virtualbox
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from virtualbox.org through tools like virtualbox_org__read_page and virtualbox_org__get_search. This creates a surface where malicious instructions embedded in web content could influence the agent's behavior.
- Ingestion points: Tools that read or search virtualbox.org (SKILL.md).
- Boundary markers: None observed in the instructions.
- Capability inventory: Tools for searching and reading virtualbox.org content.
- Sanitization: No specific sanitization or filtering of external content is described.
- [EXTERNAL_DOWNLOADS]: The skill instructions reference an external MCP server (https://unbrowse.ai/mcp/virtualbox.org) and installation via npx, both of which target the vendor's official infrastructure.
Audit Metadata