skills/mukul975/anthropic-cybersecurity-skills/performing-authenticated-scan-with-openvas/Gen Agent Trust Hub
performing-authenticated-scan-with-openvas
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructions include commands to download official Docker configuration files from Greenbone's GitHub-hosted documentation (
greenbone.github.io). This is a standard and expected operation for setting up the community edition of the software. - [INDIRECT_PROMPT_INJECTION]: The scripts
scripts/agent.pyandscripts/process.pyingest and parse XML vulnerability reports generated by the OpenVAS scanner. This presents a potential surface for indirect prompt injection if a scanned target were to return malicious data that is then interpreted by the processing scripts or an agent using them. - Ingestion points: XML report parsing in
scripts/agent.py(parse_openvas_xml_reportfunction) andscripts/process.py(export_report_csvfunction). - Boundary markers: None implemented; the scripts extract text directly from XML tags.
- Capability inventory: The scripts use the
python-gvmlibrary to interact with the GVM management socket and can trigger CLI commands viagvm-clias shown in the documentation. - Sanitization: There is no explicit sanitization of the vulnerability descriptions or names extracted from the scan results before they are printed or exported to CSV/JSON.
Audit Metadata