Release a held key.
AI agents invoke key_up to trigger actions in macOS Control MCP Server. 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.
Releasing a held key is a keyboard input action that executes external operations in the OS or applications. When combined with key_down, it can simulate keystrokes, trigger hotkeys, or perform other side-effecting desktop automation actions. The blast radius is medium since it can interact with any application or system function, but requires prior key_down state to be meaningful.
From the tool's definition Release a held key — part of keyboard automation that can trigger OS-level or application-level actions when used in combination with key_down
Attacks that exploit this kind of access
Release a held key. It is categorised as a Execute tool in the macOS Control MCP Server MCP Server, which means it can trigger actions or run processes. Use rate limits and argument validation.
Register the macOS Control MCP Server MCP server in PolicyLayer and add a rule for key_up: 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 macOS Control MCP Server. Nothing to install.
key_up 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 key_up 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 key_up. 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.
key_up is provided by the macOS Control MCP Server MCP server (lodimup/macos-control-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 →