cve-source-check

Pass

Audited by Gen Agent Trust Hub on Jul 23, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill performs HTTP HEAD requests to verify the reachability of security advisory URLs. These URLs target established organizations and well-known services (e.g., GitHub, Oracle, Google, PostgreSQL, Elastic, and Ubuntu) which are considered safe sources for security metadata.
  • [COMMAND_EXECUTION]: The script performs standard file operations (reading inventory/registry files and writing Markdown/JSON reports) and uses network requests for URL validation. It does not utilize subprocess spawning or arbitrary command execution.
  • [DATA_EXFILTRATION]: No evidence of sensitive data harvesting or exfiltration was found. The network activity is restricted to best-effort reachability checks of public security feeds.
  • [PROMPT_INJECTION]: The instructions and documentation are focused on the audit workflow and do not contain patterns designed to bypass safety filters or override agent behavior.
  • [DYNAMIC_EXECUTION]: The script uses only static logic from the Python standard library. No dynamic code generation, runtime compilation, or unsafe deserialization (e.g., pickle) is employed.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 23, 2026, 09:19 PM
Security Audit — agent-trust-hub — cve-source-check