The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to perform binary analysis by executing several command-line tools including otool, llvm-objdump, llvm-nm, codesign, lipo, file, and xxd. These commands are used to extract metadata, headers, and section information from Mach-O files for reverse engineering purposes.
[EXTERNAL_DOWNLOADS]: The documentation suggests obtaining the jtool2 utility from the Intezer GitHub repository (github.com/intezer/jtool2) either by cloning and building from source or downloading the release binary. This is a common practice in binary analysis workflows.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it involves processing untrusted external binary files and ingesting the output of analysis tools.
Ingestion points: The agent is directed to read the output of tools like otool, llvm-objdump, and llvm-nm when analyzing target binaries (SKILL.md, Operation Steps).
Boundary markers: The instructions do not define specific delimiters or warnings for the agent to ignore potentially malicious strings embedded within the binary metadata.
Capability inventory: The skill uses shell command execution to run various binary analysis and manipulation tools (SKILL.md, Tools Preparation).
Sanitization: There is no mention of sanitizing or escaping content extracted from binary strings or symbol tables before the agent processes them.

re-format-macho