mailkube
The official Python SDK for Mailkube — send transactional email and verify inbound webhooks. Fully typed, sync and async, Python 3.12+.
Full product and API documentation: mailkube.com/docs.
Install
pip install mailkube
# or
uv add mailkube
Configuration
from mailkube import Mailkube
client = Mailkube(
api_key="mk_...", # or set MAILKUBE_API_KEY
base_url="https://api.mailkube.com/mta/v1/", # or set MAILKUBE_BASE_URL; this is the default
timeout=30.0, # per-request timeout in seconds
)
AsyncMailkube takes the same arguments. Both also accept http_client= — your own
httpx.Client / httpx.AsyncClient — if you need custom transport, proxies, or mTLS; the SDK
will not close a client you pass in.
Send an email
from mailkube import Mailkube
with Mailkube() as client:
email = client.emails.send(
from_="Acme <hello@yourdomain.com>",
to="customer@example.com",
cc="manager@example.com",
reply_to="support@yourdomain.com",
subject="Hello world",
html="<p>It works!</p>",
)
print(email.id, email.message_id)
from_ maps to the wire from field (from is a reserved keyword). Supply html and/or text, or
a template_id with template_version and variables. Attachments accept raw bytes or a base64
string. See SendEmailParams for the full field list (cc, bcc,
reply_to, headers, attachments, tags, template_id, topic, idempotency_key, …).
Async
import asyncio
from mailkube import AsyncMailkube
async def main():
async with AsyncMailkube() as client:
email = await client.emails.send(
from_="Acme <hello@yourdomain.com>",
to="customer@example.com",
subject="Hello world",
html="<p>It works!</p>",
)
print(email.id)
asyncio.run(main())
Idempotency
Pass idempotency_key to safely retry a send without risking a duplicate — sent as the
Idempotency-Key header, not in the body:
email = client.emails.send(
from_="Acme <hello@yourdomain.com>",
to="customer@example.com",
subject="Your receipt",
html="<p>Thanks for your order.</p>",
idempotency_key="order-4821-receipt",
)
print(email.idempotent_replayed) # True if this key was already used
Reusing a key with a different payload raises ConflictError instead of silently sending the new
content.
Errors
Every failure raises a subclass of MailkubeError (AuthenticationError, InvalidRequestError,
RateLimitError — which carries .retry_after — ServerError, MailkubeConnectionError, …). Each
API error exposes .error_name, .message, .status_code, and .request_id (quote it to support).
The HTTP status picks the exception class; .error_name says precisely what went wrong. Compare it
against the ErrorName constants — it stays a plain string, so an error name newer than your
installed SDK is still reported verbatim rather than crashing:
from mailkube import ErrorName, InvalidRequestError
try:
client.emails.send(...)
except InvalidRequestError as exc:
if exc.error_name == ErrorName.QUOTA_EXCEEDED:
...
Threading
Echo a message's message_id in the In-Reply-To / References headers of a later send:
reply = client.emails.send(
from_="Acme <hello@yourdomain.com>",
to="customer@example.com",
subject="Re: Your order",
html="<p>An update.</p>",
headers={"In-Reply-To": first.message_id, "References": first.message_id},
)
Tags
Attach free-form {"name": ..., "value": ...} pairs to a send. The server denormalizes them
onto the sending-log, so you can filter, export, and dashboard by tag, and they ride along on
delivery webhooks:
email = client.emails.send(
from_="Acme <hello@yourdomain.com>",
to="customer@example.com",
subject="Welcome aboard",
html="<p>Glad you're here.</p>",
tags=[{"name": "campaign", "value": "spring24"}, {"name": "plan", "value": "pro"}],
)
Validation is server-side: names and values allow the [A-Za-z0-9_-] charset, a name is at most
16 characters and a value at most 32, values may be blank, at most 20 tags per send, and names must
be unique. Tag
values are not encrypted, so keep personal data out of them.
Schedule an email
Pass scheduled_at and the message is accepted now and delivered later. Give it either an
ISO-8601 string with a timezone offset or a timezone-aware datetime; it must be in the
future and within your plan's scheduling horizon (30 days by default):
from datetime import datetime, timedelta, UTC
email = client.emails.send(
from_="Acme <hello@yourdomain.com>",
to="customer@example.com",
subject="Your weekly digest",
html="<p>Here's what happened.</p>",
scheduled_at=datetime.now(UTC) + timedelta(hours=2),
batch_id="digest-2026-08", # optional: group sends so you can move or cancel them together
)
email.is_scheduled # True
email.status # "scheduled"
email.scheduled_at # "2026-08-20T07:00:00Z"
email.id # use this to retrieve, reschedule, or cancel it
An immediate send is unaffected: is_scheduled is False and status / scheduled_at /
batch_id stay None. batch_id is only valid alongside scheduled_at.
