The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes various security tools and command-line utilities to perform scanning and exploitation.
Spawns and interacts with msfconsole using start_process and send_command.
Runs nmap, msfvenom, and nc for reconnaissance and payload handling.
Executes standalone Python exploit scripts for MS17-010 and MS08-067.
[REMOTE_CODE_EXECUTION]: The skill is designed to achieve remote code execution on Windows systems by exploiting vulnerabilities in the SMB protocol.
Targets multiple CVEs including CVE-2008-4250, CVE-2017-0143, and CVE-2020-0796.
Provides specific exploitation paths for different Windows OS versions and service packs.
[DATA_EXFILTRATION]: The exploitation methodology establishes network connections from the target host to the attacker.
Uses reverse shell payloads that connect back to a user-defined LHOST and LPORT.
[CREDENTIALS_UNSAFE]: The skill provides instructions for sensitive data collection following a successful compromise.
Directs the agent to dump credentials and save evidence like AWS keys to local storage.
[PROMPT_INJECTION]: The skill processes untrusted external data that could lead to indirect prompt injection.
Target IP addresses and OS details are interpolated into shell commands and Metasploit configuration without sanitization.
Lacks boundary markers or instructions to ignore embedded commands in target-sourced data.

smb-exploitation