The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides several automation scripts and playbooks that execute shell commands on the local machine and remote servers.
Ansible playbooks in references/best-practices.md use apt, service, and command modules to manage packages and system services.
The Bun deployment script in references/deployment-automation.md uses shell execution ($) to run kubectl and docker commands for building and deploying container images.
NGINX Ingress configuration examples include snippets for rate limiting and caching logic.
[EXTERNAL_DOWNLOADS]: The skill recommends installing several management tools from well-known and trusted sources.
Instructions include using Homebrew to install ansible, sshuttle, k9s, and kubectx on macOS.
The GitHub Actions workflow in references/deployment-automation.md fetches actions from trusted organizations including actions, azure, oven-sh, and docker.
Recommends the installation of krew and several community plugins for kubectl (neat, node-shell, popeye, tree, ingress-nginx).
[INDIRECT_PROMPT_INJECTION]: The skill includes an automation script and CI/CD workflow that process external data, presenting a potential injection surface.
Ingestion points: The Bun script reads live Kubernetes Deployment manifests via kubectl get, and the GitHub Actions workflow accepts manual inputs via workflow_dispatch (references/deployment-automation.md).
Boundary markers: The skill explicitly includes security warnings advising users to assign GHA inputs to environment variables rather than direct interpolation to prevent expression injection.
Capability inventory: The automation scripts have the capability to write files to disk, execute shell commands, and interact with the Kubernetes API and Docker daemon.
Sanitization: The deployment script uses a YAML parser (parseDocument) for programmatic patching rather than text-based manipulation, reducing the risk of schema confusion.

k8s-operations