We need vpn capabilities, one to securely route traffic but also to pick different endpoints globally. Sometimes the endpoints in NA are overloaded or we're hit with higher fees for using LLM endpoints coming from a certain geographical area. I use nord and proton for routing my normal desktop browser traffic as needed. It would be nice to have this in OH. You already have a built in browswer to the subsystem is there...maybe a simple plugin engine would work. Or, just create a vpn section in the settings and let the user setup tailscale, wireguard etc. Also, have the ability to create several configs so a user can easy switch from one vpn service or endpoint to another in the UI like choosing a llm.

Hi @clean-postman-32056 thanks for the feedback. It’s possible to connect to different backends in different areas by selecting the backend. Is that what you’re looking for or something else?

Hello Graham, I wasn't looking for another backend or fee/system to manage. I just want to route the traffic from my local OH install though a vpn if needed instead of just out my default home router. A: to secure traffic as it heads from OH to the LLM. B: So I can change my IP and geo location. For example, if I'm getting throttled or rate limited coming from a North American IP block I can choose a VPN in frace or japan or wherever so that my traffic going to the llm provider now has that IP address. With our free speech and online activities under assault being able to secure our traffic from being monitored in flight would be a nice comfort feature. Being able to route from another geolocation and ip block is also part of being secure these days but it also functions as a way to avoid being throttled or avoiding higher fees for peek time hours. Even deep seek is doing this now, what's the point of paying for a llm if they throttle me because I'm sitting in North America?

Interesting, this is an interesting use case. But won’t most VPN software get you this for free? Like once you activate the VPN won’t it route all traffic for you, including any traffic from the agent canvas?

@clean-postman-32056, building on Graham’s questions, @hundreds-twilight-32833 and I would like to better understand how you’re running OpenHands locally, when you need to switch geographies, and where the current workflow breaks down. Would you be open to a 20-minute call to walk us through the use case and share more detailed product feedback?

I don't have time today for a call. But, the concept is really simple. FYI, I'm running your latest openhands 1.8.0 via docker on linux locally..as in my lab. I connect to the front end from my desktop computer where I use the UI. So, the backend and front end and my desktop are all local. I have a few llm configured like deep seek and opencode zen models that make api calls to the remote endpoints. Sometimes those remote endpoints will throttle or up charge you during peek times or outright pause your usage...as in a timeout. Look at deep seeks new usage policy they are implementing. Users can be charged double the token fee if using the system during peek times and peek times is defined by your region. So, if I'm in Asia my peek times would be different than in Canada. To get around this absurdity I could route my traffic through a vpn so my src ip changes to where the vpn server is physically located. To the provider it looks like I'm in Asia. That was my idea.