-c79d5c84cefd" /> <img width="548" height="303" alt="Screenshot 2026-07-03 at 6 27 26 PM" src="https://github.com/user-attachments/assets/b042101a-0e88-4730-adb8-1b490018f208" /> **Changed:** - `PasswordStrengthBar` now shows a warning-colored callout (with a docs link) whenever the entered password contains characters that need percent-encoding — this covers project creation, reset database password, restore-to-new-project, and the Vercel deploy-button flow - Replaced `DATABASE_PASSWORD_REGEX` (only caught `@`, `:`, `/`) with a `passwordNeedsPercentEncoding()` helper based on `encodeURIComponent`, so `#`, `%`, `+`, `?`, `&`, spaces etc. are caught too - Moved `SpecialSymbolsCallout` from `ProjectCreation/` to `components/ui/` since it's now shared **Added:** - Info admonition in the Connect sheet next to connection strings that still contain `[YOUR-PASSWORD]` (direct connection + `.env`-based file setups; hidden for psql and .NET where percent-encoding doesn't apply, and after a password reset since the substituted password is already encoded) ## To test - Project creation → type a password containing \`#\` or \`@\` → warning callout appears above the strength bar; disappears for alphanumeric passwords - Database Settings → Reset database password → same behaviour - Connect sheet → Direct connection → note shows under the connection string for URI/JDBC types, not for psql; after resetting the password from the sheet, the note disappears (password is substituted already encoded) - Connect sheet → Node.js/Python/Go/SQLAlchemy file setups show the note; .NET does not - \`pnpm vitest run lib/password-strength.test.ts\` passes <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit ## Summary by CodeRabbit * **New Features** * Added a dedicated password encoding note (with documentation link) on direct connection screens when the password is embedded in a URL. * Added an encoding hint to the password strength area when percent-encoding is required. * **Bug Fixes** * Removed regex-based “invalid password” callout and replaced it with safer percent-encoding detection logic. * **Tests** * Added test coverage for `passwordNeedsPercentEncoding`. * Removed obsolete Project Creation password regex tests. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Co-authored-by: Alaister Young <10985857+alaister@users.noreply.github.com> |
||
|
Saxon FletcherandDanny White
|
09ea558d54 |
Vercel install layout (#47550)
Bring the Vercel install layout up to date with other connect screens. Resolves DEPR-615. | Before | After | | --- | --- | | <img width="1600" height="1200" alt="CleanShot 2026-07-03 at 10 51 59@2x" src="https://github.com/user-attachments/assets/6a8c2910-8c45-4c7d-8d38-8b80c5cf4c83" /> | <img width="1150" height="1318" alt="CleanShot 2026-07-03 at 11 56 17@2x" src="https://github.com/user-attachments/assets/3a5add5c-fea4-44f3-a368-5732257b27d9" /> | ## Testing - Open the deploy preview or staging URL for `/dashboard/integrations/vercel/install` with callback params from a real Vercel Marketplace install redirect (see _Vercel_ subheading below). - Confirm the install screen renders with the selected Supabase account, organization picker, and primary install CTA. - Remove required callback params such as `code`, `configurationId`, or `source` to verify the "Missing Vercel installation details" warning state. - Clicking "Install integration" with an expired or reused `code` can show `Creating Vercel integration failed: Failed to get Vercel access token`; that is expected for preview UI validation. A full successful install requires a fresh Vercel-generated code from the install flow. To reiterate; this won’t work because the code will be invalid. But it should show that everything is hooked up right for prod. ### Vercel 1. Go to Supabase. Open either org-level or project (settings) level integrations. 2. Tap "[Install Vercel Integration](https://vercel.com/integrations/supabase-local)". 3. Follow the install instructions from that Vercel page. This should open a browser window. 4. Copy the URL of that browser window. Take the params and paste them instead at the end of the deploy preview URL. Example of #4: ```txt Before: https://supabase.com/dashboard/integrations/vercel/install?code=jDhIBDlD58zzLVtuSNjJpUSu&configurationId=icfg_v3dKllQIniSOwdVI3gypnZh3&next=https%3A%2F%2Fvercel.com%2Ftest-5706s-projects%2F~%2Fintegrations%2Ficfg_v3dKllQIniSOwdVI3gypnZh3%2Finstalled&source=marketplace&teamId=team_mPkGQZjTLBEUXh15b03iVsTg After https://studio-staging-git-chore-install-layout-supabase.vercel.app/dashboard/integrations/vercel/install?code=jDhIBDlD58zzLVtuSNjJpUSu&configurationId=icfg_v3dKllQIniSOwdVI3gypnZh3&next=https%3A%2F%2Fvercel.com%2Ftest-5706s-projects%2F~%2Fintegrations%2Ficfg_v3dKllQIniSOwdVI3gypnZh3%2Finstalled&source=marketplace&teamId=team_mPkGQZjTLBEUXh15b03iVsTg ``` <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Updated the installation experience with a cleaner, more guided layout. * Added clearer organization selection during setup, including visibility into which organizations are already installed. * **Bug Fixes** * Improved loading and error handling during installation. * Added clearer warnings for missing setup details, already-installed integrations, and cases with no available organizations. * Fixed routing behavior so organization selection is preserved more reliably. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Co-authored-by: Danny White <3104761+dnywh@users.noreply.github.com> |
||
|
Andrey A.