Manage scheduled emails
Until it is due, a scheduled email lives in client.scheduled_emails:
email = client.scheduled_emails.get(email_id)
email = client.scheduled_emails.update(email_id, scheduled_at="2026-08-21T07:00:00Z")
client.scheduled_emails.cancel(email_id)
Listing
list returns one page; iter_all walks every page lazily, following the links the API
returns:
page = client.scheduled_emails.list(status="scheduled", batch_id="digest-2026-08")
page.data # list[ScheduledEmail]
page.pagination.total_count
page.has_more
for email in client.scheduled_emails.iter_all(status=["scheduled", "failed"]):
print(email.id, email.scheduled_at, email.subject)
| Filter | Accepts |
|---|---|
status |
"scheduled", "canceled", "failed" — one, or a list. A sent email has left the collection, so "sent" is a validation error, not an empty result. |
batch_id |
The batch label used at send time. |
scheduled_at_gte / scheduled_at_lte |
ISO-8601 with an offset, or an aware datetime. |
page |
1-based page number. |
Timestamps come back as the verbatim ISO-8601 strings the API sent — call
datetime.fromisoformat if you want objects.
Batches
Everything sent under one batch_id moves or cancels together:
result = client.scheduled_emails.batches.update("digest-2026-08", scheduled_at="2026-08-21T07:00:00Z")
result.rescheduled_count # 2
result = client.scheduled_emails.batches.cancel("digest-2026-08")
result.canceled_count # 2
An unknown batch is a no-op reporting 0, not an error.
AsyncMailkube exposes the identical surface — await client.scheduled_emails.get(...), and
async for email in client.scheduled_emails.iter_all(...).
Scheduling errors
The names specific to this surface:
from mailkube import ErrorName, InvalidRequestError, NotFoundError
try:
client.scheduled_emails.cancel(email_id)
except NotFoundError:
... # scheduled_email_not_found
except InvalidRequestError as exc:
if exc.error_name == ErrorName.SCHEDULED_EMAIL_NOT_PENDING:
... # already sent or canceled
Every API error also carries .request_id — quote it when contacting support.
Verify webhooks
webhooks.verify is a pure, stdlib-only helper — call it in your request handler with the raw body:
from mailkube import verify, SignatureVerificationError, UnknownEvent
try:
event = verify(raw_body, request.headers, signing_secret)
except SignatureVerificationError:
... # reject with 400
if isinstance(event, UnknownEvent):
... # a newer event type than this SDK version knows about — still usable
else:
print(event.type)
An unrecognized event type is returned as UnknownEvent instead of raising, so a new server event
type never forces an SDK upgrade on receivers.
Logging
Silent by default. Turn on request/response logging with:
import mailkube
mailkube.enable_logging(level="DEBUG")
or set the MAILKUBE_LOG environment variable. Authorization and Idempotency-Key headers are
redacted from log output.
Client lifecycle
Create one client and reuse it. Mailkube is thread-safe; AsyncMailkube is bound to its event
loop. Use the client as a (async) context manager, or call .close() / .aclose().
More examples
Runnable scripts in examples/:
simple_send.py— basic sync sendasync_send.py— async send, then thread a reply onto itsend_with_attachments.py— attach a file from raw bytessend_with_tags.py— tag a send for filtering and reportingsend_with_template.py— send from a saved templateschedule_send.py— schedule a send, then inspect the ackmanage_scheduled_emails.py— list, paginate, retrieve, reschedule and cancelschedule_batch.py— schedule a batch, then move or cancel it as a unitwebhook_receiver_flask.py— verify and dispatch webhooks in a Flask app
Contributing
See CONTRIBUTING.md for the development setup and the quality gates every change must pass. Security issues: see SECURITY.md.
License
Apache-2.0 © 2026 Mailtactic, Corp.
Release files for mailkube 1.3.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| mailkube-1.3.0.tar.gz | 90.1 kB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| mailkube-1.3.0-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 129.2 kB
Release files / mailkube-1.3.0.tar.gz
| Download URL | mailkube-1.3.0.tar.gz |
|---|---|
| Size | 90.1 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
34a0086d87f1374af8349570733f88607d310552c0fb7667b49cbb95734e1a0c
|
|
BLAKE2b-256 checksum How to use checksums |
2b0fbd60e3fa11e6feff833fb27d4d96ddb3b66b51cf432752077e416ef42393
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Aug 10, 2026.
Transparency logRelease files / mailkube-1.3.0-py3-none-any.whl
| Download URL | mailkube-1.3.0-py3-none-any.whl |
|---|---|
| Size | 39.2 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
9b360290f0d24b2c3c9783a57120dd9bfabf0cfe132549083486e7cf9902c060
|
|
BLAKE2b-256 checksum How to use checksums |
ddba22ab31c893bc399c279518e2e6dc155f24a7a1cfcb2209ee7fa62335a922
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Aug 10, 2026.
Transparency log