Send through a Mailgun domain without mixing credentials

Start with the official Mailgun client/library instructions in the sending quickstart. Select the endpoint for your actual domain region; do not reuse a US endpoint for an EU-domain example.

Wire one sending path

1. Install the supported client for your runtime and its documented dependencies.
2. Load the domain’s authorized sending credential from your secret store. An SMTP password is not an HTTP API key.
3. Configure the verified domain, approved From address, one controlled recipient, subject, and plain-text body.
4. Submit once and retain the returned message identity alongside a preexisting application send record.

Before execution, verify that the example actually uses your intended domain and not the provider’s restricted sandbox. Running a quickstart sends real mail; do not leave a reserved documentation address as the recipient.

Check delivery evidence

Inspect domain activity and the controlled inbox. Provider acceptance means the message entered processing, not that it reached the inbox. Connect delivery/bounce feedback and define suppression handling before sending important production notifications.

A 401/403 needs credential/permission review. A missing domain can be an endpoint/region problem. An ambiguous timeout needs reconciliation before another submission; do not silently switch to MailKite and risk two sends.

For a full application comparison see sending with Mailgun. MailKite’s SDK is an alternative application transport with independent domain and credential requirements.