skills/andsnw/orca/orca-emulator/Gen Agent Trust Hub

orca-emulator

Warn

Audited by Gen Agent Trust Hub on Nov 11, 2026

Risk Level: MEDIUMPRIVILEGE_ESCALATIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [PRIVILEGE_ESCALATION]: The skill instructions explicitly direct the agent to 're-run it with escalated permissions' if a 'runtime_access_denied' error occurs. This encourages the agent to use elevated privileges (e.g., sudo) to bypass system-level sandbox or permission constraints.
  • [COMMAND_EXECUTION]: The skill provides detailed logic for the agent to resolve a local executable (e.g., orca, orca-ide, orca-dev) and execute shell commands to control simulators, hardware buttons, and accessibility trees.
  • [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to run 'ORCA skills get orca-emulator' to load a 'version-matched guide'. This pattern creates a surface for indirect prompt injection, as data retrieved from a tool's output is subsequently processed as instructions.
  • Ingestion points: Output from the 'ORCA skills get' command.
  • Boundary markers: None specified; the agent is instructed to load and follow the guide directly without delimiters or isolation.
  • Capability inventory: The skill has the capability to execute shell commands and perform simulator management tasks.
  • Sanitization: No validation or sanitization of the fetched guide content is described.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Nov 11, 2026, 05:29 AM