gram
Warn
Audited by Gen Agent Trust Hub on Apr 10, 2026
Risk Level: MEDIUMCOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides commands to forcefully terminate system processes including 'x-cmd', 'curl', 'python', 'nodejs', 'bun', and 'deno' based on escalating safety levels.
- [COMMAND_EXECUTION]: High safety levels (3-5) allow the tool to archive and delete folders containing specific patterns ('soul.md', 'memory.md'), with level 5 expanding the search and deletion scope to the entire root directory ('/').
- [COMMAND_EXECUTION]: The tool includes functionality to 'sever' all HTTP/HTTPS network connections starting at safety level 2, which can disrupt system connectivity and external communications.
Audit Metadata