disk-cleaner

Pass

Audited by Gen Agent Trust Hub on May 7, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill makes extensive use of Bash commands including du, find, df, and docker to inspect the filesystem. It is designed to perform interactive file deletions (rm) based on user feedback after providing a report.
  • [DATA_EXPOSURE]: The skill scans potentially sensitive locations such as ~/Library/Application Support/ and all hidden directories in the user's home folder (e.g., .ssh, .aws, .env). While it only reports on file sizes and paths, this exposes filesystem metadata to the model.
  • [PROMPT_INJECTION]: User-provided arguments are directly interpolated into the prompt via $ARGUMENTS. This is a potential vector for direct prompt injection where a user could provide crafted input to alter the agent's intended behavior or bypass the 'Safety Rules' defined in the skill.
Audit Metadata
Risk Level
SAFE
Analyzed
May 7, 2026, 02:45 AM
Security Audit — agent-trust-hub — disk-cleaner