The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill provides numerous actionable payloads to establish remote shell access across multiple languages and protocols.
Evidence includes ready-to-use Python, Bash, PHP, and Netcat reverse shells (e.g., python3 -c 'import socket,subprocess,os;s=socket.socket...').
Instructions include shell stabilization techniques using Python PTY spawning (pty.spawn("/bin/bash")).
[DATA_EXFILTRATION]: The skill implements a structured workflow for post-compromise data harvesting.
It provides commands to search for and read sensitive 'flag' files (find / -name "user.txt").
It includes logic to store captured flags into local state files for later reporting.
It provides commands to read sensitive system files like /etc/passwd via multiple vectors (SQLi, File Upload, Web Shells).
[EXTERNAL_DOWNLOADS]: The skill automates the discovery and retrieval of exploit code from untrusted third-party sources.
It includes commands to query the GitHub API for CVE exploits and download them using wget from unverified repositories.
It facilitates the use of external exploitation frameworks like Metasploit (msfconsole) and sqlmap.
[COMMAND_EXECUTION]: The skill encourages the bypass of security controls such as WAFs and keyword filters.
It instructs on the use of Base64 encoding to obfuscate payloads and pipe them directly into a shell (echo BASE64 | base64 -d | bash).
It provides lists of alternative commands to evade detection (e.g., using head or nl instead of cat).
[CREDENTIALS_UNSAFE]: The skill advocates for credential brute-forcing and the testing of default credentials against SSH, FTP, and MySQL services using tools like hydra.

exploitation-knowledge