cloud-system-administrator
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill establishes operational procedures that require the agent to ingest and act upon data from external, untrusted sources, creating a potential attack surface.
- Ingestion points: The agent is instructed to process access request tickets (
references/iam_access_operations.md), triage infrastructure alerts, and monitor vendor status pages (references/monitoring_incident_response.md). - Boundary markers: The instructions do not define delimiters or explicit "ignore embedded instructions" warnings for the agent when it parses content from external tickets or alert descriptions.
- Capability inventory: The skill describes high-impact capabilities including IAM role assignment, credential rotation, DNS record modification, and compute instance management (
references/compute_os_maintenance.md,references/operational_hygiene.md). - Sanitization: There are no instructions for validating, escaping, or filtering external ticket or alert content before the agent processes it.
- [NO_CODE]: The skill consists exclusively of markdown documentation and reference guides; no executable scripts, binary files, or package manifests are present.
Audit Metadata