elf-inspection
Audited by Gen Agent Trust Hub on Sep 14, 2026
The 'elf-inspection' skill is designed to assist agents in analyzing Linux binaries. It uses standard tools like readelf, nm, and objdump. 1. Privilege Escalation: The skill contains an instruction to run 'sudo ldconfig'. This command updates the system's shared library cache and requires root privileges. While necessary for certain troubleshooting tasks (e.g., after installing new libraries), using sudo is a high-privilege action. Given it aligns with the skill's purpose, the severity is adjusted to medium. 2. Indirect Prompt Injection: By design, this skill processes external binary data. There is a low-severity risk that specially crafted binaries could contain strings or metadata intended to influence the agent or exploit underlying analysis tools. 3. External References: The skill mentions 'bloaty' (hosted by Google on GitHub) and 'checksec', and provides links to Red Hat and Linux man pages. These references are to trusted organizations or well-known services and are considered safe. No obfuscation, data exfiltration, or persistence mechanisms were detected.