A lot of people write so much software to sandbox their AI agents. I say don't bother. Just give your agent its own non-admin local user account. It's easier and much more effective.
@stagas68 notes287 replies26 tags, 197 users following72 followers |\__/,| (`\
_.|o o |_ ) )
-(((---(((--------enter to followenter to reply
I give my AI Full-Access. #yolo1 note0 followersenter to follow :p
@swiley5 notes24 replies0 tags, 3 users following3 followersswiley.net
Try my zero dependency agent harness: public.swiley.net/agent.py
If you're spending too much time on sites like these install bingeblock.comenter to followenter to reply
My whole machine is set up to make way too much easily accessible with shell tools (email, stock trading, payments, mms group chats.) I can't trust it.
@fastidious19 notes97 replies1 tag, 3 users following4 followersBetter don’t ask, just fastidious.enter to followenter to reply
That's why one leases a VPS, or runs isolated VMs, so that you can trust it. ;-)
@stagas68 notes287 replies26 tags, 197 users following72 followers |\__/,| (`\
_.|o o |_ ) )
-(((---(((--------enter to followenter to reply
*hacks into @swiley5 notes24 replies0 tags, 3 users following3 followersswiley.net Try my zero dependency agent harness: public.swiley.net/agent.py If you're spending too much time on sites like these install bingeblock.comenter to follow's machine, pulls all funds* :p
So it has full access to textlog!? ;)
@stagas68 notes287 replies26 tags, 197 users following72 followers |\__/,| (`\
_.|o o |_ ) )
-(((---(((--------enter to followenter to reply
Locally only, remotely it would need to know my ssh login password, which I think it doesn't ;P