The AI Agent Safety Stack That Costs $10 Total (No VM Required)

Everyone's building elaborate safety stacks for AI agents:

• Permission systems
• Approval gates
• Policy engines
• Sandbox frameworks
• Audit logs

And they're all missing the simplest, most effective safety measure of all.

The $10 stack

1. Create a folder for each AI task
2. Map that folder to a drive letter on Windows
3. Point the agent at that drive

Total cost: $9.99 for ScsSubst (https://subst.scsoi.com/), one-time purchase.

That's it. No subscriptions. No complex setup. No learning curve.

Why this works better than fancy systems

• It's enforced by the OS, not by the AI tool or the model
• It works with every AI tool — Cursor, Claude Code, Cline, Aider, whatever
• It's invisible to the agent — it just sees a drive, doesn't know it's mapped
• It's free to reset — delete the drive, start over, no cleanup needed

What you don't need

• You don't need a VM for every task
• You don't need Docker containers
• You don't need a policy engine that intercepts tool calls
• You don't need to spend 2 hours setting up a sandbox framework

You just need a drive letter.

The workflow

C:i-workspaces\client-a\  →  Z:
C:i-workspaces\client-b\  →  Y:
C:i-workspaces\experiment\ → X:

Each client gets their own drive. Each experiment gets its own drive. The agent can't accidentally mix files from client A into client B's project.

The point

AI safety doesn't have to be complicated. The best safety measure is the one you'll actually use.

A mapped drive letter takes 10 seconds to set up, costs $10 one time, and gives you a real OS-level boundary. That's the whole stack.