The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute untrusted binaries for dynamic analysis using tools like strace, ltrace, and direct execution (e.g., ./suspicious). These commands are intended for sandboxed environments but pose a risk if executed on the host system.
[PRIVILEGE_ESCALATION]: Several commands require administrative privileges to function, specifically for network operations and packet manipulation. Evidence includes the use of sudo for arpspoof, route add, tcpdump, and wireshark components.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data which could contain malicious instructions.
Ingestion points: Malware samples (suspicious_file, mal.exe), network captures (capture.pcap), Debian packages (package.deb), and memory dumps (memory.dmp).
Boundary markers: None provided in the instructions to distinguish between analysis tool output and potential embedded instructions.
Capability inventory: The skill uses bash, python3, strace, ltrace, yara, tshark, and arpspoof, providing a significant surface for exploitation if an injected instruction is followed.
Sanitization: There is no explicit sanitization or filtering of the content extracted from the files being analyzed.
[DYNAMIC_EXECUTION]: The skill provides Python scripts for automated malware execution and shellcode emulation.
Evidence: scripts-and-obfuscation.md contains a run_sample function that uses subprocess.run to execute a file from a provided path. It also includes an emulator using the unicorn engine and capstone for programmatic disassembly.
[EXTERNAL_DOWNLOADS]: The skill requires the installation of numerous third-party tools and libraries from public registries and repositories.
Evidence: Extensive pip install commands for security libraries, apt/brew commands for analysis utilities, and manual installation links to GitHub repositories for tools like dnSpy and PyArmor-Unpacker.

ctf-malware