computer-use-playbook
Pass
Audited by Gen Agent Trust Hub on Mar 3, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [SAFE]: The skill incorporates extensive safety guidelines for handling sensitive automation tasks. It explicitly forbids attempting to bypass CAPTCHAs or security gates, instead requiring a 'human-in-the-loop' handoff. This design significantly reduces the risk of unauthorized account access or bot-like behavior detection.
- [COMMAND_EXECUTION]: The skill instructions utilize standard system utilities such as xdotool, xprop, rg, and find to perform desktop and filesystem automation. These tools are used according to their intended functions within the defined scope of the skill's computer-use playbook.
- [PROMPT_INJECTION]: No malicious prompt injection or instruction override patterns were detected. The skill uses natural instructional language to define its operational logic and escalation paths, which is appropriate for its intended use case.
Audit Metadata