Microsoft’s 4B Mage-VL-Cowork has me thinking about testing a small computer-use agent at home. This is a plan I’m considering, not something I’ve done.
What I’m thinking about
- Throwaway VM only: a fresh desktop VM with a snapshot I can roll back, no shared folders, and no clipboard sharing
- No real accounts: no saved logins, keys or personal files inside it, just dummy test data
- Locked-down network: outbound limited to what the test needs, and nothing reachable on the rest of my network
- Simple tasks first: rename files, fill a spreadsheet, change a setting, and watch every step
- Logs: record what it clicks and runs so I can check it afterward
Open questions
- Is a 4B model reliable enough for real chores, or just a demo?
- What’s the easiest kill switch: pause the VM, or cut its network?
Review before you install: read the release notes yourself, back up your config, and try new versions somewhere you can afford to break first.
Isolate it: if you try this, run it inside a disposable VM with no access to your real accounts, keys or files, and snapshot before you start.
My take
An agent that can click anything should start in a box it can’t break out of. Ideas welcome.
God Bless, and Use the Source!
