The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill provides instructions to download and execute scripts directly from non-trusted external repositories via shell pipes, such as piping linpeas.sh from GitHub into sh.
[REMOTE_CODE_EXECUTION]: It details procedures for downloading pre-compiled exploit binaries like PwnKit from untrusted sources and executing them on the target system.
[DATA_EXFILTRATION]: The skill contains commands specifically designed to exfiltrate sensitive files, including the system shadow file, to attacker-controlled remote servers using tools like wget.
[CREDENTIALS_UNSAFE]: It directs the agent to scan for and read sensitive files containing credentials, such as AWS configuration files, SSH private keys, and environment variables.
[PRIVILEGE_ESCALATION]: Extensive instructions are provided for gaining root privileges by modifying critical system files like /etc/passwd and /etc/shadow, or exploiting kernel-level vulnerabilities.
[PERSISTENCE]: The playbook includes methods for maintaining unauthorized access, such as appending SSH keys to authorized_keys files and setting up malicious cron jobs for reverse shells.
[DYNAMIC_EXECUTION]: It employs advanced exploitation techniques like shared library hijacking via LD_PRELOAD and process memory injection using gdb to execute code with elevated permissions.
[COMMAND_EXECUTION]: The skill lists numerous standard Linux binaries that can be abused to break out of restricted shells and execute commands with the permissions of the binary owner.

linux-privilege-escalation