The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides Python scripts in files such as reference/nosql-injection-advanced.md and reference/sql-injection-advanced.md that utilize the requests library to send HTTP requests to external, user-specified targets for vulnerability scanning.
[COMMAND_EXECUTION]: Documentation files including reference/os-command-injection-cheat-sheet.md and reference/os-command-injection-quickstart.md contain an extensive catalog of OS command injection payloads. This includes numerous one-liners for spawning reverse shells via Bash, Python, Perl, and PHP interpreters.
[REMOTE_CODE_EXECUTION]: The reference materials detail specific techniques for achieving remote code execution across multiple vulnerability classes, such as Server-Side Template Injection (SSTI) and SQL Injection (e.g., via SQLite's load_extension).
[DATA_EXFILTRATION]: The skill includes payloads designed to demonstrate data exfiltration via out-of-band channels like DNS and HTTP, often using utilities such as nslookup, curl, or wget to transmit system information or file contents to a remote listener.
[PROMPT_INJECTION]: As a tool designed for vulnerability testing, the skill provides the agent with techniques to bypass security filters and inject commands into target systems. While these are intended for security research, the availability of these techniques represents an inherent risk surface for indirect injection if applied to untrusted data.

injection