|
514c3aa0d0 | fix(self-hosted): type generation should respect exposed schemas (#47577) | ||
|
Ivan Vasilov
|
fae2c5b96e |
chore: Add org slug and project ref as tags to the Sentry data (#47442)
<!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Bug Fixes** * Enhanced telemetry to include the active project reference and selected organization on events. * Cleared and re-applied project and organization tags when navigating to prevent stale tagging. * Only applies user identification when user account details are available. * **Tests** * Updated telemetry tests to cover the additional tagging behavior by extending Sentry mocks accordingly. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Pamela Chia
|
98cfe3307e |
feat(telemetry): fix creation-funnel tracking gaps (#47386)
## Summary The creation-funnel instrumentation that shipped Jun 25 (#47291, #47293) had real gaps, surfaced by the weekly telemetry audit and confirmed against production PostHog data before I touched code. The two automated reports also contradicted each other on `errorReason`; I checked production (every value is a controlled slug) and the emit path (only `useTrackFunnelError` sets it, and it only accepts classified slugs), so I left the type as-is rather than add a cross-package abstraction for a risk that cannot occur today. ## Changes - Classify HTTP 401/403/404 API errors as `unauthorized` / `forbidden` / `not_found` instead of the catch-all `other`. In production the `org_creation` `other` bucket was ~96% 401s (~1,300 real over 4 days), invisible in reason breakdowns. The status-code fallback runs after the message-pattern match, so specific reasons still win and it only rescues errors that would otherwise be `other`. - Add a single `tier` property (`tier_free` / `tier_pro` / `tier_payg` / `tier_team`) to `organization_creation_completed`, which previously carried no properties. One canonical billing slug (matching `SubscriptionTier`) instead of two overlapping plan/tier fields, so the org-creation funnel segments cleanly by tier and joins against subscription data. `tier_payg` is uncapped PRO. - Freeze the submitted tier at submit time (snapshot in `createOrg`) rather than reading live form state in the success callback, so the event records the tier that was actually created even if the user edits the form during the async payment flow. - Emit `project_creation_form_exposed` with `surface: 'vercel'` on the integration deploy-button project-creation page (the enum value existed but was never fired). Gated on the URL `slug` so the impression is captured as soon as the form renders, matching the sibling exposure hook on that page. I also checked the confirm-modal error path flagged in the insights post: it already classifies via the shared `useProjectCreateMutation.onError`, so adding instrumentation there would double-count. No change made. ## Testing These are analytics events with no UI change, so correctness is in what lands in PostHog. Post-deploy validation I will run against production (project 34344): - `dashboard_error_created` where `origin='org_creation'` and `errorReason='other'` drops ~96%, with `unauthorized` / `not_found` appearing. - `organization_creation_completed.tier` populated on 100% of new events with one of the four tier slugs. - `project_creation_form_exposed` with `surface='vercel'` goes from 0 to greater than 0. ## Linear - fixes GROWTH-948 <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Added telemetry for organization creation completion that includes the selected billing tier. * Added one-time telemetry when the Vercel project creation form is exposed. * **Bug Fixes** * Improved API error classification to more accurately distinguish unauthorized, forbidden, and not found responses. * **Documentation** * Updated telemetry event definitions to require tier metadata for organization creation events. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Gildas Garcia
|
0e3364bbad |
Chore/cleanup studio deps (#47399)
## Problem Knip reported some unused dependencies. Some are actually used in builds, etc but others are not. ## Solution Remove the really unused dependencies <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Chores** * Cleaned up unused dependencies and removed some obsolete test/support files. * Updated project ignore rules to better match current app structure and generated files. * **Bug Fixes** * No user-facing behavior changed; this release is focused on maintenance and cleanup. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Andrey A.
