pentest-cve-vulnerability-research-helper
Pass
Audited by Gen Agent Trust Hub on Sep 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted content from external sources, including web search results, GitHub repositories, and vulnerability databases such as Exploit-DB and Sploitus. This creates a surface where malicious instructions embedded in CVE descriptions or exploit PoCs could potentially influence agent behavior.
- Ingestion points: Web search queries and curl operations targeting external URLs derived from research results.
- Boundary markers: The skill lacks explicit boundary markers or instructions to treat external data as strictly non-instructional text.
- Capability inventory: The skill utilizes subprocess execution for the vulnx tool and performs network operations via curl and web search utilities.
- Sanitization: Data is filtered using jq and grep, but no semantic sanitization is applied to prevent the agent from interpreting instructions within the data.
- [COMMAND_EXECUTION]: The skill executes the vulnx security tool and standard CLI utilities like curl, grep, and jq. These operations are intended for vulnerability research and include safety checks, such as verifying binary existence and testing authentication before execution. It correctly handles the PDCP_API_KEY using environment variable placeholders rather than hardcoding credentials.
Audit Metadata