AI agents invoke cmux_browser to trigger actions in Cmux. 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.
Browser control via an MCP agent involves executing arbitrary actions in a web browser, which can navigate to sites, interact with web applications, and trigger external operations. This is Execute-class because the effects depend on agent-supplied arguments and can include navigation, form submission, clicks, and other browser interactions.
From the tool's definition Tool name 'cmux_browser' and server description states it 'let agents control terminal panes, workspaces, input, and browser.' The description is incomplete ('Run a'), but the context indicates browser control capability.
Attacks that exploit this kind of access
Run a. It is categorised as a Execute tool in the Cmux MCP Server, which means it can trigger actions or run processes. Use rate limits and argument validation.
Register the Cmux MCP server in PolicyLayer and add a rule for cmux_browser: 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 Cmux. Nothing to install.
cmux_browser 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 cmux_browser 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 cmux_browser. 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.
cmux_browser is provided by the Cmux MCP server (puchkoff/cmux-mcp). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Every MCP server has a record like this.
Type a name, get the same breakdown: verified identity, auth posture, risk grade, capabilities, recommended policy.
Teams ship this data inside their own products. See what a licence covers →