|
8a9a9948a8 | fix(studio): self-hosted folder listings return metadata only (#47403) | ||
|
Gildas Garcia
|
c6fc456910 |
chore: cleanup duplicate exports studio (#47387)
## Problem Knip reports many duplicate exports (both named and default). Besides, we're moving away from default exports and even have an eslint rule to enforce it on new code. ## Solution - Cleanup those exports - Update imports when necessary No functional changes. If it builds, it's fine |
||
|
|
2aa1b52234 |
feat(studio): add feature to rewrite queries DEBUG-145 (#47266)
## Problem Moving the Logs Explorer to ClickHouse means users' saved BigQuery queries no longer run. <img width="2430" height="1010" alt="CleanShot 2026-06-29 at 11 36 04@2x" src="https://github.com/user-attachments/assets/ae0ab155-7d3d-4ae9-81c3-22bf3a88cf8c" /> ## Fix Rewrite the query with AI instead of a SQL transpiler. AI handles the long tail of nested fields and dialect differences far better than a rule-based rewriter, and it needs no extra runtime dependency. - `rewriteLogsSqlWithAI` posts the current query to `/api/ai/code/complete` with `dialect: 'clickhouse'`. The endpoint skips the Postgres schema and best-practices for that dialect and uses logs-specific instructions and model so the output is ClickHouse logs SQL (FROM `logs` + `source` filter, no `unnest` joins, nested fields read from `log_attributes['...']`). - The query's `source` is detected and its real `log_attributes` keys are fetched and passed to the model, so it maps to exact paths instead of guessing. - The rewrite runs in the background and is proposed as a side-by-side accept/discard diff in the editor. The AI Assistant panel is not opened. - Entry points: a banner shown only for legacy-looking queries (dismissal persisted), and a "Fix Query" button next to Field Reference. - The Field Reference drawers discover `log_attributes` keys from real data so the listed fields match what the source actually emits. ## Dependencies Built on top of #47265 (Logs Explorer -> OTEL endpoint) — that is the base branch of this PR. Merge #47265 first. Behind `otelLegacyLogs` (off by default). Part of DEBUG-145 (split from #47087). ## How to test - Open the Logs Explorer with a BigQuery logs query (the templates have some), click "Fix Query", and confirm the diff shows valid ClickHouse SQL. Accept it and confirm the applied query runs. <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Added an OTEL legacy logs workflow (behind a feature flag) with an interactive banner and a “Fix Query” ClickHouse rewrite action, including an accept/discard diff review overlay. * Introduced OTEL-aware field reference rendering with dynamic discovery of `log_attributes` keys and updated OTEL source insertion behavior. * Enabled dialect-aware SQL completion for ClickHouse logs, using logs-specific instructions and output constraints. * **Bug Fixes** * Improved rewrite flow validation and handling, including log source detection and cleanup of AI-generated SQL formatting. * **Tests** * Added Vitest coverage for rewrite prompt generation, detection/classification utilities, SQL fence stripping, OTEL field mapping, and OTEL log attribute key discovery. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Co-authored-by: Joshen Lim <joshenlimek@gmail.com> |
||
|
Ivan Vasilov
|
a9565c1645 |
refactor: Optimize Sentry user ids (#47342)
<!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Bug Fixes** * Improved signed-in user identification in error tracking to use the current authenticated user ID directly. * Removed reliance on previously stored anonymous identifiers, reducing inconsistent user matching across sessions. * **Tests** * Added coverage to confirm error tracking user identity is set when a user is signed in and not set when signed out. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Charis
|
16526bd6bf |
refactor(studio): extract SQL editor save mechanism + model folder lifecycle (4/9) (#47276)
## What
PR 4 of a stacked refactor of the SQL editor snippet/folder state. It
pulls the persistence logic out of the store into an injectable
mechanism, and replaces the folder `'new-folder'` id sentinel with an
explicit lifecycle — plus a concurrency bug fix that surfaced along the
way.
