linux-shell-scripting

Warn

Audited by Gen Agent Trust Hub on Aug 8, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCREDENTIALS_UNSAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill contains multiple templates that require or utilize administrative privileges to modify the system state.
  • Evidence: Usage of sudo apt-get install in Phase 7 for automated package installation.
  • Evidence: Usage of sudo systemctl in Phase 7 for service management.
  • Evidence: Instructions for user account creation (useradd) and password management (passwd, chage) in Phase 3 which typically require root access.
  • [COMMAND_EXECUTION]: The skill provides logic to establish persistence on a Linux system.
  • Evidence: Phase 7 includes a 'Task Scheduler' script that programmatically modifies the user's crontab using (crontab -l; echo ...) | crontab - to ensure scripts run on a recurring schedule.
  • [REMOTE_CODE_EXECUTION]: The skill facilitates the execution of code on remote systems.
  • Evidence: Phase 10 features a 'Remote Script Execution' template using ssh "$remote_server" "bash -s" < "$remote_script", which executes local script content on a remote target.
  • [EXTERNAL_DOWNLOADS]: The skill includes functionality to interact with and download data from the internet.
  • Evidence: Phase 6 uses curl to perform head requests against external websites for uptime monitoring.
  • Evidence: Phase 7 uses apt-get to download and install packages from remote repositories.
  • [CREDENTIALS_UNSAFE]: The skill templates demonstrate insecure credential handling practices.
  • Evidence: Phase 1 includes a database backup template that hardcodes placeholders for db_user and db_pass, and passes the password directly to the mysqldump command line, which can expose credentials in process lists (e.g., via ps).
Audit Metadata
Risk Level
MEDIUM
Analyzed
Aug 8, 2026, 01:43 AM
Security Audit — agent-trust-hub — linux-shell-scripting