kubernetes-specialist
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches configuration and installation manifests from official sources for ArgoCD, Istio, Linkerd, and Submariner. These are well-known services within the Kubernetes ecosystem.
- [REMOTE_CODE_EXECUTION]: The documentation provides instructions to install Kubernetes tools by piping remote scripts to the shell. These scripts originate from established projects (istio.io, linkerd.io, submariner.io) and are documented neutrally as standard installation procedures for these services.
- [COMMAND_EXECUTION]: The skill uses
kubectl,helm, andoperator-sdkto manage clusters, troubleshoot workloads, and deploy manifests. These operations are core to the skill's primary purpose as a Kubernetes specialist tool. - [INDIRECT_PROMPT_INJECTION]: The skill has a surface for processing untrusted data.
- Ingestion points: Reads cluster state and processes user-supplied resource names (e.g., pod names) in
SKILL.mdandreferences/troubleshooting.md. - Boundary markers: Includes a strict "Constraints" section in
SKILL.mddefining mandatory safety practices (e.g., using Secrets, resource limits, and least privilege). - Capability inventory: Full cluster management capabilities via
kubectlandhelmacross all reference files, including executing commands in containers and modifying deployments. - Sanitization: Relies on instructional constraints and safety guidelines rather than programmatic sanitization of inputs.
Audit Metadata