skills/nplez1/orca/computer-use/Gen Agent Trust Hub

computer-use

Fail

Audited by Gen Agent Trust Hub on Oct 9, 2026

Risk Level: HIGHPRIVILEGE_ESCALATIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [PRIVILEGE_ESCALATION]: The skill explicitly instructs the agent to escalate its privileges if a command fails due to sandbox restrictions. Under the 'Load the version-matched guide before running Orca commands' section, it states: 're-run it with escalated permissions' if it fails with 'runtime_access_denied'. This encourages the agent to bypass security controls implemented by the host environment.
  • [COMMAND_EXECUTION]: The instructions detail a process for resolving and executing various local CLI tools (orca, orca-dev, orca-ide) depending on the environment variables and operating system. It directs the agent to run commands like ORCA skills get computer-use and ORCA open --json to manage the GUI automation service.
  • [INDIRECT_PROMPT_INJECTION]: The skill relies on dynamically loading instructions via the command ORCA skills get computer-use. This creates an attack surface where the output of the CLI tool—if influenced by an attacker or external state—could inject malicious instructions into the agent's context. The skill lacks specific boundary markers or sanitization instructions for this ingested content.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Oct 9, 2026, 12:51 PM