The Agent Skills Directory

[Command Execution] (HIGH): The skill includes the command sudo ldconfig. Executing commands with sudo is a form of privilege escalation that allows the agent to perform administrative actions. In an automated environment, this could be abused to modify system-wide library search paths or library caches, potentially leading to system compromise.
[Indirect Prompt Injection] (LOW): The skill is designed to process untrusted ELF binaries provided by the user. Tools such as ldd can, in some environments or with specific binary configurations, execute code from the binary being analyzed to resolve dependencies.
Ingestion points: Untrusted ELF executables and shared libraries analyzed by ldd, nm, readelf, and objdump (e.g., prog, ./libfoo.so).
Boundary markers: None identified. The skill does not provide instructions to the agent to treat binary outputs as untrusted data.
Capability inventory: Execution of binary analysis tools (nm, readelf, objdump, ldd) and system commands (sudo, ldconfig, strip, ln).
Sanitization: No sanitization or sandboxing is suggested for the analysis of potentially malicious binaries.

elf-inspection