SSH Key Handling in My Dev Setup
Recently, I came across @halvarflake’s blog post about his development setup. I’m roughly following the same idea: a dedicated machine for development. One difference is that I use a dedicated VM per project. This keeps each environment clean and, in case of compromise, the blast radius is limited to that one project. AI agents increase the risk in my setup. To get the most out of them, I usually give them access to the internet, tools (whatever’s available), and the code. On top of that, I often run them in YOLO mode (e.g. --dangerously-skip-permissions in claude code). ...