---
title: "Security"
canonical: "https://machinemail.dev/security"
---

# Security

Effective November 17, 2026.

MachineMail holds mail on behalf of your agents, so we treat every inbox as sensitive. This page says how we protect it, which independent reports cover the infrastructure we run on, and how to report a vulnerability. What we collect and how long we keep it is in our [Privacy Policy](/privacy), and who processes it for us is on [Subprocessors](/subprocessors).

## Encryption

- **In transit:** all traffic to machinemail.dev, the API and the MCP server uses HTTPS. Plain HTTP is redirected. Our calls to providers use TLS.
- **At rest:** our servers' disks at Fly.io, our Fly Managed Postgres database and our Tigris object storage, where raw messages and attachments live, are encrypted at rest.

## Credentials and sign-in

- **API keys:** stored only as hashes, never in plain text. A key can be limited to one pod or one inbox and to the permissions it needs, and a limited key can't create a key with more access than its own.
- **Dashboard sign-in:** a one-time code by email, a social account, your organization's single sign-on (SSO), or a passkey. With a passkey you can turn email codes off entirely.
- **Sensitive changes** need a recent sign-in, confirmed with a passkey or an emailed code.
- **Sessions:** cookies are `HttpOnly`, `Secure` and `SameSite=Lax`.
- **Abuse:** sign-in, sign-up and the API are rate limited.

## Access in your organization

- **Roles:** owners, admins and members. Only owners and admins manage people, and admins can't change or remove owners.
- **Offboarding:** removing someone ends their access at once. The API keys they created are revoked, the apps they connected are disconnected and the invitations they sent are revoked.
- **Separation between customers:** every request is limited to the organization, pod or inbox its credential belongs to. Our tests check that one organization can't read another's threads, messages or events.

## Protecting your agents

- **Trust report:** every inbound message is checked for SPF, DKIM and DMARC, and scored for phishing and prompt injection.
- **Quarantine:** mail judged dangerous is labelled `QUARANTINE` and withheld from your agents until a person in your organization reviews it.
- **Agent-safe reads:** a key you give a model can be limited to the agent-safe view, which strips quoted history, signatures and text hidden from human readers.
- **Webhooks:** each delivery is signed so you can check it came from us. Webhook and attachment URLs can't point at private networks.

## Infrastructure

- **Hosting:** our servers and database run on Fly.io in Ashburn, Virginia (region `iad`). Email is sent and received through Amazon SES. Raw messages and attachments are stored in Tigris.
- **Independent reports:** Fly.io, Amazon Web Services, Tigris and Cloudflare each hold a SOC 2 Type II report covering the infrastructure we run on.
- **Backups:** database backups are kept for 10 days and server disk snapshots for 5 days.

## How we build and ship

- **Review before merge:** every change goes through a pull request, and automated type checks, tests and lint run before it can merge.
- **Staging first:** every release runs first in a separate staging environment with its own database, keys and mail domain. Only after it passes a smoke test there is the same image released to production.
- **Security review:** we reviewed the API's security in October 2026. Every finding was fixed or mitigated, and each fix has a regression test.

## Staff access

- **Admin tool:** our internal admin tool is not reachable from the public internet. It shows account metadata only, never subjects, bodies or recipients, and every change and search made in it is recorded in an append-only audit log.
- **Production database:** only MachineMail's operator can access it directly. The service itself and the admin tool connect to it to run the service, and only to run, secure and support MachineMail.

## Compliance

- **SOC 2:** MachineMail does not have its own SOC 2 report yet. The providers we run on do (see Infrastructure).
- **Security questionnaires:** we're happy to fill in yours. Email [support@machinemail.dev](mailto:support@machinemail.dev).
- **Data processing agreement:** available on request at the same address.

## Report a vulnerability

If you think you've found a security problem in MachineMail, email [support@machinemail.dev](mailto:support@machinemail.dev) with "Security" in the subject. Include what you found, how to reproduce it and what it could affect. We'll reply, keep you posted while we fix it, and tell you when it's done.

While testing, use only accounts and inboxes you own, don't access or change other customers' data, and don't run denial-of-service or spam tests.

## Incidents

If we learn of a breach that affects your data, we will tell you without undue delay and as the law requires.
