enterprise-vpn-attack
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONCREDENTIALS_UNSAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides numerous shell commands using
curl,python3,grep, andwcto perform version fingerprinting and vulnerability testing against external targets. - [CREDENTIALS_UNSAFE]: The
SKILL.mdfile contains a reference table of well-known factory default credentials for various enterprise VPN appliances including Cisco, Fortinet, Citrix, Palo Alto, Pulse Secure, F5, and SonicWall. - [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of untrusted data by instructing the agent to fetch and analyze HTTP responses and SAML metadata from external target servers.
- Ingestion points: Network responses fetched via
curlin the 'Vendor identification', 'CVE matrix', and 'AAA backend identification' sections ofSKILL.md. - Boundary markers: Absent. The skill does not provide instructions to the agent to ignore potentially malicious content within the fetched data.
- Capability inventory: The skill environment permits command execution via
curl,python3, and shell utilities, which could be exploited if an ingested response contains instructions recognized by the agent. - Sanitization: Absent. Fetched content is processed directly through shell pipes or redirected to temporary files for inspection.
Audit Metadata