I'm not seeing a clean/easy way to deny all outbound traffic by default and allow only specifically configured endpoints through (via rate-limited rules).
If I add a Block rule with host-pattern * and manual approval, it matches all traffic and nothing gets through even if a "rate-limit" rule is also present that would allow the traffic. The URL match logic feels...backward.
Is "block by default" not possible? Can it be added? If this project is essentially a smart firewall, why is the default egress "allow anywhere"?
If an agent can HTTP-call some obscure website that opens up a reverse shell or runs a container jailbreak, doesn't that kind of defeat the point here..?
...also there appears to be no way to do manual approvals outside of the SDK. Onecli CLI doesn't seem to have an "approve pending request" function/param it and there's no pop-up or similar in the web UI to let a request through.
I'm not seeing a clean/easy way to deny all outbound traffic by default and allow only specifically configured endpoints through (via rate-limited rules).
If I add a Block rule with host-pattern
*and manual approval, it matches all traffic and nothing gets through even if a "rate-limit" rule is also present that would allow the traffic. The URL match logic feels...backward.Is "block by default" not possible? Can it be added? If this project is essentially a smart firewall, why is the default egress "allow anywhere"?
If an agent can HTTP-call some obscure website that opens up a reverse shell or runs a container jailbreak, doesn't that kind of defeat the point here..?
...also there appears to be no way to do manual approvals outside of the SDK. Onecli CLI doesn't seem to have an "approve pending request" function/param it and there's no pop-up or similar in the web UI to let a request through.