AI agents call cmux_capture to retrieve information from Cmux without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This tool retrieves and queries terminal output without modifying, executing, or deleting anything. It allows agents to observe terminal state across workspaces but cannot change that state. The blast radius of misuse is minimal — an agent could only view sensitive information displayed on screen, not act on it.
From the tool's definition Tool description states 'Read the current screen (or scrollback) of a terminal surface (wraps read-screen)' — explicitly a read operation with no side effects.
Attacks that exploit this kind of access
Read the current screen (or scrollback) of a terminal surface (wraps read-screen). Pass workspace/window to read a surface that lives in another workspace. It is categorised as a Read tool in the Cmux MCP Server, which means it retrieves data without modifying state.
Register the Cmux MCP server in PolicyLayer and add a rule for cmux_capture: 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_capture is a Read tool with low risk. Read-only tools are generally safe to allow by default.
Yes. Add a rate_limit block to the cmux_capture 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_capture. 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_capture 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 →