The Agent Skills Directory

[PERSISTENCE]: The skill provides explicit instructions for establishing persistence on a target system. This includes utilizing Meterpreter persistence modules and modifying system-level files by appending malicious commands to /etc/crontab via echo operations.
[PRIVILEGE_ESCALATION]: Detailed guidance is included for escalating privileges on compromised systems. The skill instructs on searching for SUID binaries (find / -perm -4000), checking sudo permissions (sudo -l), and utilizing external enumeration scripts like linpeas.sh, linux-exploit-suggester.sh, and winpeas.exe to identify kernel and configuration vulnerabilities.
[COMMAND_EXECUTION]: The skill facilitates the execution of powerful and potentially destructive CLI tools, including nmap for aggressive network scanning, msfconsole for exploit delivery, hydra for brute-forcing credentials, and dd for writing disk images.
[DATA_EXFILTRATION]: Instructions cover techniques for identifying and accessing sensitive data, such as leveraging Google Dorks to find .env and configuration files, and using command injection to read sensitive local files like /etc/passwd.
[EXTERNAL_DOWNLOADS]: The skill describes downloading external software and scripts, including Kali Linux ISO images and various third-party security enumeration tools. While these are common in penetration testing, they represent external code being introduced into the environment.
[INDIRECT_PROMPT_INJECTION]: The skill's workflow involves ingesting data from external, potentially hostile targets (e.g., DNS records, web page content, service banners), which could contain malicious instructions designed to influence the agent's behavior.
Ingestion points: Data gathered during reconnaissance and scanning phases from external target servers enters the agent's execution context (e.g., theHarvester, whatweb, and nmap outputs).
Boundary markers: The skill includes a 'Mandatory confirmation gate' requiring the agent to wait for explicit user confirmation before acting on gathered data or executing intrusive commands.
Capability inventory: The skill has access to shell execution and network operations across multiple scripts and workflow phases.
Sanitization: There is no evidence of automated sanitization or filtering of external content before it is processed by the agent.

ethical-hacking-methodology