The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructs users to execute remote scripts directly in the shell without prior inspection, which is a high-risk pattern.
Evidence: curl -Ls https://get.submariner.io | bash in references/multi-cluster.md.
Security Risk: The target URL https://get.submariner.io is currently blacklisted by automated reputation scanners. Executing unverified content from a blacklisted source can lead to full system compromise.
[EXTERNAL_DOWNLOADS]: Multiple references and scripts download configuration files and binaries from external sources.
Fetches ArgoCD installation manifests from raw.githubusercontent.com/argoproj/argo-cd.
Downloads the Istio installation script from istio.io/downloadIstio.
Fetches the Linkerd installation script from run.linkerd.io/install.
Downloads the Sealed Secrets controller from github.com/bitnami-labs/sealed-secrets.
Fetches Kiali and Jaeger addons from raw.githubusercontent.com/istio/istio.
[CREDENTIALS_UNSAFE]: The documentation includes multiple hardcoded sensitive strings in configuration examples.
Evidence: api-key: "sk-1234567890abcdef" and db-password: "MySecurePassword123!" in references/configuration.md.
While these are presented as examples, the presence of common API key prefixes (like sk-) and plaintext passwords can be exploited if accidentally used in production or if parsed by automated tools.
[COMMAND_EXECUTION]: The skill provides numerous commands that perform administrative actions on a Kubernetes cluster.
Various kubectl commands for pod management, service configuration, and cluster debugging across all reference files.
Subprocess execution patterns in references/cost-optimization.md and references/troubleshooting.md for scaling and diagnostics.

kubernetes-specialist