The Agent Skills Directory

CREDENTIALS_UNSAFE (HIGH): The skill explicitly accesses sensitive identity and configuration files, including the specific kubeconfig ~/.kube/aks-rg-hypera-cafehyna-hub-config and SSH private keys such as ~/.ssh/id_rsa and ~/.ssh/azure_devops_key within Workflows/RepoManage.md and SKILL.md.\n- COMMAND_EXECUTION (HIGH): The skill facilitates administrative operations on Kubernetes and ArgoCD, including the deletion of applications (argocd app delete), removal of clusters (argocd cluster rm), and modification of RBAC project roles, which could lead to unauthorized resource modification if misused.\n- PROMPT_INJECTION (HIGH): The skill is susceptible to Indirect Prompt Injection. 1. Ingestion points: External data is ingested via argocd app logs and application status manifests in Workflows/Troubleshoot.md. 2. Boundary markers: There are no delimiters or instructions to ignore embedded commands in the processed data. 3. Capability inventory: The skill possesses broad write, delete, and execute capabilities across the managed infrastructure. 4. Sanitization: No sanitization or validation of ingested content is performed before processing or display.\n- PERSISTENCE (HIGH): In SKILL.md, the instructions recommend modifying the user's shell configuration files (~/.bashrc or ~/.zshrc) to add persistent aliases, creating a permanent footprint on the host system.\n- DYNAMIC_EXECUTION (MEDIUM): The workflows in Workflows/AppManage.md and Workflows/AppSetManage.md use cat <<EOF patterns to generate and then apply local YAML manifests, which constitutes runtime configuration generation and execution.

ArgoCdCommand