The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/agent.py script executes calicoctl and kubectl using subprocess.run with arguments passed as a list and without the shell=True parameter. This implementation follows standard security best practices to prevent command injection.
[INDIRECT_PROMPT_INJECTION]: The agent ingests external data from the Kubernetes environment (such as namespace and pod names) to perform connectivity tests via kubectl exec. This represents a potential injection surface, but the risk is effectively mitigated by the script's use of structured argument passing and its execution of specific, limited commands.
[EXTERNAL_DOWNLOADS]: The skill references the installation of the kubernetes Python library and the calicoctl CLI tool from official, trusted sources (kubernetes.io and tigera.io). These are well-known resources required for the skill's documented functionality.

implementing-container-network-policies-with-calico