The Agent Skills Directory

[PERSISTENCE]: The skill describes methods to maintain access to a target system across sessions, including modifying the system crontab (/etc/crontab) to run backdoor scripts and adding unauthorized public keys to the SSH authorized_keys file.
[PRIVILEGE_ESCALATION]: The instructions include commands to identify privilege escalation paths, such as searching for SUID binaries (find / -perm -4000), checking sudo permissions (sudo -l), and utilizing automated enumeration scripts like linpeas.sh and winpeas.exe to find system misconfigurations.
[COMMAND_EXECUTION]: The skill facilitates the execution of powerful security tools for scanning, brute-forcing, and exploitation, including Nmap, Metasploit, Hydra, and SQLMap, which can be used to perform significant actions on target systems.
[DATA_EXFILTRATION]: Reconnaissance phases involve techniques to discover and potentially extract sensitive data, such as using Google Dorks to locate publicly exposed .env files, configuration files, and directories containing passwords.
[INDIRECT_PROMPT_INJECTION]: This skill possesses an indirect injection attack surface as it requires the agent to ingest and process data from untrusted external sources (e.g., HTTP responses from target URLs, scanning tool outputs).
Ingestion points: Target URLs, IP addresses, and the output of tools like Nikto, Nmap, and Gobuster (SKILL.md).
Boundary markers: The skill includes a 'Mandatory confirmation gate' that requires the agent to show exact commands and wait for user confirmation before proceeding.
Capability inventory: Access to powerful tools including msfconsole, nmap, hydra, sqlmap, and general shell command execution.
Sanitization: No automated sanitization is provided; the skill relies on the user to manually review commands and outputs at the confirmation gate.

ethical-hacking-methodology