Dangerously-skip-permission is carte blanche for the model to do whatever it pleases with your system. If you happen to have access to a production database on your system, then the model also has access to it, should you use that option.
I’ve worked at places where senior devs have access to prod for emergency fixes but usually the procedure is to use a VPN AND an ssh key with a passphrase. Usually.
Human developers should not develop with the production DB, why the hell would you give an AI the rights to touch the prod DB?
You don’t test in production?
Dangerously-skip-permission is carte blanche for the model to do whatever it pleases with your system. If you happen to have access to a production database on your system, then the model also has access to it, should you use that option.
that assumes
Possible? Sure. It’s also possible that I drink half a bottle of vodka on a Friday night and mess up with production.
Naturally.
You should still probably not use dangerously-skip-permissions, though.
Yes, but the question is why the hell do you have access to a production database in the first place?!
And if so, how is it on the same machine you can run Claude code on?!
I’ve worked at places where senior devs have access to prod for emergency fixes but usually the procedure is to use a VPN AND an ssh key with a passphrase. Usually.
I’ve worked at several places where I’ve been able to access production databases.
No need to be so dramatic about it, really.
Startups and small companies where there simply aren’t enough resources to set up proper operational controls
No no no I’m not running Dev ops on the production system. I’m running prod ops on the developer system.