### Save mechanism (`sql-editor-save.ts`)
`createSaveMechanism({ state, upsertContent, createSQLSnippetFolder,
updateSQLSnippetFolder, invalidate, notify, debounceMs })` → `{
saveSnippet, createFolder, updateFolder }`. The store's subscribe now
dispatches to it; *when* to save still lives in the subscribe (the
scheduler/provider move is PR 5). Per-id debounce cache lives in the
factory closure (no module-global leak).
- **`saveSnippet`** reads the live store snippet, guards
`isLoadedSnippet` so a content-less snippet can **never PUT an empty
body** (directly unit-tested), then builds the payload + drives status
transitions + gated invalidation.
- **`toast` is injected** as a `Notifier` (new generic DI contract in
`lib/notifier.ts`) — the mechanism no longer imports sonner.
- **create vs rename are two named-arg functions**, not an `isNew`
branch; rollback is deterministic per operation instead of matching on
`error.message` text.
- **caught errors are `unknown`**, narrowed via the existing
`getErrorMessage` util with a generic fallback — no `any`.
### Folder lifecycle (replaces the `NEW_FOLDER_ID` sentinel)
- **`FolderStatus`** enum (`new_editing | new_saving | editing | saving
| idle`) collapses the persistence and progress axes into one enum —
same pattern as `SnippetStatus` — with `isNewFolder` / `isFolderEditing`
/ `isFolderSaving` predicates. Tagging a folder as new/persisted is now
an explicit field, not an id convention.
- New placeholders get a **unique local id** (`crypto.randomUUID`);
`NEW_FOLDER_ID` is deleted, which also lifts the accidental
one-unsaved-folder-at-a-time limit.
### Bug fix: folder-rename rollback race
The shared `lastUpdatedFolderName` field let two in-flight renames
clobber each other's rollback target (and a shared `finally` could wipe
it). Replaced by a **per-folder `previousName`** on
`StateSnippetFolder`, so concurrent renames of different folders are
isolated. A new test runs two failing renames concurrently and asserts
each restores its own previous name.
## Tests
`sql-editor-save.test.ts` (mechanism — fakes + fake timers, incl.
content-less no-PUT and concurrent-rename isolation) and
folder-lifecycle predicate tests. `pnpm --filter studio typecheck`
clean; 82 state/sql-editor unit tests pass.
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Improved SQL editor folder handling with clearer create, rename, and
save states.
* Added a more consistent notification flow for successful and failed
save actions.
* **Bug Fixes**
* Improved rollback handling when folder renames fail, helping restore
the previous name reliably.
* Updated save behavior to better protect against duplicate or
out-of-order updates.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
|
||
|
Pamela Chia
|
2d0bcd4714 |
feat(telemetry): classify funnel creation errors (#47293)
## Summary The KPI-3 friction dashboard needs to know *why* users hit errors on the signup, project-creation, and org-creation funnels, not just that they did. The existing `dashboard_error_created` event already fires for these paths (10% sampled, with `$pathname`), but carries no reason: ~98.5% of events have no `errorType` and no property carries an error message. This adds PII-safe classification computed client-side from a controlled vocabulary, so raw error text never leaves the browser. Validation errors (previously invisible, since they are inline form errors that never raise a toast) are now captured on invalid submit. ## Changes - Extend `dashboard_error_created` with `origin`, `errorCategory`, `errorReason`, `errorCode`, and a `form` source value - Add a pure, unit-tested classifier (`funnel-errors.ts`) and a 10%-sampled tracking hook (`use-track-funnel-error.ts`); the classifier maps errors to stable slugs and emits only slugs + HTTP status, never raw message text - Classify signup errors (API failures + validation) in `SignUpForm` - Classify project-creation errors (API failures, OrioleDB guard, validation) in the new-project wizard - Classify org-creation errors (API failures, payment/card declines, confirm-subscription, validation) in `NewOrgForm` ## Testing 13 unit tests cover every classifier branch (validation / api / network / payment, status-code handling, message-pattern matching, and fallbacks). To verify on the Vercel preview (events are 10% sampled; set the sample rate to 1 locally to observe each fire): - Signup with a weak but non-empty password: `origin=signup, source=form, errorCategory=validation, errorReason=password_invalid` - Signup with an already-registered email: `origin=signup, source=toast, errorCategory=api, errorReason=email_already_registered` - New project with an empty name: `origin=project_creation, source=form, errorReason=project_name_invalid` - New org with an empty name: `origin=org_creation, source=form, errorReason=org_name_missing` - New org with a declined test card: `origin=org_creation, errorCategory=payment` PII: raw `error.message` is never sent; only controlled slugs and HTTP status. Dashboard consumers must filter `origin IS NOT NULL` so these do not collide with the generic toast events the global tracker still emits. ## Linear - fixes FE-3691 <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Added improved, categorized telemetry for signup, project creation, and organization creation errors, including payment, subscription-change, and validation failures. * Extended dashboard error events with optional structured diagnostics (origin, category, reason, and optional error code) and support for form-origin reporting. * **Bug Fixes** * Improved project-creation handling to record a validation telemetry event when an Oriole image is unavailable. * Ensured payment-related and subscription-change failures are captured consistently alongside existing user toasts. * **Tests** * Added unit tests covering API/network/validation/Stripe error classification and reason mapping. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Joshen Lim
|
9f7d300354 |
Check flags loaded before fetching region data (#47289)
## Context
Realised that we were calling both `useDefaultRegionQuery` and
`useOrganizationAvailableRegionsQuery` in the new project page.
`smartRegionEnabled` defaults to `false` at the beginning while the
flags are still being loaded, to this calls `useDefaultRegionQuery`. But
once the flags are loaded and `smartRegionEnabled` becomes `true`, then
the other hook is called
## Changes involved
- Checks that the flags are loaded first before calling either hooks to
prevent unnecessarily triggering both
- Adjust `defaultRegion` to remove hardcode
- Check smart region first, then default back to default specific region
- Renamed variables to be clearer:
- `autoDefaultRegion` (check based on location)
- `fixedDefaultRegion` (hardcoded in repo)
- Update `useEffect` on `regionError` to only reset the region value if
the default value is not undefined
- Also just a tiny nit to re-arrange to group the `useEffects` together
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **Bug Fixes**
* Region selection now waits for feature flags to load before fetching
default and “smart” region options, preventing premature or incorrect
region choices.
* New project setup now updates the database region and smart
recommendations more consistently, with improved fallback behavior when
region lookup fails.
* AWS Nimbus default region is now environment-aware: non-prod uses
Southeast Asia, while prod uses East US.
* Default privilege settings now stay in sync more reliably during
setup, updating only when appropriate.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
|
||
|
|
9eab4f8fbf |
build(studio): Vite/TanStack-Start build pipeline behind flag (stack 1/6, from #46424) (#47107)
**Stack 1/6** of the TanStack Start migration (#46424), split into reviewable, independently-mergeable PRs. > [!IMPORTANT] > **Next stays the default and only active framework after this PR.** This wires up the Vite/TanStack-Start build pipeline behind the `STUDIO_FRAMEWORK` flag, but there are no TanStack routes yet — so the TanStack build isn't functional or tested until later PRs in the stack. Nothing about the Next build, dev, or deploy changes behaviourally here. ## What's in this PR - **Dispatch:** `dev`/`build`/`start` now go through `scripts/dispatch.js`, which runs the Next variant unless `STUDIO_FRAMEWORK=tanstack`. The original commands are preserved as `dev:next`/`build:next`/`start:next`. - **Build pipeline:** `vite.config.ts`, `serve.js`, `smoke-server.mjs`, vite/tanstack deps, `turbo.jsonc`. - **`tsconfig.json`:** `jsx: react-jsx`, `moduleResolution: Bundler`, `target: ES2022`. Because `include` is `**/*.ts(x)`, this re-typechecks the whole app, so the companion adaptations below land with it. - **Shared adaptations (companions to the tsconfig change):** `BufferSource` casts, `packages/ui` unused-`React` import removals, etc. - **Routing/middleware plumbing:** `next.config.ts` + `redirects.shared.ts` (redirect rules now shared with `vercel.ts`), `proxy.ts`/`start.ts` middleware + `hosted-api-allowlist.ts`. ## Verification Run locally off `master`: frozen install ✓, `studio` typecheck ✓, **Next build ✓** (compiles + generates all routes), lint ratchet ✓ ("some rules improved"), prettier ✓. <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Added a hosted API endpoint allowlist to return 404 for non-supported `/api/*` routes. * Introduced a TanStack route-migration checklist and expanded TanStack Start routing support. * **Improvements** * Enhanced deployment refresh/detection by tightening cookie handling for “latest deployment” updates. * Centralized redirect/maintenance-mode rules for consistent platform vs self-hosted behavior. * Improved production serving with a dedicated static + proxy server and a post-build smoke test. * **Dependencies** * Updated TanStack-related packages and React Table/query tooling versions. * **Documentation / Chores** * Updated formatting and tooling config; added shared build environment parsing utilities. