Security

One connection.
Yours to verify.

FORTIQ is built so that its security story fits in a sentence: the client talks SSH to the one server you picked, and nothing else.

Network

SSH only, by design

No analytics, no telemetry, no hidden services.

  • The only connection is SSH

    The client has no direct connection to mail providers, web servers, databases or update services. Applications run their tools on your server and receive the results over SSH.

  • No update checks

    The client does not look for updates on its own, because that would be a second connection. You download new versions yourself.

  • Two servers never mix

    Every result is tied to the server and connection that produced it. A late answer from a previous host is discarded.

Trust

Know who you are talking to

Trust is established before a secret is ever sent.

  • Fingerprint pinned first

    The SHA-256 fingerprint of the server key is shown and pinned before authentication. If the key later changes, sign-in is blocked until you decide explicitly.

  • Obsolete algorithms refused

    Weak or outdated key exchange, ciphers and host key types are rejected rather than silently accepted. A test covers this.

  • Password or key

    Sign in with a password or a private key. Sessions are reused so you are not asked again and again.

Secrets

Handled with care

Where your credentials live, and where they never go.

  • Operating system store

    A remembered password lives only in the system secure store: Windows Credential Manager or Android Keystore. If it is unavailable, FORTIQ refuses to save. It never falls back to a file.

  • Not in arguments or logs

    Secrets are not passed on command lines and are not written to logs. Tests check that no copy remains in memory after use.

  • Known limits

    ssh-agent is not supported yet. The Android Keystore path has been verified only by a self-test on an emulator.

On the server

Careful with what it touches

Safer defaults, and honesty about their limits.

  • Read-only where it matters

    System and Databases only read. Database protection is layered, but it is a safeguard, not a security boundary: use a read-only role for sensitive data.

  • Browser guard rails

    Internal server addresses open only after your confirmation, and cloud metadata addresses never open.

  • Mail passwords live on the server

    Mail credentials are kept in the server's own configuration, protected by file permissions (0600), but stored in clear text there.

What we do not claim

These properties are backed by the project's own automated tests on development builds. They have not yet been re-run on packaged release binaries, and no release has been security-qualified.

Questions about how it works?

Read the FAQ