Codestin Search App

jiggyjigsj · 2026-08-11T07:21:33Z

Summary

move public status-page email signup into a service-layer workflow
create the pending subscriber and send its verification email in one server request
remove the client-side follow-up mutation and redundant public email endpoint
stop returning subscriber verification tokens to the browser

Root cause

The status-page UI first created a pending subscriber, then relied on a second client-side tRPC mutation to send the verification email. If that follow-up request did not run or failed, the subscriber remained pending without receiving a link, and the pending guard blocked another attempt.

Validation

pnpm --filter @openstatus/services check
changed-file oxlint
focused service test against the libSQL image pinned by upstream CI: 5 steps passed
formatting and documentation-reference checks passed as part of pnpm verify

The repository-wide verifier still fails on pre-existing Deno type errors involving Buffer in apps/workflows/src/cron/checker.ts and packages/api; the affected service package type-checks cleanly.

vercel · 2026-08-11T07:21:40Z

@jiggyjigsj is attempting to deploy a commit to the OpenStatus Team on Vercel.

A member of the Team first needs to authorize it.

cubic-dev-ai

All reported issues were addressed across 6 files

_{Reply with feedback, questions, or to request a fix.

Re-trigger cubic}

mxkaske · 2026-08-11T08:41:28Z

@claude review

claude · 2026-08-11T08:41:45Z

Claude finished @mxkaske's task in 2m 51s —— View job

Claude's Review

Todos:

Gather context (read diff, changed files)
Understand request (code review requested)
Review packages/services/src/page-subscriber/subscribe.ts
Review router changes (email/index.ts, statusPage.ts)
Review test file and client-side changes
Evaluate cubic-dev-ai's two flagged issues (TOCTOU race, email-failure rollback)
Post final review

Solid refactor overall — consolidating the create + verification-send into one server-side transaction closes the original client-round-trip gap, and the router/UI cleanup is consistent (no leftover consumers of the removed id/token fields or the deleted sendPageSubscriptionVerification endpoint; form-subscribe-email.tsx's toast.promise already surfaces errors from the single mutation, so dropping the header's redundant onError is safe).

On cubic-dev-ai's two findings:

#2 (P1, subscribe.ts:51, email-provider failure) — confirmed, real regression risk. upsertSelfSignupSubscriber's transaction commits before channel.sendVerification is called. If the provider call throws (rate limit, transient 5xx, invalid domain config, etc.), the pending row is already durable, and hasPendingSubscriber will block the same email/page pair for up to 7 days (VERIFICATION_EXPIRY_MS) with no path to retry. That's the exact "stuck pending, no link" failure mode the PR's own root-cause section describes — it's just moved from "client mutation didn't fire" to "provider call failed." Worth at least catching the send failure and clearing/expiring the row (or deleting it if newly created) so the user can immediately retry, rather than leaving them locked out for a week.

#1 (P2, subscribe.ts:22, TOCTOU on the pending check) — the race is real, but the stated consequence is partly wrong. hasPendingSubscriber reads outside the transaction that upsertSelfSignupSubscriber opens, so two near-simultaneous requests for the same (email, page) can both pass the guard and both reach the upsert. However, the table has a partial unique index (idx_page_subscriber_email_page_active in packages/db/src/schema/page_subscribers/page_subscribers.ts:69) enforcing one active email subscriber per page, and SQLite/libSQL serializes concurrent writers — so this cannot create two pending rows as the comment claims. What can happen: request A inserts and sends a verification email; request B's transaction then finds A's just-committed row, takes the "merge pending" branch (refreshes expiresAt, reuses the same token), and subscribeSelfSignupSubscriber sends a second verification email for that same token. So the real, narrower bug is a duplicate-email race, not duplicate rows. Given this is a public/unauthenticated endpoint, moving the pending check inside the same transaction as the upsert (or having the merge branch short-circuit the caller's send when it just refreshed an already-pending row) would close it.

No other correctness issues found in the diff. One minor note: there's no test exercising the concurrent-request race for issue #1, understandably hard to do reliably against libSQL — not a blocker, just noting the gap given the guard's stated purpose is anti-spam.

cubic-dev-ai

1 issue found across 6 files (changes from recent commits).

