Is it safe to use Agentic AI? To be honest, it depends on how you've set it up, whether you have restricted the agent's ability to code on your computer & which files it sees. In today's video, I show why you should restrict Agentic AI access, how to use Docker to "sandbox" your agent & how to control which files your agent sees using Hermes ✨
https://www.youtube.com/watch?v=OMQo6YMZYzo https://www.patreon.com/cw/wanderlootsDocker desktop is an easy way for anyone to create a "container" for running code. A container is an isolated sandbox that separates your AI agent from the rest of your computer. This separation is especially important for running the "execution backend", or the terminal aka coding environment of your agentic AI, as this can reduce the risk of malicious behaviour and sensitive data leaking from your computer.
But the best part is that you can then easily "mount" specific volumes (aka folders or files) to your docker container to control the level of access your agent has to your data. That way, you're always in control of what you want your agent to see and how you want it to use that information.
Hermes Agent, by Nous Research (
https://hermes-agent.nousresearch.com/), is a self-evolving AI Agent harness. That means it has skills, tools, and memories that enable the agent to tap into past conversations, codify style & preferences, & build custom workflows that improve every time you use them.
Hermes has persistent memory, which means it extracts key information and remembers it over time, building a profile of you & how you work. These memories are used to build skills that improve over time as you repeat your workflows. Over time, you'll need to ask Hermes to do less, while still achieving more (without repeating yourself).
Hermes is free & open source, so you can be confident that it is an agent you can use for the long haul. You can easily connect to local models like Gemma4 or Qwen through Ollama, so you can run your agent privately & freely. The new desktop app makes it easier than ever to get started.
Whether you're new to agentic AI or are confident leveraging agent harnesses, this tutorial walks through everything you need to know to get Hermes working for you.
I also touch on why the skills & architecture of Hermes make Obsidian & the LLM Wiki a natural companion of this incredible agentic AI tool. More on LLM Wiki here:
https://youtu.be/n4EVksU_EOsI hope you enjoy! ✨
P.S. I greatly appreciate any feedback, please let me know what you think 😊
Join My Membership:
Timestamps:
00:00 What Can Go Wrong With Agentic AI
00:27 Intro To Docker: Reducing Agentic AI Risk
01:41 Brief Demo: Why Containers Matter
03:46 Installing & Setting Up Docker Desktop
04:28 Making Docker Your Terminal Backend (Hermes Example)
05:57 Testing Docker Container Isolation
07:22 Container Volume Mounting Explained
08:41 Demo: Volume Mount & Container File Access
10:34 Example: Obsidian Vaults
12:06 Understanding Docker Limitations & Risks
13:52 Advanced: Containers & Gateways
15:15 Next Steps For Using Hermes Agent
LINKS (MY WORLD)
SOCIALS
MY FAVOURITE TOOLS
EQUIPMENT USED