The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The skill requires administrative privileges (sudo) for critical forensic tasks such as loading kernel modules (insmod lime.ko) and accessing raw system memory devices (/dev/mem, /proc/kcore). This level of access is expected and necessary for memory acquisition on Linux and macOS systems.
[COMMAND_EXECUTION]: The skill utilizes multiple command-line tools for forensic analysis, including the Volatility 3 framework (vol), string extraction utilities (strings, floss), and system commands for file copying and disk dumping (dd, cp).
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze memory dumps (.raw, .lime, .elf, .vmem), which are untrusted external data sources.
Ingestion points: Forensic images and memory dumps are processed by analysis tools, and the results (e.g., process lists, command history, extracted strings) are presented to the agent in the SKILL.md workflows.
Boundary markers: The instructions do not define specific delimiters to separate the untrusted data extracted from memory from the agent's instructions.
Capability inventory: The skill environment includes file system operations (cp, dd, mv), package installation capabilities (pip), and the execution of analysis frameworks that may interact with the host system.
Sanitization: No specific sanitization or filtering of the content within the memory dumps is described prior to the agent interaction with output from commands like strings or volatility.

memory-forensics