DB migration compat / Back-compat — Current branch migrations with ${{ needs.check-migrations-changed.outputs.base_branch }} branch code (push) Has been cancelled
DB migration compat / Forward-compat — Current branch code with ${{ needs.check-migrations-changed.outputs.base_branch }} branch migrations (push) Has been cancelled
<!--
Make sure you've read the CONTRIBUTING.md guidelines:
https://github.com/stack-auth/stack-auth/blob/dev/CONTRIBUTING.md
-->
<!-- CURSOR_SUMMARY -->
---
> [!NOTE]
> **High Risk**
> Touches authentication and OAuth token/authorize flows and changes how
client requests are validated, so regressions could cause widespread
login/client-access failures. Also includes a data migration that alters
effective security posture for existing projects.
>
> **Overview**
> Adds a **project-level toggle**
(`project.requirePublishableClientKey`) to control whether client
requests/OAuth flows must include a publishable client key, including a
DB migration that backfills existing projects to require it.
>
> Backend auth now treats the publishable client key as *optional when
allowed*, introducing a public sentinel (`__stack_public_client__`) and
returning a new specific error
(`PUBLISHABLE_CLIENT_KEY_REQUIRED_FOR_PROJECT`) across smart request
auth + OAuth `authorize`/`callback`/`token` endpoints.
>
> Dashboard and SDKs update key generation/display and request
construction to handle missing publishable keys, expose an advanced
toggle on the Project Keys page, and extend internal config overrides to
support a new `project` level; E2E/tests and schema fuzzing are expanded
accordingly, and CI adds a forward-compat migration check job when
back-compat fails.
>
> <sup>Written by [Cursor
Bugbot](https://cursor.com/dashboard?tab=bugbot) for commit
5d06c08613. This will update automatically
on new commits. Configure
[here](https://cursor.com/dashboard?tab=bugbot).</sup>
<!-- /CURSOR_SUMMARY -->
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Project-level config to require publishable client keys; migration
applied to existing projects.
* **Improvements**
* Auth flows now support optional publishable client keys with explicit
validation and a sentinel for keyless OAuth.
* Dashboard/UI and SDKs handle publishable keys as optional and
conditionally show/generate them.
* Admin/client APIs extended to manage project-level overrides.
* **Bug Fixes**
* Key validation behavior aligned with project config.
* **Tests**
* Expanded E2E and unit tests covering optional/required publishable-key
scenarios.
* **Documentation**
* Spec and knowledge docs updated to describe the sentinel and config
behavior.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
DB migrations are backwards-compatible / Test migrations with ${{ needs.check-migrations-changed.outputs.base_branch }} branch code (push) Has been cancelled
https://www.loom.com/share/09a89533039d4bd4814332ec0728a30f
<!--
Make sure you've read the CONTRIBUTING.md guidelines:
https://github.com/stack-auth/stack-auth/blob/dev/CONTRIBUTING.md
-->
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Added batch analytics event submission API endpoint
* Enhanced session replay timeline with visual markers for page views
and click events
* Display click event counts on replay list items
* Implemented client-side event tracking for page views and clicks
* **Bug Fixes**
* Session replay now properly errors when analytics feature is disabled
* **Tests**
* Added end-to-end tests for analytics events batch API with validation
and querying
* Updated session replay test expectations for analytics error handling
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
<!--
Make sure you've read the CONTRIBUTING.md guidelines:
https://github.com/stack-auth/stack-auth/blob/dev/CONTRIBUTING.md
-->
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Subscription IDs are now included in product listings and UI data for
clearer subscription tracking.
* Cancellation can be performed by subscription ID as well as by
product; client and template APIs support passing a subscription
identifier.
* **Tests**
* End-to-end tests added/updated to cover canceling subscriptions via
subscription ID and updated listing snapshots.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
<!--
Make sure you've read the CONTRIBUTING.md guidelines:
https://github.com/stack-auth/stack-auth/blob/dev/CONTRIBUTING.md
-->
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Browser-side event tracker with batching, navigation & click capture
and background/keepalive delivery
* Server endpoint to accept batched analytics events and associate them
with session replay segments
* Client APIs to send analytics batches and integrate with session
replay
* **Bug Fixes / UX**
* Pausing replay now uses the UI-facing playback time for more accurate
pause positions
* Replay endpoint now returns a clear analytics-disabled error
(ANALYTICS_NOT_ENABLED) when analytics is off
* **Tests**
* End-to-end tests covering batch ingestion, validation, and replay
timing behavior
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
<!--
Make sure you've read the CONTRIBUTING.md guidelines:
https://github.com/stack-auth/stack-auth/blob/dev/CONTRIBUTING.md
-->
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Added new session replay analytics columns to ClickHouse for enhanced
tracking and reporting
* **Refactor**
* Renamed session recording segment identifier across APIs and data
models from `tab_id` to `session_replay_segment_id`
* Updated internal data structures and type definitions to align with
new naming convention
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
https://www.loom.com/share/3b7c9288149e4f878693281778c9d7e0
## Todos (future PRs)
- Fix pre-login recording
- Better session search (filters, cmd-k, etc)
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Analytics → Replays: session recording & multi-tab replay with
timeline, speed, seek, and playback settings; dashboard UI for listing
and viewing replays.
