re-sample-acquire
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill recommends downloading third-party security tools such as pe-sieve and hollows_hunter from GitHub and System Informer from its official website.
- [COMMAND_EXECUTION]: Instructions include executing high-privilege commands like procdump, pe-sieve, and gcore to scan and dump process memory on Windows, Linux, and macOS.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to capture raw memory artifacts from potentially malicious processes, creating an indirect prompt injection surface. 1. Ingestion points: Memory scanning via ReadProcessMemory (Windows), process_vm_readv (Linux), and mach_vm_read (macOS). 2. Boundary markers: None provided for the captured data. 3. Capability inventory: The skill can write memory dumps to disk and execute external analysis tools. 4. Sanitization: No sanitization is performed on the raw memory content before it enters the agent's context.
- [DYNAMIC_EXECUTION]: The skill suggests using Python with ctypes and Frida scripts to dynamically interact with and analyze memory regions at runtime.
Audit Metadata