AI agents invoke execute_bash to trigger actions in Computer Use. 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.
This tool runs arbitrary bash commands with effects determined entirely by the command arguments. Bash execution can perform any operation the user has permissions for: read files, modify systems, install software, exfiltrate data, or pivot to other systems. The blast radius is maximal when an AI agent controls the input.
From the tool's definition Tool name 'execute_bash' and description 'Execute a bash command on the virtual computer' explicitly indicate arbitrary command execution capability.
Documented attack patterns abuse exactly the kind of access execute_bash gives an agent:
PolicyLayer is an MCP gateway — it sits between your AI agents and Computer Use, and nothing reaches the server without passing your rules. This is the rule we recommend for execute_bash:
{
"version": "1",
"default": "deny",
"tools": {
"execute_bash": {
"limits": [
{
"counter": "execute_bash_rate",
"window": "minute",
"max": 10,
"scope": "grant"
}
]
}
}
}execute_bash 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.
Execute a bash command on the virtual computer. It is categorised as a Execute tool in the Computer Use MCP Server, which means it can trigger actions or run processes. Use rate limits and argument validation.
Register the Computer Use MCP server in PolicyLayer and add a rule for execute_bash: 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 Computer Use. Nothing to install.
execute_bash 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 execute_bash 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 execute_bash. 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.
execute_bash is provided by the Computer Use MCP server (spencerkinney/computer-use-mcp). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Start from Computer Use, 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.
15 Computer Use tools catalogued and risk-classified — across an index of 43,000+ MCP servers.