exploit-poc-development
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of untrusted data from vulnerability advisories and public exploit code, which could contain malicious instructions designed to influence the agent's behavior.
- Ingestion points: External vulnerability advisories, searchsploit output, and public exploit code (SKILL.md).
- Boundary markers: Absent. No specific delimiters are used to wrap external content.
- Capability inventory: Python execution and use of the pwntools library for network and binary exploitation.
- Sanitization: Absent. Procedural manual review is suggested, but no programmatic sanitization is defined.- [COMMAND_EXECUTION]: The skill methodology utilizes command-line tools like searchsploit and programming environments (Python/pwntools) to trigger vulnerability primitives and deliver payloads. This capability, while necessary for the skill's function, allows for arbitrary code execution during the exploit development process.
Audit Metadata