The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill relies heavily on tools and scripts hosted on external GitHub repositories (e.g., SysWhispers3, pe-sieve, CallStackSpoofer, Ekko, Foliage) that are not associated with recognized trusted organizations or well-known services.
[REMOTE_CODE_EXECUTION]: Instructions direct the user to run external scripts, specifically "python3 syswhispers.py", which is downloaded from a third-party repository. This pattern of downloading and executing code from unverified sources is a significant risk.
[COMMAND_EXECUTION]: The skill utilizes shell and PowerShell commands to perform system-level operations, such as fingerprinting services ("Get-Service"), identifying kernel drivers ("fltmc filters"), and executing security-evasion binaries ("pe-sieve64.exe", "sysmon64.exe").
[DYNAMIC_EXECUTION]: Detailed procedures are provided for runtime memory modification, including patching "EtwEventWrite" to disable Event Tracing for Windows (ETW) and "AmsiScanBuffer" to bypass the Antimalware Scan Interface (AMSI). It also describes using indirect syscalls and hardware breakpoints to manipulate program execution flow.
[INDIRECT_PROMPT_INJECTION]:
Ingestion points: The skill gathers data from the host environment, including service lists, driver lists, and process memory dumps.
Boundary markers: There are no explicit markers or instructions to isolate or sanitize this environmental data.
Capability inventory: The skill possesses capabilities for Python script execution, binary execution, and deep manipulation of system processes and memory permissions.
Sanitization: No sanitization or validation of external tool outputs or environmental data is described before processing or use in further commands.

reverse-eng-edr-analysis