AI agents invoke os_click to trigger actions in OScribe. 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.
Clicking UI elements can trigger arbitrary application actions (opening files, submitting forms, deleting data, launching programs, confirming dialogs). The blast radius is high because the exact effect depends on what is under the cursor at the time, and in the context of a vision-based desktop automation server it could interact with any application on the system.
From the tool's definition 'Click at current cursor position' — triggers a UI click action on whatever element is under the cursor, driving desktop automation via simulated user input
Attacks that exploit this kind of access
Click at current cursor position. Use os_move first to position the cursor, then os_click to click. This ensures precise clicking by separating movement and click. It is categorised as a Execute tool in the OScribe MCP Server, which means it can trigger actions or run processes. Use rate limits and argument validation.
Register the OScribe MCP server in PolicyLayer and add a rule for os_click: 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 OScribe. Nothing to install.
os_click 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 os_click 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 os_click. 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.
os_click is provided by the OScribe MCP server (mikealkeal/oscribe). 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 →