The Agent Skills Directory

[PROMPT_INJECTION]: The skill demonstrates an indirect prompt injection surface by interpolating user-controlled variables directly into Ansible tasks and shell commands without explicit sanitization.
Ingestion points: User-defined variables such as vm_name, cluster_name, and ceph_network in SKILL.md and reference files.
Boundary markers: Absent; variables are interpolated directly into command strings (e.g., pvecm create {{ cluster_name }}).
Capability inventory: The skill utilizes ansible.builtin.command, ansible.builtin.shell, and privileged Proxmox modules with root-level access to the virtualization infrastructure.
Sanitization: No input validation or escaping logic is implemented for the provided variables.
[COMMAND_EXECUTION]: The skill provides numerous examples of executing system-level commands (e.g., pvecm, pveceph, wipefs, ip link) to manage Proxmox clusters and storage. While these are necessary for the skill's primary purpose, they require high privileges (root/sudo).
[EXTERNAL_DOWNLOADS]: The skill recommends installing the community.proxmox collection from Ansible Galaxy using ansible-galaxy. This is a standard operation targeting a well-known, trusted repository.
[SAFE]: Sensitive information such as API tokens and passwords are managed securely using the infisical lookup plugin and the no_log: true parameter, preventing credential exposure in execution logs.

ansible-proxmox