AI agents invoke create_sandbox to trigger actions in Virtualization. What it does depends on the arguments the agent supplies, and its effects often reach beyond the immediate call — builds kicked off, notifications sent, workflows started.
Creating a sandbox spins up an isolated virtual environment, which is an Execute-level action — it triggers external operations (VM/container instantiation) with real resource implications. The description is empty, so confidence is reduced, but context from server description and sibling tools strongly implies this creates a new sandbox environment.
From the tool's definition Tool name 'create_sandbox' on a server that explicitly supports 'Windows Sandbox' creation and VM lifecycle management. The sibling tool 'create_windows_sandbox' confirms sandbox creation is a core capability.
Documented attack patterns abuse exactly the kind of access create_sandbox gives an agent:
PolicyLayer is an MCP gateway — it sits between your AI agents and Virtualization, and nothing reaches the server without passing your rules. This is the rule we recommend for create_sandbox:
{
"version": "1",
"default": "deny",
"tools": {
"create_sandbox": {
"limits": [
{
"counter": "create_sandbox_rate",
"window": "minute",
"max": 10,
"scope": "grant"
}
]
}
}
}create_sandbox stays usable, but rate-capped — a runaway agent can't fire it dozens of times a minute. Everything else on the server is denied unless you say otherwise.
Free to start. No card required.
create_sandbox. It is categorised as a Execute tool in the Virtualization MCP Server, which means it can trigger actions or run processes. Use rate limits and argument validation.
Register the Virtualization MCP server in PolicyLayer and add a rule for create_sandbox: allow, deny, rate-limit, or require approval. Point your MCP client at the PolicyLayer proxy URL and the rule is enforced on every call, before it reaches Virtualization. Nothing to install.
create_sandbox is a Execute tool with high risk. Execute tools should be rate-limited and have argument validation enabled.
Yes. Add a rate_limit block to the create_sandbox rule in your PolicyLayer policy. For example, setting max: 10 and window: 60 limits the tool to 10 calls per minute. Rate limits are tracked per agent session and reset automatically.
Set action: deny in the PolicyLayer policy for create_sandbox. The AI agent will receive a policy violation error and cannot call the tool. You can also include a reason field to explain why the tool is blocked.
create_sandbox is provided by the Virtualization MCP server (sandraschi/virtualization-mcp). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Start from Virtualization, add the rest of your stack, and see everything your agents can call. Then put policy on all of it.
Free to start. No card required.
56 Virtualization tools catalogued and risk-classified — across an index of 43,000+ MCP servers.