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Co-authored-by: Alaister Young <10985857+alaister@users.noreply.github.com> Co-authored-by: Ivan Vasilov <vasilov.ivan@gmail.com> |
||
|
|
e81c714aae |
refactor(studio): lazy self-hosted admin client + enforce in API routes (from #46424) (#47104)
Extracted from the TanStack Start migration (#46424) to shrink that PR. The self-hosted storage/auth API routes each constructed a module-scope admin client (`createClient(process.env.SUPABASE_URL!, process.env.SUPABASE_SERVICE_KEY!)`). Those env vars only exist on self-hosted, so eager module-scope construction is wasteful on platform and fragile on any runtime that evaluates an API module before its route is hit (constructing with `undefined` credentials throws on import). **Changed:** - Add `lib/api/self-hosted-admin.ts` — `selfHostedSupabaseAdmin`, a `Proxy` that defers `createClient(...)` until first property access (inside a handler, i.e. on self-hosted where the vars are set). - Swap **all 17** storage/auth/vector-bucket handlers from module-scope `createClient(...)` to `import { selfHostedSupabaseAdmin as supabase }`. - **Enforce it:** add an eslint `no-restricted-syntax` rule banning module-scope `createClient` in `pages/api/**` + `routes/**` (now that every flagged handler is lazy). The same eslint config block also carries an analytics-SQL boundary rule — 0 violations on master. Behaviour is unchanged (the client is still built lazily inside the handler). This is also the change that makes those routes safe under TanStack's single-handler module evaluation. ## To test - Self-hosted Studio: storage buckets/objects, vector buckets, and auth users operations work as before. ## Verification studio lint (0 errors, both rules active) ✓ · studio typecheck ✓. <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Refactor** * Standardized self-hosted Supabase admin client usage across platform authentication and storage endpoints, removing per-route client setup. * Improved reliability by lazily creating the admin client only when first used. * **Chores / Tooling** * Updated ESLint rules to prevent module-scope Supabase client creation in API routes and to enforce safe analytics SQL access patterns. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Co-authored-by: Alaister Young <10985857+alaister@users.noreply.github.com> Co-authored-by: Ali Waseem <waseema393@gmail.com> |
||
|
|
e491182054 |
Auth flow improvements (#46967)
## I have read the [CONTRIBUTING.md](https://github.com/supabase/supabase/blob/master/CONTRIBUTING.md) file. YES/NO ## What kind of change does this PR introduce? Bug fix, feature, docs update, ... ## What is the current behavior? Please link any relevant issues here. ## What is the new behavior? Feel free to include screenshots if it includes visual changes. ## Additional context Add any other context or screenshots. <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit ## Release Notes * **New Features** * Added “Continue with {provider}” sign-in and sign-up flows using enabled external identity providers. * Enabled inbound branding to focus a specific provider for customized sign-in/sign-up experiences. * **Improvements** * Refined the sign-in options layout and “last used” tracking for clearer authentication choices. * Updated account identity/provider connection experiences (link/unlink and management UI). * **Bug Fixes** * Fixed hydration mismatches in sign-in and password-related layouts. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com> Co-authored-by: Joshen Lim <joshenlimek@gmail.com> |
||
|
Matt Rossman
|
c54168842b |
fix(assistant): prevent parallel approval-required tool calls (#47008)
