reverse-eng-binary-triage
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill describes a methodology for analyzing untrusted native binary files (ELF, PE, Mach-O), which constitutes an ingestion surface for potentially malicious data designed to influence the analyzing agent.
- Ingestion points: Native binaries and their associated metadata/strings loaded during triage (SKILL.md).
- Boundary markers: None explicitly specified within the skill instructions for separating binary content from agent instructions.
- Capability inventory: The skill utilizes diagnostic and debugging tools including
strings,gdb,ltrace,ghidra, andradare2to inspect and execute binary logic. - Sanitization: The instructions do not define automated sanitization but include a manual safety control by advising the user to analyze untrusted binaries only within an isolated VM environment.
Audit Metadata