Prompt for AI agents (unresolved issues)


Check if these issues are valid — if so, understand the root cause of each and fix them. If appropriate, use sub-agents to investigate and fix each issue separately.


<file name="packages/services/src/page-subscriber/expire-verification.ts">

<violation number="1" location="packages/services/src/page-subscriber/expire-verification.ts:33">
P2: A delayed failure from an older verification send can invalidate a newer resend, leaving the user with an unusable link. The compensation needs an attempt/version guard (or a rotated token per resend) before expiring the row.</violation>
</file>

_{Reply with feedback, questions, or to request a fix.

Re-trigger cubic}

cubic-dev-ai

All reported issues were addressed across 4 files (changes from recent commits).

_{Tip: Review your code locally with the cubic CLI to iterate faster.

Re-trigger cubic}

cubic-dev-ai

All reported issues were addressed across 4 files (changes from recent commits).

_{Tip: Review your code locally with the cubic CLI to iterate faster.

Re-trigger cubic}

cubic-dev-ai

All reported issues were addressed across 5 files (changes from recent commits).

_{Tip: Review your code locally with the cubic CLI to iterate faster.

Re-trigger cubic}

jiggyjigsj · 2026-08-13T16:46:31Z

@mxkaske This should be ready to review.

This was referenced Aug 11, 2026

fix: send subscriber verification in signup service Cooklist/openstatus#2

Closed

[Self Hosted]: Email status page subscriptions create pending subscriber but do not send verification email #2571

Open

jiggyjigsj marked this pull request as ready for review August 11, 2026 07:28

cubic-dev-ai Bot reviewed Aug 11, 2026

View reviewed changes

Comment thread packages/services/src/page-subscriber/subscribe.ts Outdated

Comment thread packages/services/src/page-subscriber/subscribe.ts Outdated

cubic-dev-ai Bot reviewed Aug 11, 2026

View reviewed changes

Comment thread packages/services/src/page-subscriber/expire-verification.ts

Comment thread packages/services/src/page-subscriber/upsert.ts Outdated

Comment thread packages/services/src/page-subscriber/subscribe.ts Outdated

cubic-dev-ai Bot reviewed Aug 11, 2026

View reviewed changes

cubic-dev-ai Bot reviewed Aug 12, 2026

View reviewed changes

Comment thread packages/services/src/page-subscriber/upsert.ts

Comment thread packages/services/src/page-subscriber/upsert.ts

cubic-dev-ai Bot reviewed Aug 12, 2026

View reviewed changes

Comment thread packages/services/src/page-subscriber/upsert.ts Outdated

jiggyjigsj added 7 commits August 18, 2026 00:45

fix: send subscriber verification in signup service

e53064d

fix: recover subscriber verification delivery

38a53fa

fix: harden subscriber verification retries

6b34c9e

fix: serialize subscriber verification claims

488ef4d

fix: route legacy subscriber verification through service

3c342e5

fix: isolate concurrent subscriber audit changes

360ba97

fix: allow configuring notification email sender

eb3a311

jiggyjigsj force-pushed the codex/status-subscriber-email-service branch from bc48542 to eb3a311 Compare August 18, 2026 05:45

Uh oh!

Uh oh!

Conversation

jiggyjigsj commented Aug 11, 2026 • edited by cubic-dev-ai Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Root cause

Validation

Uh oh!

vercel Bot commented Aug 11, 2026

Uh oh!

cubic-dev-ai Bot left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

mxkaske commented Aug 11, 2026

Uh oh!

claude Bot commented Aug 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Claude's Review

Uh oh!

cubic-dev-ai Bot left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

cubic-dev-ai Bot left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

cubic-dev-ai Bot left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

cubic-dev-ai Bot left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

jiggyjigsj commented Aug 13, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

jiggyjigsj commented Aug 11, 2026 •

edited by cubic-dev-ai Bot

Loading

cubic-dev-ai Bot left a comment •

edited

Loading

claude Bot commented Aug 11, 2026 •

edited

Loading

cubic-dev-ai Bot left a comment •

edited

Loading

cubic-dev-ai Bot left a comment •

edited

Loading

cubic-dev-ai Bot left a comment •

edited

Loading

cubic-dev-ai Bot left a comment •

edited

Loading