The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides detailed instructions for executing high-risk commands to achieve persistence and escalate privileges:
Persistence methods include modifying cron jobs (echo "* * * * * /tmp/backdoor.sh" >> /etc/crontab) and adding unauthorized SSH keys to ~/.ssh/authorized_keys.
Privilege escalation techniques include checking for sudo -l permissions and searching for SUID binaries (find / -perm -4000).
Exploitation commands involve the Metasploit Framework (msfconsole) and automated tools like sqlmap and hydra.
[CREDENTIALS_UNSAFE]: The methodology encourages targeting and exposing sensitive information:
Google Dorking techniques specifically look for .env files and configuration files.
Command injection testing examples include accessing protected system files like /etc/passwd.
[DATA_EXFILTRATION]: The reconnaissance phase includes tools and techniques for gathering organizational intelligence, such as email harvesting with theHarvester and extensive DNS enumeration using dnsrecon and dig.
[EXTERNAL_DOWNLOADS]: The skill references external resources for security testing, such as the Kali Linux ISO from kali.org and third-party privilege escalation enumeration scripts including linpeas.sh, winpeas.exe, and linux-exploit-suggester.sh.
[COMMAND_EXECUTION]: (Vulnerability Surface for Indirect Prompt Injection):
Ingestion points: The skill processes data from external sources and tools such as theHarvester, nikto, and web directory scanners in SKILL.md.
Boundary markers: No delimiters or instructions are provided to the agent to ignore potentially malicious instructions embedded in tool outputs.
Capability inventory: The skill has broad capabilities to execute shell commands, perform network operations, and write to the file system.
Sanitization: There is no evidence of output sanitization or validation before the agent acts on information retrieved from targets.

Ethical Hacking Methodology