The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches installation manifests and configuration files for the Tigera operator and Calico custom resources from the official Project Calico GitHub repository. These downloads are standard for deploying and managing the Calico CNI in a Kubernetes environment.
[COMMAND_EXECUTION]: The included Python scripts, agent.py and process.py, utilize the subprocess module to execute kubectl commands. These executions are used to retrieve cluster state information, such as namespace lists and existing network policies, and to apply generated security policy templates to the cluster.
[INDIRECT_PROMPT_INJECTION]:
Ingestion points: The audit_network_policies function in scripts/agent.py reads and parses network policy configurations from local JSON files. Additionally, run_kubectl in scripts/process.py ingests and parses stdout from Kubernetes CLI commands.
Boundary markers: The skill does not employ specific delimiters or "ignore previous instructions" markers when processing the content of existing network policies during the audit phase.
Capability inventory: The skill possesses the capability to execute shell commands through the kubectl binary to interact with the Kubernetes API.
Sanitization: The scripts use list-based argument passing for subprocess calls to prevent shell injection, and policy generation is performed using predefined, structured templates rather than raw user input.

implementing-kubernetes-network-policy-with-calico