The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to analyze untrusted Linux ELF binaries, creating a potential indirect prompt injection surface.
Ingestion points: The scripts/agent.py script and instructions in SKILL.md process user-provided binary files to extract metadata and strings.
Boundary markers: No specific delimiters are used to isolate binary content from the agent's reasoning process.
Capability inventory: The skill uses subprocess to call system utilities and has file read access.
Sanitization: Binary content is inspected directly without sanitization, which is expected for forensic analysis but represents an inherent risk surface.
[COMMAND_EXECUTION]: The script scripts/agent.py executes system tools like strings, readelf, and upx via subprocess.run(). While these are legitimate analysis actions using safe argument passing, they constitute a capability to interact with the underlying system based on the target binary path provided to the script.
[EXTERNAL_DOWNLOADS]: The skill utilizes the pyelftools library, which is a well-known Python package for ELF analysis. If not present, the documentation directs the user to install it via pip, the official Python package manager.

analyzing-linux-elf-malware