@fenrircodes here.
Recently I was using Cursor Agents for a full testing and implementation workflow and ran into an interesting limitation.
I’m based in Australia and was testing integrations with a digital asset exchange. Because Agent traffic was originating from US-based IPs, I ended up getting blocked by the exchange’s geo-restrictions. There was nothing unusual about the activity—I was simply trying to perform some testing from my own account.
My workaround was to move into the terminal and run everything locally. This solved the IP issue immediately, but it also took me away from the Agents window experience, which is honestly one of my favourite parts of Cursor.
It got me thinking: could there be a future option to run Agent execution locally while still maintaining the Agents interface and workflow? I completely understand there would be latency, resource, and architectural trade-offs, but having the choice could be valuable for users dealing with region-specific services, compliance requirements, private infrastructure, or IP-based restrictions.
Curious if others have run into similar issues and whether local Agent execution is something already being considered.