computer-use
Pass
Audited by Gen Agent Trust Hub on Sep 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes UI accessibility trees from arbitrary desktop applications, creating a surface for indirect prompt injection. * Ingestion points: The
ORCA computer get-app-statecommand returns thetreeTextof targeted applications (SKILL.md). * Boundary markers: No specific delimiters or safety wrappers are defined for the ingested UI content. * Capability inventory: The skill possesses extensive system interaction capabilities, includingclick,set-value,type-text, andhotkey(SKILL.md). * Sanitization: There is no evidence of sanitization or filtering applied to the UI data before it is processed by the agent. - [COMMAND_EXECUTION]: The skill operates by executing local CLI commands (
orca,orca-dev, etc.) with various flags and arguments to control the desktop environment.
Audit Metadata