huawei-cloud-cdn-dns-resolution-diagnosis

Pass

Audited by Gen Agent Trust Hub on Sep 10, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill demonstrates an indirect prompt injection attack surface because it processes untrusted user-provided domain names and resolves them to IP addresses that are subsequently passed as arguments to cloud CLI commands and included in final reports.
  • Ingestion points: The domain_name input parameter is accepted via user triggers in SKILL.md and passed to the scripts/dns_resolve.py script.
  • Boundary markers: The agent produces a structured text diagnosis report defined in references/task-report-generation.md, using clear headers (e.g., ==================== CDN DNS Resolution Diagnosis Report ====================) to separate tool outputs from agent instructions.
  • Capability inventory: The skill has the capability to execute local shell commands via the hcloud CLI (ShowDomainDetailByName, ShowIpInfo/v2) and run a local Python script (scripts/dns_resolve.py).
  • Sanitization: The skill includes proactive sanitization in scripts/dns_resolve.py, which uses a strict regular expression (_DOMAIN_PATTERN) to validate domain names against RFC 1035 standards before performing any DNS resolution or cloud API calls.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 10, 2026, 08:45 AM
Security Audit — agent-trust-hub — huawei-cloud-cdn-dns-resolution-diagnosis