AI agents invoke mouse_drag to trigger actions in Kwin. 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.
Mouse drag is a GUI interaction action that triggers external operations on a live or virtual desktop. It can move UI elements, select content, or interact with applications in ways whose effects depend on arguments. The description is empty, lowering confidence, but the server context (desktop GUI automation) and sibling tools (keyboard_key, focus_window, etc.) strongly imply this is an Execute-category action.
From the tool's definition Tool name 'mouse_drag' on a server that 'automates Linux desktop GUI by launching and interacting with Wayland applications'; sibling tools include keyboard_type, focus_window, dbus_call suggesting GUI automation actions
Documented attack patterns abuse exactly the kind of access mouse_drag gives an agent:
PolicyLayer is an MCP gateway — it sits between your AI agents and Kwin, and nothing reaches the server without passing your rules. This is the rule we recommend for mouse_drag:
{
"version": "1",
"default": "deny",
"tools": {
"mouse_drag": {
"limits": [
{
"counter": "mouse_drag_rate",
"window": "minute",
"max": 10,
"scope": "grant"
}
]
}
}
}mouse_drag 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.
mouse_drag. It is categorised as a Execute tool in the Kwin MCP Server, which means it can trigger actions or run processes. Use rate limits and argument validation.
Register the Kwin MCP server in PolicyLayer and add a rule for mouse_drag: 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 Kwin. Nothing to install.
mouse_drag 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 mouse_drag 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 mouse_drag. 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.
mouse_drag is provided by the Kwin MCP server (isac322/kwin-mcp). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Start from Kwin, 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.
30 Kwin tools catalogued and risk-classified — across an index of 43,000+ MCP servers.