* **Admin APIs**
* Admin endpoints to list recordings, list chunks, fetch chunk events,
and retrieve all events (paginated).
* **Client**
* Client-side rrweb recording with batching, deduplication, upload API
and a send-batch client method.
* **Configuration**
* New STACK_S3_PRIVATE_BUCKET for private session storage.
* **Tests**
* Extensive unit and end-to-end tests for replay logic, streams,
playback, and APIs.
* **Chores**
* Removed an E2E API test GitHub Actions workflow.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
### Context
Some of our users' emails were getting stuck in sending. The long delays
in processing the retries caused a vercel function timeout.
### Summary of Changes
We refactor the low level email sending functions to remove the retry
logic there. We kick it up to the email queue step. Additionally, we
flag emails to be retried when they encounter issues but leave it for a
future iteration to actually perform the retry. We perform an
exponential backoff with a random component to decide when they have to
be retried. We also make some small adjustments to the queuing function
to not queue skipped emails.
When an email fails to send during the sending function, we check to see
if it is a retryable error or not. Some errors are transient and trying
again may succeed while others indicate deeper issues. If it is
retryable, and the max number of retry attempts hasn't been reached, we
set `nextSendRetryAt` to a time determined by an exponential backoff
calculation function. When the queuing function looks for emails to
queue, it doesn't just pick up the `SCHEDULED`. emails whose
`scheduledAt` time <= `NOW()`, but also those emails whose
`nextSendRetryAt` time <= `NOW()`. What this means in practice is that
one iteration of the `email-queue-step` will mark emails as retryable
while another iteration will perform the retry. This should be cleaner
and prevent long delays in the `email-queue-step` process due to
retries. This also makes it easier to scale up the number of retries if
need be.
DB migrations are backwards-compatible / Test migrations with ${{ needs.check-migrations-changed.outputs.base_branch }} branch code (push) Has been cancelled
### Context
Recently, a user raised [this
issue](https://github.com/stack-auth/stack-auth/issues/1144), which
indicated that `tokenOverrides` were not being respected/used in the
`getUser()` function. If we trace the flow through this function, we see
`this._getSession -> this._getOrCreateTokenStore -> _createCookieHelper
-> createCookieHelper -> createNextCookieHelper -> await rscHeaders()`.
What this means is that even when a `requestLike tokenOverride` was
passed, we would not end up using it because the `createCookieHelper`
call occurs before the extant override checking logic in
`getOrCreateTokenStore`, and the `createCookieHelper` didn't check the
override but only the default `tokenStoreInit`. This caused the error to
propagate up.
### Summary of Changes
We check the `tokenStoreOverride` in the `createCookieHelper` function
now, preventing this issue from happening. We also add extra test
coverage to verify that overrides are respected, and don't overwrite the
default token store.
### Out of Scope Discussion
The original issue was raised with a `bun` runtime running `next.js`
code. There seems to be some incompatibility between `bun 1.3.8` and
`nextjs 15+`, not just with our backend but with fetching and working
with responses from any `nextjs` server.
<!-- CURSOR_SUMMARY -->
> [!NOTE]
> **High Risk**
> Touches core sign-up/auth flows and user restriction semantics
(including new DB constraints) and introduces dynamic rule
evaluation/logging; misconfiguration or CEL/parser bugs could block
sign-ups or incorrectly restrict users.
>
> **Overview**
> Introduces **CEL-based sign-up rules** (config-driven) that are
evaluated during password/OTP/OAuth sign-ups and anonymous upgrades;
matching rules can reject sign-ups or mark users as admin-restricted,
and triggers are logged for analytics.
>
> Extends `ProjectUser` with `restrictedByAdmin` plus public/private
restriction details, updates restriction computation/filtering, and
exposes these fields via user CRUD (including validation + DB constraint
enforcing consistency when unrestricted).
