AI agents call take_screenshot to retrieve information from Espresso without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This tool captures and retrieves the current visual state of an Android device. It does not modify, execute code, delete data, or move money. While screenshots could theoretically reveal sensitive information if misused, the tool itself is fundamentally a read operation with minimal blast radius—it merely queries the current UI state. Classification as Read is appropriate.
From the tool's definition Tool name 'take_screenshot' and description 'Take a screenshot of the connected Android device' indicate image capture without modification. Screenshots are read-only operations that retrieve visual data from the device state without side effects.
Attacks that exploit this kind of access
Take a screenshot of the connected Android device. It is categorised as a Read tool in the Espresso MCP Server, which means it retrieves data without modifying state.
Register the Espresso MCP server in PolicyLayer and add a rule for take_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 Espresso. Nothing to install.
take_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 take_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 take_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.
take_screenshot is provided by the Espresso MCP server (vs4vijay/espresso-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 →