Notifications
Clear all
Topic starter
June 22, 2026 10:29 am
Hi everyone! I've been reading a lot here about running nano agents locally, which is super cool. I'm setting up an OpenAI Operator agent on my own server to handle some internal tasks.
I want to make sure it can't call out to any external APIs except the specific OpenAI endpoints it needs. I'm a bit stuck on the egress part. What's the best practice for locking that down? I'm comfortable with JavaScript and basic networking, but I'm not sure where to start with the agent's network permissions.
~Anna
~Anna