Capture a screenshot with UI elements and cursor position. Returns: (1) the image, (2) cursor coordinates (x, y), (3) UI elements from Windows UI Automation with their screen coordinates. Use this to see the screen AND know where to click precisely. ⚠️ CRITICAL - HOW TO CLICK ON ELEMENTS: - ALWAY...
AI agents call os_screenshot to retrieve information from OScribe without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
os_screenshot is a passive observation tool that retrieves the current state of the desktop via screenshot and UI element metadata. It does not execute commands, modify data, delete anything, or commit financial actions. While it is a prerequisite for other tools on the server that perform actions (os_click, os_drag, etc.), this tool itself only reads and returns information.
From the tool's definition Tool description states it 'Capture[s] a screenshot' and 'Returns: (1) the image, (2) cursor coordinates, (3) UI elements'.
Attacks that exploit this kind of access
Capture a screenshot with UI elements and cursor position. Returns: (1) the image, (2) cursor coordinates (x, y), (3) UI elements from Windows UI Automation with their screen coordinates. Use this to see the screen AND know where to click precisely. ⚠️ CRITICAL - HOW TO CLICK ON ELEMENTS: - ALWAYS use the center=(x,y) coordinates from the Elements list below - NEVER guess or estimate positions visually from the image - The JSON coordinates are EXACT screen positions, the image is only for visual context Example: To click on Button. It is categorised as a Read tool in the OScribe MCP Server, which means it retrieves data without modifying state.
Register the OScribe MCP server in PolicyLayer and add a rule for os_screenshot: 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_screenshot 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 os_screenshot 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_screenshot. 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_screenshot 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 →