>
> Adds a new dashboard **Sign-up Rules** page with a visual condition
builder (CEL <-> visual tree), drag-reorder by priority, per-rule 48h
sparkline analytics via a new hidden internal endpoint, and adds
user-page UI to view/edit manual restrictions. Also refactors ClickHouse
client initialization to require env vars (removing
`isClickhouseConfigured` checks) and adjusts CI container startup wait
time.
>
> <sup>Written by [Cursor
Bugbot](https://cursor.com/dashboard?tab=bugbot) for commit
2141e689e8c1b72303b805e9234f996010d0880. This will update automatically
on new commits. Configure
[here](https://cursor.com/dashboard?tab=bugbot).</sup>
<!-- /CURSOR_SUMMARY -->
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Sign-up Rules: visual rule builder, in-project CRUD with drag-reorder,
per-rule analytics, backend evaluation, and admin UI.
* Admin user restrictions: dashboard controls, banners/status,
public/private admin details surfaced in user views.
* **APIs & Schema**
* Config and user schemas extended; new SignUpRejected error and sign-up
rule types added.
* **Tests**
* Extensive unit and E2E coverage for rules, parser, evaluator,
analytics, and restricted-user flows.
* **Docs**
* Editorial guidance added to AGENTS.md.
* **Chores**
* DB statement timeout, updated clean script, minor dependency
additions.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
https://www.loom.com/share/bb7abfde507f40d386ee856f5ffbd506
<!--
Make sure you've read the CONTRIBUTING.md guidelines:
https://github.com/stack-auth/stack-auth/blob/dev/CONTRIBUTING.md
-->
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* USD-based refund system enabling partial and full refunds with
explicit USD amounts
* Per-entry refund selection with granular quantity controls in refund
dialogs
* **Bug Fixes**
* Stronger refund validation and error handling to prevent invalid or
out-of-bounds refunds
* **Tests**
* Expanded end-to-end coverage for refund edge cases and scenarios
* **Style**
* Improved refund dialog UI with contextual alerts and better controls
<sub>✏️ Tip: You can customize this high-level summary in your review
settings.</sub>
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
### Summary of Changes
Previously, on the Swift SDK, the `signInWithOAuth` function wasn't
working. In this PR, we fix it by having the `getOAuthUrl` function to
actually redirect correctly. Note that to do so, we updated the
`validRedirectUrl` check on the backend to accept app native redirects
(from our new trusted url scheme). Another thing to note is that we
added functionality to the `TokenStore` abstraction to conditionally
refresh the access token that the user is trying to fetch if it is
expired/close to expiring if possible. `getOAuthUrl` will attempt to get
a valid access token, and thus will rely on our algorithm documented in
`utilities.md`.
The specs serve as the source of truth.
We go further and implement Apple Native sign in. To do so, we have it
hit a new route on the backend and verify the `jwtToken` retrieved by
the sdk against an Apple-provided set of `jwks`. We use jose to do so,
in line with the rest of the codebase.
We take this opportunity to refactor the oauth provider route owing to
the amount of duplicated logic. Additionally, to enable the apple sign
in, users will have to update the Apple authentication method modal on
the dashboard and add accepted bundle ids. These are identifiers for
projects, and we will check the `JWT` on the backend to make sure the
audience is set to an accepted bundleId.
We also update the Apple modal to be more informative.
### Using the new Features
To use the Apple native sign in, users will have to 1) sign up with an
apple developer account, 2) set up their bundleids for their projects by
connecting them to the apple developer account, 3) update the Stack-Auth
Authentication Methods dashboard apple modal with the relevant fields.
Then, trying to sign in with apple with our Swift SDK will use the apple
native sign in.
### UI Changes
Renamed the fields in the apple modal. Added a new field for bundle ids.
See below.
https://github.com/user-attachments/assets/0e760c0e-3198-4818-ac7f-4900d7a125bb
Co-authored-by: Konstantin Wohlwend <n2d4xc@gmail.com>
<!--
Make sure you've read the CONTRIBUTING.md guidelines:
https://github.com/stack-auth/stack-auth/blob/dev/CONTRIBUTING.md
-->
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **Security**
* Added client-side access checks on payments endpoints and expanded
customer-type handling (including a new "custom" type).
* **SDK / Client**
* Client interface methods now accept explicit request types
(client/server/admin) to route requests appropriately.
* **Server**
* New server-side product listing to support server requests and
caching.
