skills/aibot88/sec_skill_store/kubernetes-live-network-architecture-mutation-guard/Gen Agent Trust Hub
kubernetes-live-network-architecture-mutation-guard
Pass
Audited by Gen Agent Trust Hub on Jun 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute kubectl commands including apply, patch, and create for modifying live Kubernetes networking architecture components.
- [EXTERNAL_DOWNLOADS]: Fetches configuration patterns and validation rules from well-known official sources such as kubernetes.io, docs.cilium.io, and coredns.io using the WebFetch tool.
- [DATA_EXFILTRATION]: The skill accesses sensitive cluster data including ServiceAccount tokens and ClusterMesh peer Secrets. Although the instructions explicitly forbid printing raw credentials and require sanitization, the technical capability to read and process these secrets exists through bash and Read tools.
- [PROMPT_INJECTION]: The skill processes untrusted data from user-provided YAML manifests and live cluster state, creating a surface for indirect prompt injection.
- Ingestion points: User-provided YAML files and kubectl resource output (SKILL.md).
- Boundary markers: No specific delimiters or "ignore instructions" warnings are defined for processed data.
- Capability inventory: Shell execution (bash), file system modification (Edit), and network access (WebFetch).
- Sanitization: The skill mentions using "sanitized YAML" but does not specify automated escaping or validation logic for the ingested content.
Audit Metadata