exploiting-memory-corruption
Pass
Audited by Gen Agent Trust Hub on Aug 7, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious patterns, obfuscation, or unauthorized data access were detected. The skill provides technical reference material for security research and CTF tasks.
- [PROMPT_INJECTION]: The skill describes a process that ingests untrusted data (binaries and crash inputs), which represents an indirect prompt injection surface inherent to binary exploitation tasks.
- Ingestion points: Native binaries (ELF/PE) and crashing inputs provided by the user for analysis in the 'When to Use' and 'Enumerate the Mitigations First' sections.
- Boundary markers: None explicitly mentioned for isolating the agent from potential instructions embedded within the analyzed binaries.
- Capability inventory: Extensive use of binary analysis tools such as pwntools, GDB (pwndbg/GEF), and ROPgadget for research and payload development.
- Sanitization: Not described, as the tools are applied to target binaries in a research, CTF, or lab context.
Audit Metadata