* **Tests**
* E2E tests updated to use a fast sign-up flow and pass authentication
tokens for authorized requests.
<sub>✏️ Tip: You can customize this high-level summary in your review
settings.</sub>
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
<img width="553" height="471" alt="Screenshot 2026-01-14 at 12 16 36 PM"
src="https://github.com/user-attachments/assets/9f32473d-5294-4cf7-b527-0668fb04ae47"
/>
<img width="630" height="514" alt="Screenshot 2026-01-14 at 12 17 06 PM"
src="https://github.com/user-attachments/assets/b17f57f7-148d-4438-b337-df7516d1793e"
/>
<!--
Make sure you've read the CONTRIBUTING.md guidelines:
https://github.com/stack-auth/stack-auth/blob/dev/CONTRIBUTING.md
-->
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Expanded Stripe webhooks: handles invoice and one‑time/subscription
events, sends templated payment receipt and failure emails, posts
chargeback alerts to Telegram.
* Customer invoices API plus client and UI support for listing invoices;
backend stores invoice status, total, and hosted URL.
* **Tests**
* Added end‑to‑end tests for new webhook scenarios (receipts, failures,
chargebacks) and invoices API with email outbox checks.
* **Chores**
* Centralized Telegram helpers and improved formatting, validation, and
reliability.
<sub>✏️ Tip: You can customize this high-level summary in your review
settings.</sub>
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
<!-- CURSOR_SUMMARY -->
---
> [!NOTE]
> Introduces end-to-end invoice visibility and payment notifications.
>
> - **Emails:** Adds default `payment_receipt` and `payment_failed`
templates and sends them from Stripe webhooks for one-time and
subscription payments (skips non‑uncollectible failures); resolves
recipients for users/teams.
> - **Webhooks:** Expands handled events; upserts invoices on
`invoice.*`; stricter unknown-type handling; adds Telegram chargeback
alert; refactors init script Telegram sending.
> - **Data model:** Extends `SubscriptionInvoice` with `status`,
`amountTotal`, `hostedInvoiceUrl` and writes them via
`upsertStripeInvoice`.
> - **API/SDK/UI:** New paginated `GET
/payments/invoices/{customer_type}/{customer_id}`; client interface
(`listInvoices`, hooks) and template Payments panel render an invoices
table.
> - **Tests:** E2E for invoices access, webhook behaviors, and email
delivery.
>
> <sup>Written by [Cursor
Bugbot](https://cursor.com/dashboard?tab=bugbot) for commit
edc8fe5651. This will update automatically
on new commits. Configure
[here](https://cursor.com/dashboard?tab=bugbot).</sup>
<!-- /CURSOR_SUMMARY -->
### Summary of Changes
We would like to setup a payment settings page. Here, developers should
be able to toggle test mode, see their stripe connection status, and
adjust the payment method configs.
## Test Mode Toggle
This will exist in concert with the test mode banner.
## Stripe Connection Status
While users cannot see the page unless they create a stripe account,
they can still see it if they've created the stripe connected account
but haven't finished onboarding. This is a handy place for them to
finish their onboarding.
## Payment Methods
We would like developers using our payments feature to be able to set
what payment options should be made available to their users.
Consequently, we create a route and a page on the dashboard which hits
that route to update what payment options are made available. The UI
stores "pending changes" which represent updates to be made to the
payment method configs corresponding to that project's connected stripe
account. These are then sent to the backend, validated with a schema,
and then updated using stripe.
We also note that some payment methods have dependencies on others: for
example, the "apple pay" method cannot be enabled if the "debit/credit
cards" method is not enabled. We note the two cases where it is observed
to happen and raise an alert using `toast` to make it clear to the
developer, and make it extensible in case other dependencies are added
in the future. To ensure synchronization between the frontend dashboard
UI and the backend route handler on the payment method names, we have
both pull from a shared utility file. This ensures only one update will
need to be made.
**NOTE 1:** We chose to build our own component rather than using the
Stripe embedded component as the Stripe component is still in
pre-release mode.
**NOTE 2:** To disable specific payment methods for all our users, we
should update the platform account config in Stripe for stack-auth. This
will prevent said payment method from being made available to them.
**NOTE 3:** We skip the multi-account method config isolation test
because the stripe mock server does not support testing with multiple
accounts. However, the logic of the test has been verified with a real
stripe account.
