The Agent Skills Directory

[SAFE]: The skill provides documentation and templates for Kubernetes manifest generation. It does not execute arbitrary code or scripts in the agent's runtime environment. Any shell scripts provided (e.g., init.sh and healthcheck.sh in assets/configmap-template.yaml) are intended for deployment within user-managed Kubernetes clusters, not for execution during the skill's operation.\n- [SAFE]: External URLs refer to the official Kubernetes documentation site (kubernetes.io), which is a well-known and trusted source for technical specifications. These references are handled neutrally as technical resources.\n- [SAFE]: The skill promotes secure configuration practices, explicitly recommending the use of Pod Security Standards such as runAsNonRoot: true, allowPrivilegeEscalation: false, and readOnlyRootFilesystem: true in its templates and best practices guide.\n- [PROMPT_INJECTION]: The skill facilitates the generation of Kubernetes manifests based on user input, which presents a surface for indirect prompt injection.\n
Ingestion points: User requirements for workload configuration (e.g., application name, image, and environment variables) provided during the workflow described in resources/implementation-playbook.md.\n
Boundary markers: Absent; user inputs are interpolated into templates without specific delimiters or isolation instructions.\n
Capability inventory: The skill is limited to text generation and does not utilize dangerous capabilities like subprocess calls, file-system writes, or network-post operations during its processing phase.\n
Sanitization: Absent; the skill relies on the user or downstream validation tools (like kubectl --dry-run or kube-score) mentioned in the 'Validate and Test' section to verify the generated YAML.

k8s-manifest-generator