Addresses the issue of parallel tool approvals freezing Assistant and causing UX ambiguity around what happens if only part of multiple dependent invocations is approved. Two layers at which this is addressed: 1. Tightens prompt to clarify approval tools must be issued one per step, not in parallel. 2. In case something slips past the prompt, this also auto-denies all but the first `approval-required` tool call when the model issues multiple in the same step, so the model is forced to reissue them sequentially. **Demo** The following chats demo me explicitly asking Assistant to run those approval-gated tools in parallel, and the Assistant correctly invokes them sequentially instead. | Parallel `execute_sql` request | Parallel `deploy_edge_function` request | |--------|--------| | <img width="1820" height="4240" alt="CleanShot 2026-06-16 at 16 20 23@2x" src="https://github.com/user-attachments/assets/3ddad61d-24d6-4e5b-8572-c261755f3a03" /> | <img width="1738" height="2726" alt="CleanShot 2026-06-16 at 16 22 59@2x" src="https://github.com/user-attachments/assets/77a0aa20-e3cb-4061-b270-ab2a96d1e64a" /> | As shown in [this trace](https://www.braintrust.dev/app/supabase.io/p/Assistant/trace?object_type=project_logs&object_id=5a8d02e5-b3b6-40cc-ba76-ecee286478f4&r=a3a37857-95df-4a95-a5a8-818ea305b2a5&s=a3a37857-95df-4a95-a5a8-818ea305b2a5), parallel tool calls are still allowed for context gathering tools that don't require approval: <img width="1051" height="517" alt="CleanShot 2026-06-16 at 16 25 42@2x" src="https://github.com/user-attachments/assets/00d208b1-6131-4b8e-910e-e92fd2a79a5b" /> Closes AI-803 <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Improvements** * Approval-required operations are now processed sequentially rather than in parallel within the AI Assistant. * **Tests** * Added comprehensive test coverage for the parallel approval prevention logic. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Joshen Lim
|
1baaded0bb |
Consolidate execute-sql-query into execute-sql-mutation (#46944)
## Context Just some clean up as I was going through stuff - `useExecuteSqlQuery` is deprecated and not used at all - As such `execute-sql-query` is technically irrelevant, the more relevant file is `execute-sql-mutation` - Hence opting to consolidate `execute-sql-query` into `execute-sql-mutation` - Also removing `ExecuteSqlError` since its just re-exporting the `ResponseError` type There's a lot of file changes but its essentially just updating the importing statements across the files |
||
|
Ali Waseem
|
289accd7d4 |
chore: Update time from 10 to 20 minutes for pause projects fallback (#46479)
## 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? Minor update for long running state thresholds when updating project sizes <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Chores** * Increased the minimum threshold for recognizing long-running operations from 10 to 20 minutes, reducing premature fallback behavior and improving stability for longer tasks. * **Tests** * Updated test suite to validate the new 20-minute threshold across default, missing, small, and larger restore scenarios. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Ali Waseem
|
1c2d28d5b3 |
chore: wrap local storage into helper methods that are safer (#46628)
## 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? - Noticing our code we have many patterns of calling localstorage and handling those errors - We should add those in a single well tested file - Handle those errors in the singleton which makes it easier for us to debug customer issues. Logger is outputing local storage warnings for feature we expose - Side effect of this is random crashes on studio when local storage isn't available or handled correctly <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Refactor** * Improved browser storage handling across the app for more reliable persistence and graceful behavior in restricted or non-browser environments (settings, previews, charts, tabs, sign-in/session flows, integrations, and UI state). * **New Features** * Introduced a safe storage layer to standardize and harden local/session persistence. * **Tests** * Added comprehensive tests covering the new safe storage behavior. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Wen Bo Xie
|
f041ec4124 |
feat: update API keys instruction for AI Assistant (#45401)
<!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Documentation** * Updated authentication docs: require startup-time validation of required env vars and introduce JSON-mapped publishable/secret key handling. * Added JWKS-based asymmetric JWT verification guidance and clarified symmetric (HS256) handling; deprecated anon/service-role keys. * Clarified Authorization control flow: when handlers must fully authenticate vs. when module-scoped JWKS verification applies. <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
||
|
Andrey A.
|
b28f91741f | fix(self-hosted): reveal and copy secret api key in project settings (#46592) | ||
|
Andrey A.
