oz-upload-file

Pass

Audited by Gen Agent Trust Hub on Aug 23, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
  • [COMMAND_EXECUTION]: The skill provides a shell command template: "$OZ_CLI" artifact upload '<path>' --run-id "$OZ_RUN_ID" --description '<description>'. This template relies on the agent to interpolate values for <path> and <description>. While single quotes are used, the lack of robust sanitization or escaping mechanisms for these parameters means that an agent could inadvertently or maliciously include shell metacharacters that lead to command execution in the host shell environment.
  • [DATA_EXFILTRATION]: The skill is designed to upload local files to an external platform (Oz). By providing a command that reads a file at an arbitrary path and sends it to a remote endpoint, the skill creates a data exfiltration surface. If the agent is successfully targeted by a prompt injection or behaves autonomously to access sensitive files (such as .env files, SSH keys, or cloud credentials), this tool could be used to transmit that sensitive data to the external Oz platform.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 23, 2026, 04:59 AM
Security Audit — agent-trust-hub — oz-upload-file