The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use the Bash tool to execute llvm-dwarfdump and readelf commands. These are intended behaviors for analyzing binary files provided by the user.
[INDIRECT_PROMPT_INJECTION]: The skill involves parsing and analyzing DWARF metadata from external binary files. This creates a surface for indirect prompt injection, as a malicious binary could contain specially crafted debug information designed to influence the agent's output.
Ingestion points: Binary files processed by analysis tools (llvm-dwarfdump, dwarfdump, readelf) as described in SKILL.md.
Boundary markers: None present; the agent is instructed to read and interpret the output of these tools directly.
Capability inventory: The skill utilizes Bash for command execution, Read for file access, and WebSearch for external information retrieval.
Sanitization: No specific sanitization or validation of the binary content is performed prior to parsing.
[EXTERNAL_DOWNLOADS]: The skill references legitimate external technical resources, including the official DWARF standard (dwarfstd.org) and the libdwarf reference implementation on GitHub. These references are used for information retrieval and are consistent with the skill's stated purpose.

dwarf-expert