The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use several command-line tools for binary analysis, including binwalk for file signature scanning, dd for data extraction, objcopy for section dumping, tshark for network data extraction, and xxd for hex manipulation.
[DYNAMIC_EXECUTION]: The instructions include a Python script that uses the Unicorn engine to perform CPU emulation of extracted shellcode. This involves mapping memory and executing arbitrary binary code taken from untrusted external sources.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted binary data, creating a surface for indirect prompt injection.
Ingestion points: Binary blobs are extracted from host files (sample.bin), memory dumps, and network captures (c2.pcap).
Boundary markers: There are no explicit boundary markers or instructions to isolate the data content from the agent's reasoning process in the provided templates.
Capability inventory: The skill utilizes file system access, command execution, and code emulation.
Sanitization: No data sanitization or filtering is performed on the binary content before analysis or emulation.
[EXTERNAL_DOWNLOADS]: The skill references and suggests installing several well-known security libraries and tools, such as unicorn, capstone, binwalk, and metasploit-framework, using standard package managers.
[PRIVILEGE_ESCALATION]: The tool setup instructions include commands like apt install, dnf install, and pacman -S, which typically require administrative (root/sudo) privileges to execute on Linux systems.

re-shellcode