computer-use-remote

Pass

Audited by Gen Agent Trust Hub on May 1, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: No malicious patterns or security risks were identified. The instructions follow best practices for grounding AI actions in visual evidence and respecting user control signals.- [INDIRECT_PROMPT_INJECTION]: The skill describes tools that ingest visual data from a remote machine, which represents a potential attack surface. However, the instructions encourage safe grounding practices to mitigate accidental obedience.
  • Ingestion points: Remote desktop screenshots via computer_use_remote (SKILL.md).
  • Boundary markers: Absent.
  • Capability inventory: Desktop control and remote code execution (SKILL.md).
  • Sanitization: Absent.
Audit Metadata
Risk Level
SAFE
Analyzed
May 1, 2026, 07:34 AM