The Agent Skills Directory

[COMMAND_EXECUTION]: Technical workflows involve the use of aggressive security tools including Nmap for service discovery and script-based enumeration, and Hydra or Medusa for high-volume credential brute-forcing attacks.
[DATA_EXFILTRATION]: The skill instructs the agent to locate and access highly sensitive files, including SSH private keys (id_rsa, id_dsa, etc.) and shell command history (~/.bash_history) which may contain plaintext credentials. It also includes commands to scan external web servers for exposed private keys using curl.
[PERSISTENCE]: A core component of the post-exploitation methodology involves establishing persistent access by appending an attacker-controlled public key to the ~/.ssh/authorized_keys file.
[CREDENTIALS_UNSAFE]: The workflow relies on the harvesting and management of sensitive cryptographic keys and the use of external wordlists for brute-force and password spraying attacks.
[DYNAMIC_EXECUTION]: Provides a functional Python script template using the paramiko library to programmatically connect to SSH hosts and execute arbitrary shell commands based on runtime logic.
[PRIVILEGE_ESCALATION]: Includes commands to identify potential privilege escalation vectors, specifically checking for sudo permissions with sudo -l and searching for sensitive files with elevated access requirements.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from target systems (service banners, file contents, configuration parameters) and lacks sanitization or boundary markers (delimiters) to prevent embedded instructions from influencing the agent's logic. Capabilities for further exploitation (command execution, network access) are present if the agent is influenced by malicious target data.

ssh-penetration-testing