The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill provides comprehensive exploitation chains to achieve arbitrary code execution in various template engines, including Jinja2, Twig, Mako, FreeMarker, and Pug.
[COMMAND_EXECUTION]: Instructions guide the agent to execute system commands such as id, env, find, and cat via templates on target servers.
[DATA_EXFILTRATION]: Payloads are included to specifically extract sensitive system files like /etc/passwd and /flag.txt, as well as application configuration secrets like SECRET_KEY and database connection strings.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface. Ingestion points: Target server responses are processed via analyze_response to locate injection points. Boundary markers: None are defined to isolate untrusted server output from the agent's context. Capability inventory: The agent uses http_request and python3 to interact with and exploit targets. Sanitization: No validation or sanitization of target server output is performed before the agent uses it to determine subsequent actions.

ssti-methodology