2 posts tagged with "server access"

Remote development used to mean a chain of compromises: carry a sufficiently powerful laptop, connect through a sometimes unreliable VPN, then open separate tools for SSH, remote desktop, and development. The server may have more CPU, more memory, and a more stable runtime, yet the working environment remains tied to the device in front of us.

I recently inverted that model: make the Linux server the real development machine, and use the Larktun app as the portable way in.

Once Larktun is installed on the server, I can SSH from a phone or tablet and hand coding, troubleshooting, and environment setup to an agent such as Hermes. When the agent builds a web page, I open the development service directly in Larktun's built-in WebView to inspect the real result. When I need an IDE or another desktop application, I enter the same Linux machine through RDP. I can even ask Hermes to prepare the available ChatGPT/Codex desktop package before switching to the graphical session.

The mobile device no longer does all the computation. Its job is to bring me safely back to the same persistent development workspace.

I recently provisioned a free Oracle Cloud server with 4 CPU cores and 24 GB of memory. For a development server, that is generous enough for test services, build jobs, database experiments, and the occasional production-support workload.

But as soon as I needed remote SSH access, the classic question appeared: should I expose port 22 to the public internet?

My answer is clear: no.