Is http://git-secret.io/ a safe method to store credentials in public Git repositories?
Conversation
Notices
-
FlowFX (flowfx@chaos.social)'s status on Sunday, 08-Jul-2018 19:42:03 CEST FlowFX -
MacLemon (maclemon@chaos.social)'s status on Sunday, 08-Jul-2018 19:57:36 CEST MacLemon @flowfx Conveniently disregards the fact that you must change all passwords after changing access. Since the credentials stored in git include all their history and you could just go back to a commit where you still had access.
Now you have a password management problem, a git problem, a key distribution problem without meaningful revocation mechanism, and a GPG problem, including a denial of service attack vector for pull requests by changing passwords though a commit.Just keep it separate.
FlowFX repeated this. -
FlowFX (flowfx@chaos.social)'s status on Sunday, 08-Jul-2018 20:03:35 CEST FlowFX @MacLemon Thank you.
-
Henryk Plötz (henryk@chaos.social)'s status on Monday, 09-Jul-2018 10:56:01 CEST Henryk Plötz @MacLemon @flowfx To be fair, though, that's a general point on *all* password sharing mechanismus. (See also: https://www.passwordstore.org/)
Inherently, it's impossible to rescind knowledge.
A dishonest authorized user can always just make a copy/memorize everything. Git just makes this slightly easier by automatically providing old copies.
FlowFX repeated this. -
MacLemon (maclemon@chaos.social)'s status on Monday, 09-Jul-2018 11:47:28 CEST MacLemon @henryk I totally agree. Having a public ledger of all the password history, even if encrypted, puts up a great level of attack surface, IMHO. I mean, how good do you expect the passwords that protect that thing to be statistically?
I can just clone the repo and start attacking it offline with as many CPUs I'm willing to afford.
Keeping the same info in an offline repo allows for versioning likewise and keeps the attack surface lower. I wouldn't risk it.
@flowfxFlowFX repeated this.
-