The Solva Group · Mailroom

How Mailroom is built to be safe.

Mailroom reads your mail, so the bar has to be high. This page states plainly how it is protected, including the parts that are limitations rather than strengths. For what we collect and who receives it, see privacy.

Your mailbox credential

You sign in on Google's or Microsoft's own page. We never see your password and could not store it if we wanted to. What we hold is a revocable access credential, encrypted at rest with AES-256-GCM using a fresh initialisation value every time. If the stored value is altered by so much as one byte, decryption fails and is refused rather than guessed at. Only the long-lived credential is stored; the short-lived one used for each request stays in memory and is never written to disk.

Every stored file is readable only by the account that runs the service, inside a directory closed to every other account on the machine. The application does not run with administrator rights.

Your connection

Everything travels over TLS 1.2 or 1.3; older, weaker protocols are switched off. The site is served under HSTS, so browsers refuse to connect insecurely even if asked, and plain HTTP is redirected. The application listens only on the machine's internal loopback interface, so it cannot be reached directly from the internet, only through the hardened front door. Sign-in callbacks are excluded from access logs, so the codes exchanged during sign-in are never written down.

Keeping accounts separate

Each account's data lives in its own directory, and which directory to open is decided only by a cryptographically signed session cookie, never by anything the browser can ask for. There is no way to request another account's data, because the request has no way to name one. Signatures are checked in constant time, and signing out invalidates every existing session for that account immediately, on every device.

There is no administrator view of your mailbox. Nobody at Solva has a screen that shows your mail.

What leaves the server

To tell you which emails need a reply, message content is sent to Anthropic's Claude API and the result comes straight back to your screen. That is the only third party that receives message content, it is used solely to produce what you see in the app, the data is not used to train any model, and it is retained only briefly before deletion. Nothing goes to advertisers, data brokers or resellers: we do not sell or share your data in any form.

What we log

Operational logs deliberately contain no message content, no subjects, no email addresses and no credentials. Errors record a short reason and an anonymous identifier, which is enough to fix a fault and not enough to read your mail.

Built small on purpose

Most of a modern application's risk arrives through its dependencies. Mailroom has one third-party code library; everything else is the platform's own standard toolkit. There is far less here to be vulnerable, and far less to keep patched.

Being straight about the limits

A security page that claims everything is worth nothing, so here is what Mailroom does not do:

Reporting a problem

If you believe you have found a security issue, email mark@thesolvagroup.com with the detail. We will confirm receipt, investigate, and tell you what we found. We will not take legal action against anyone who reports a genuine issue in good faith and gives us a reasonable chance to fix it before going public.

The Solva Group, Victoria, Australia. Last updated 28 July 2026.