The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes curl commands and a Python script (scripts/agent.py) to interact with user-defined web endpoints. These tools are designed to automate the testing of authentication mechanisms by providing type-coerced payloads (such as boolean true or integer 0) to detect loose comparison vulnerabilities in PHP.
[EXTERNAL_DOWNLOADS]: The skill references external security research materials, specifically the PayloadsAllTheThings repository, to provide users with lists of magic hashes. These references are common in the cybersecurity domain and point to reputable community-driven resources.
[DATA_EXFILTRATION]: Technical review of the scripts/agent.py file confirmed that it only performs network requests to target URLs specified by the user via command-line arguments. The script does not access local sensitive files, system environment variables, or hardcoded credentials.
[DYNAMIC_EXECUTION]: The skill does not contain any patterns for unsafe dynamic execution, such as eval(), exec(), or the runtime loading of external code from untrusted sources. All Python code is static and follows standard practices for a command-line testing tool.

exploiting-type-juggling-vulnerabilities