The latest updates on your projects. Learn more about Vercel for GitHub.

1 Skipped Deployment

Project	Deployment	Actions	Updated (UTC)
docs	Skipped		May 22, 2026 2:15am

PR Summary

Medium Risk
Moderate risk because it refactors the core email-sending path and introduces new provider selection/fallback behavior that could affect delivery, batching, and unsubscribe handling if misconfigured.

Overview
Refactors mailer.ts to use a MailProvider abstraction with startup-time provider auto-detection and sequential fallback dispatch (Resend → SES → SMTP → Azure), including batch-send support when available.

Adds new SES (AWS_SES_REGION) and SMTP (SMTP_HOST/PORT/USER/PASS/SECURE) providers (via nodemailer), factors email validation/unsubscribe/header injection into prepare.ts, and updates batch behavior to treat per-email preparation/unsubscribe errors as per-entry results.

Extends env schema with the new variables (plus an envBoolean helper to correctly parse string booleans), and updates .env.example, Helm values/schema, and self-hosting docs to document the expanded email provider configuration.

^{Reviewed by Cursor Bugbot for commit 7705d31. Configure here.}

Greptile Summary

This PR refactors the email subsystem into a provider-abstraction layer, adding AWS SES (via nodemailer + @aws-sdk/client-sesv2) and SMTP as new providers alongside the existing Resend and Azure backends. An envBoolean helper is introduced to fix z.coerce.boolean("false") === true under skipValidation: true, and all four providers are documented in .env.example, Helm values, and self-hosting docs.

• Provider abstraction: a MailProvider interface, one file per provider, and a factory-based activeProviders array with auto-detection in priority order (Resend → SES → SMTP → Azure). Each provider implements send and optionally sendBatch.
• Batch send refactor: sendBatchEmails now prepares all emails via prepareBatch (deduplicating unsubscribe checks), routes to a sendBatch-capable provider when available, and falls back to per-message sends on failure.
• New SMTP provider: authenticates when both SMTP_USER/SMTP_PASS are set, logs a warning for partial-credential misconfigurations, and auto-enables TLS on port 465.

Confidence Score: 5/5

Safe to merge — the new SES and SMTP providers are well-guarded at startup, the provider fallback chain works correctly, and the existing test suite passes without modification.

The refactor cleanly separates concerns with no breaking changes. The one finding is a timing-race edge case in the per-message batch path where data.count can over-report by the number of users who unsubscribe in the very small window between batch preparation and dispatch — it requires both a race condition and a non-batch provider to trigger, so it has negligible real-world impact.

apps/sim/lib/messaging/email/mailer.ts — specifically the per-message fallback in sendBatchEmails.

Important Files Changed

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[sendEmail / sendBatchEmails] --> B{shouldSkipForUnsubscribe?}
    B -- yes --> C[Return SKIPPED_UNSUBSCRIBED_RESULT]
    B -- no --> D[processEmailData]
    D --> E{activeProviders.length === 0?}
    E -- yes --> F[Return MOCK_EMAIL_RESULT - console log only]
    E -- no --> G[dispatchWithFallback]
    G --> H{Try Resend}
    H -- success --> I[Return result]
    H -- fail --> J{Try SES}
    J -- success --> I
    J -- fail --> K{Try SMTP}
    K -- success --> I
    K -- fail --> L{Try Azure}
    L -- success --> I
    L -- fail --> M[Return all-providers-failed error]

    subgraph sendBatchEmails
        N[prepareBatch - check unsubscribe per email] --> O{sendable.length === 0?}
        O -- yes --> P[Return skipped/failed result]
        O -- no --> Q{Provider with sendBatch?}
        Q -- yes Resend --> R[batchProvider.sendBatch]
        R -- success --> S[mergeBatchResults]
        R -- fail --> T[Per-message fallback via sendEmail]
        Q -- no --> T
        T --> S
    end

_{Reviews (4): Last reviewed commit: "fix(mailer): batch degrades isUnsubscrib..." | Re-trigger Greptile}

@greptile

@cursor review

@greptile

@cursor review

@greptile

@cursor review