|
ca9b02b5ac | feat(self-hosted): add minimal project settings (#46554) | ||
|
Joshen LimandAli Waseem
|
1d203f6c93 |
feat: Support CLI for Vector buckets (#46381)
## Context > [!IMPORTANT] > Will open up for review once CLI PR is merged and deployed so that it's easier to test Related PR: https://github.com/supabase/cli/pull/5230 Adding support for vector buckets for local CLI - will need to be tested locally via `pnpm run dev:studio-local` ## To test There's a bit of testing instructions in the linear ticket [here](https://linear.app/supabase/issue/FE-3474/show-vector-buckets-in-local-admin-studio) as it involves using a branch of CLI - otherwise do reach out to Fabrizio if any help might be needed, but generally: ### Local CLI You might need to manually set `isCli` to `true` in `StorageMenuV2` if the "Vectors" nav item isn't showing up on the storage UI given we're testing via `pnpm run dev:studio-local` - [x] Can create bucket - [x] Can delete bucket - [x] Can create indexes - [x] Can insert data into indexes (via FDW) - [x] Can delete indexes Known issues (that aren't directly solvable from FE end) Reach out to Fabrizio for context as we were both investigating this - PG database needs to be on 17.6 (otherwise there's no S3 vectors FDW) - Storage version needs to be on 1.59.0 ### Self-hosted (This might be tricky to actually test, but just ensure that the code satisfies this) - [x] Cannot see vector buckets ### Hosted - [x] Everything works status quo <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Vector bucket management UI and platform APIs (create/list/delete buckets & indexes) * Local S3 credentials endpoint and client-side hook for self‑hosted/CLI use * **Bug Fixes** * Improved S3 vector setup notifications and clearer error guidance for manual installation * **Refactor** * Deployment-mode gating: platform vs CLI/self‑hosted now controls feature visibility and page behavior * **Tests** * Added suites covering deployment-mode gates and vector bucket error/usage scenarios * **Chores** * Build env updated to expose local S3 credential vars <!-- review_stack_entry_start --> [](https://app.coderabbit.ai/change-stack/supabase/supabase/pull/46381?utm_source=github_walkthrough&utm_medium=github&utm_campaign=change_stack) <!-- review_stack_entry_end --> <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Co-authored-by: Ali Waseem <waseema393@gmail.com> |
||
|
Matt Rossman
|
65fab30935 |
feat(ai): judge tool inputs, add storage guidance and permissive RLS evals (#46168)
Adding broad RLS policies to public buckets can cause users to expose more than they expected, like the ability to list all profile pictures on an app. This patches Assistant with knowledge to follow our latest guidance on restrictive RLS policies for storage buckets https://github.com/supabase/supabase/pull/46172 **Changes** - Adds Storage bucket evals for public website assets and avatar access patterns to distinguish public vs private bucket use cases - Adds eval for overly permissive table policies - Adds `storage` knowledge so Assistant distinguishes public buckets, private buckets, object reads, and object listing. - Adds `includeToolCallInputs` option for scorer transcripts so LLM judges can evaluate proposed SQL/tool actions. - Bumps max step count to 10 since storage knowledge may incur another tool call (also 10 is recommended [here](https://vercel.com/academy/ai-sdk/multi-step-and-generative-ui#why-multi-step-is-required) for complex multi-tool scenarios) **References** - https://supabase.com/docs/guides/storage/buckets/fundamentals#public-buckets - https://supabase.com/docs/guides/storage/security/access-control - https://github.com/supabase/supabase/pull/46172 **Notes:** - These prompt tweaks are not meant to be exhaustive fixes, they are mainly hotfixes intended to hold us out until these cases can be addressed more deeply in skills/docs and tracked in a central evals Closes AI-676 Closes AI-756 <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Added Storage knowledge resource for the assistant covering Supabase Storage access patterns and RLS guidance. * Added three evaluation cases: two for Storage (marketing assets, avatars) and one for RLS policy generation for user profiles. * **Improvements** * Evaluators now include tool call inputs when judging conversations. * Assistant prompts and generation enhanced with richer Storage/RLS guidance and extended streaming limits. * **Tests** * Added test ensuring tool call inputs are included in serialized thread context. <!-- review_stack_entry_start --> [](https://app.coderabbit.ai/change-stack/supabase/supabase/pull/46168?utm_source=github_walkthrough&utm_medium=github&utm_campaign=change_stack) <!-- review_stack_entry_end --> <!-- end of auto-generated comment: release notes by coderabbit.ai --> |
