Please report security issues privately — do not open a public issue for a suspected vulnerability.

• Preferred: use GitHub's private vulnerability reporting for this repository (the Security tab → Report a vulnerability).
• You will get an acknowledgement, and a fix or mitigation plan once the report is triaged.

Please include enough detail to reproduce: affected version, configuration shape (sanitized), and the observed vs. expected behavior.

RestoreGuard is a read-only auditor. It is designed to never write state to an audited host, and it stores no credentials of its own — it rides on the SSH setup already present on the operator machine (ssh -o BatchMode=yes against your ~/.ssh/config aliases and keys).

Because of that, the security-relevant surface is mostly:

• Command construction — probe commands sent over SSH. Reports of a path that could inject or mutate a target are in scope and treated seriously.
• Report / config handling — parsing untrusted remote command output and the local config/suppressions files.
• Secret handling — RestoreGuard reads references to repo passwords / passphrase files on the audited hosts; it must not log or emit their contents.

Out of scope: the security of the SSH setup, keys, and host access you provide — that's yours to manage.

RestoreGuard is pre-1.0 and ships from main. Security fixes target the latest released version and main.