Take screenshot (base64) + UI tree in one call. Recommended for
AI agents call adb_observe to retrieve information from ADB MCP Server without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
The adb_observe tool captures visual and structural information from the device screen without modifying any state, executing commands, or triggering actions. Screenshots and UI parsing are passive observations with no side effects, fitting the Read category. Severity is low because screen captures alone pose minimal risk even in adversarial scenarios.
From the tool's definition Tool takes a screenshot and UI tree, both read-only operations with no side effects. Description states 'Take screenshot (base64) + UI tree' which are data retrieval operations.
Attacks that exploit this kind of access
Take screenshot (base64) + UI tree in one call. Recommended for. It is categorised as a Read tool in the ADB MCP Server MCP Server, which means it retrieves data without modifying state.
Register the ADB MCP Server MCP server in PolicyLayer and add a rule for adb_observe: 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 ADB MCP Server. Nothing to install.
adb_observe 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 adb_observe 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 adb_observe. 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.
adb_observe is provided by the ADB MCP Server MCP server (lll-404/adb-mcp-server). 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 →