### UI Demo
For this demo, I had a pre-created checkout link for a one-time purchase
of a product for 100$.
https://github.com/user-attachments/assets/a0139ee8-a9ce-480c-b8b5-9b5fb1e9c15f
We update the sdk dependencies (the ones present in
`package-template.json`) to the latest versions. Since several packages
have major version bumps, this results in a variety of breaking changes
that have been handled here. Incidentally, when possible, we update
similar dependencies across the codebase.
We decide to defer the tailwind update to another PR owing to its scale.
The rest of the updates and changes have been catalogued below:
1.
[Bumping](https://github.com/panva/oauth4webapi/blob/v3.x/CHANGELOG.md)
`oauth4webapi` to 3.8.3: this was a major version changed. While there
were no compatibility issues in the sdk, there were several breaking
changes in `stack-shared`. Namely:
a. The removal of `isOauth2Error`. We used this to check if the results
of our `oauth4webapi` api invocations had issues. The functions were
changed to explicitly throw either `ResponseBodyErrors` or
`AuthorizationResponseErrors`, so the code was reworked to account for
that with no loss in error handling.
b. Dropping of support for http broadly: `oauth4webapi` now only accepts
https. This is desired, but I add a carve out for our test environments
only.
c. `refreshTokenGrantRequest` and `authorizationCodeGrantRequest` now
require `clientAuthentication` to be passed explicitly to them.
d. Changes in how we handle our `MultiFactorAuthenticationRequired`
error: This is an error that we created and is passed to the
`oauth4webapi` API if there are MFA issues. Since the
`processAuthorizationCodeResponse` now explicitly throws a
`ResponseBodyError`, we access the error cause from the body of the
error instead.
2. [Bumping](https://github.com/Qix-/color/releases) `color` to 5.0.4:
this was a major version bump. Simple type checking change, I checked
the API for the correct interface.
3.
[Bumping](https://github.com/MasterKale/SimpleWebAuthn/blob/master/CHANGELOG.md)
`simplewebauthn` to 13.2.2: two major version bumps, but no
incompatibilities surprisingly
4. [Bumping](https://github.com/jshttp/cookie/releases) `cookie` to
1.1.1: this was a major version bump.
a. Changing `parse` to `parseCookie`. In the most recent version,
`parse` is still maintained as an alias for `parseCookie` for backwards
compatibility, but I thought it would be best to change it over now. No
change in functionality.
b. Typing is now strongly enforced. A cookie can be `string |
undefined`, and the `Cookies` are now `Record<string, string |
undefined>`. We already have code to handle if a cookie is returned as
undefined/ null, so the changes here were more to ensure type
compatibility rather than big changes in functionality.
5. [Bumping ](https://github.com/isaacs/rimraf#readme)`rimraf` to 6.1.2:
No breaking changes, mostly just bug fixes.
6. [Bumping](https://github.com/panva/jose/releases?page=1) `jose` to
6.1.3: This is another major version bump. We update it across the
codebase to ensure compatibility. We use this for importing and
processing jwk tokens. There are a few big changes in the version bump,
but the only one that applies to us is that `importJwk` now yields a
`CryptoKey` instead of a `KeyObject` in Node.js. However, this doesn't
appear to break our code. We use `importJwk` in
`stack-auth/packages/stack-shared/src/utils/jwt.tsx`.
7. [Bumping](https://github.com/react-hook-form/resolvers/releases)
`hookform/resolvers` to 5.2.2 (two major version jumps), and
consequently bumping `react-hook-form` to 7.70.0: We already use the
patterns that `hookform/resolvers`' latest versions seem to be
enforcing. The only other breaking change is that it requires version
7.55.0+ of `react-hook-form`. Though we should pay attention to any
interactions with zod and `hookform/resolvers`, some people have
reported compatibility issues if they aren't using the latest compatible
versions of both.
8. [Bumping](https://github.com/jquense/yup/blob/master/CHANGELOG.md)
`yup` to 1.7.1: this was a minor version change, but we had
incompatibility issues with this change. Versions 1.4.1 and 1.7.1 cannot
exist in the same codebase due to incompatibility, so we bumped it up
across the codebase, including in peer dependencies.
9. Some minor version changes for some packages, but these were mostly
bug fixes.
10. **Edited to add**: Bumping freestyle to 0.1.6, and reworking the
freestyle mock server. In 0.1.6, freestyle changed their API in two
ways:
a. We're now supposed to hit their `execute/v2/...` endpoint and
b. They've flattened the `config` argument to `serverless.runs.create`.
These changes are minor, but are important. As part of a general suite
of dependency bumps, this was judged to fit here.
We have linked the changelogs for the packages on each line.
