Post

...so it's a quick connect endpoint.

Oh fuck no!

Keeping the password away from the agent is a good step. The next question is what the agent does after it logs in. A session can still pay, delete or send. Those actions need their own checks, outside the model, with a record of what was allowed and why.

How about no? Doesn’t that mean they don’t use end-to-end encryption for the password data?

Isn't this just an API call w/ a cert attached?

Fuck that... and then fuck it some more

I think the computer people (and I'm not one of them) call this an exploit! Correct me if I'm wrong.

Things are moving fast in this space right now.

So glad I ditched them a year ago!

Absolutely not

Fuck no. Will never use that platform.

Goddamn it is there any product on the planet where we don't have to deal with this fucking shit

Finally, some good news.

I'd give my pers'l (100% local and selftuned) AI acc. to my SM accounts if I had an) but only that. I love that Bluesky allows AI accounts (with transparency). I have been thinking of letting it on here to engage and learn more from other leftists/libs. MAGAts use AI to spread lies, why not counter?

Okay Why would I ever want that?

What could possibly go wrong?

Sounds like a secure feature with no possible flaws at all.

nope

typo, you mean computer viruses

Delegated auth for AI agents is the right move — giving agents credentials directly was always the weakest link in agentic workflows.

You should be bashing this, not reporting it. Obviously this is a terrible idea.