orca-emulator-android
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill facilitates the execution of local binaries (
orca,orca-ide, ororca-dev) to perform Android device management tasks. These tasks include booting AVDs, simulating user input (taps, swipes, typing), installing applications, and accessing system logs (logcat) and accessibility trees. - [INDIRECT_PROMPT_INJECTION]: The skill implements a dynamic instruction loading pattern where the agent is instructed to fetch the complete guide using the command
ORCA skills get orca-emulator-android. This creates a vulnerability surface where the output of a tool (the CLI binary) provides the core instructions that the agent will subsequently follow. - Ingestion points: The agent context is updated with the output of the
skills getcommand from the local binary. - Boundary markers: The stub does not specify delimiters or sanitization steps for the instructions fetched at runtime.
- Capability inventory: The skill provides extensive control over Android devices, including UI interaction, app management, and hardware button simulation.
- Sanitization: No explicit sanitization or validation of the fetched guide content is defined in the stub instructions.
Audit Metadata