News for hackers. Stuff that matters.
Re:Protecting SSH Keys with macOS Secure Enclave by ajross on Wednesday November 11, 2026 @10:48AM (#50454800)
The details of using the SE are interesting. I still don't understand the threat analysis here. Basically here's the lede:

> Any malicious script that gets access to my filesystem could just copy them and get into all my servers.

So this is postulating a "malicious script" that has access to your filesystem but not the running system. Because if it's running on the actual system it doesn't need the key, it can just execute your remote stuff on its own. Or it can get the key by MitMing you by making your next "ssh" invokation steal your passphrase. Or...

Like, there's near zero added security here. The only scenario I can think of that would have any meaning is someone getting access to an offline image of your filesystem, an attack already prevented by drive encryption (and, perhaps, already defeated by insecure backups!).

You see occasional attempts to use TPMs for similar purposes, and they're similarly silly. People need to understand that the TPM/SE exists for the benefit of third party authentication. It's not for you, the operator. It's there so that your bank/employer/intelligence-agency/whatever can know that the human being doing authentication is using the same device they've been known to use before.

Re:Protecting SSH Keys with macOS Secure Enclave by what on Wednesday November 11, 2026 @01:31PM (#50456770)
> it can just execute your remote stuff on its own

Presumably biometrics are required to use the key, so they have to convince you to prove presence when you aren’t expecting it?

> Or it can get the key by MitMing you by making your next "ssh" invocation steal your passphrase

What passphrase?

> or just read the unencrypted key out of the ssh process

The key never leaves the SE, so this is not possible.

> or even just reimplement ssh to do the SE negotiation on its own

See first point and biometrics.

Re:Protecting SSH Keys with macOS Secure Enclave by ajross on Wednesday November 11, 2026 @01:34PM (#50456819)
So, rather than argue the minutiae, let me rephrase: an article explaining a SE-based SSH key storage paradigm that was provably resistant to an attacker able to run code locally as the user would be an extremely interesting blog post to read.

This one is not that, and doesn't even seem aware of the complexities. Have a link?