Cole Medin

How to Actually Run Your Coding Agent Safely (And Avoid the Horror Stories)

Every coding agent ships a mode that skips the approval prompts, and we all turn it on. The problem is what stands between that agent and your machine, which is nothing. It has your filesystem, your database, your SSH keys. It doesn't take malice or a hack for that to go wrong, just one bad decision, and it has already happened to plenty of builders. Rules aren't enough to keep your agent from going off the rails, and the sandboxes built into the agents are a soft boundary (Anthropic says so themselves). What works is running the agent somewhere it can't reach anything that matters, which is what Docker Sandboxes does: its own microVM right on your machine. Free, local, works with any coding agent. This video is a partnership with Docker but this is genuinely a platform I use myself. Docker has always been a big part of my stack! ~~~~~~~~~~~~~~~~~~~~~~~~~~ - Docker Sandboxes (free, by Docker): https://www.docker.com/products/docker-sandboxes/ - Join my free live workshop on July 29th, where I show you how to become an AI native engineering organization - building a reliable standard for how teams use AI coding agents: https://dynamous.ai/ai-native-engineering-org ~~~~~~~~~~~~~~~~~~~~~~~~~~ 0:00 YOLO mode and why you already use it 1:53 What your agent can do to your machine 4:00 How fast Claude Code folds on risky commands 8:14 Real disasters from Claude Code 9:52 Why Docker Sandboxes 10:37 Install it and run your agent in one command 11:40 Docker Sandboxes in action 12:27 Network security 13:09 Asking Claude to break out of its own sandbox 13:56 Direct mount vs clone mode 14:48 The four isolation layers 16:54 Wrap up ~~~~~~~~~~~~~~~~~~~~~~~~~~ Join me as I push the limits of what is possible with AI. I'll be uploading videos weekly - at least every Wednesday at 7:00 PM CDT!