The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use the vendor-provided wmill CLI tool to perform actions such as script preview, script run, and generate-metadata. These commands are standard for interacting with the Windmill development environment.
[EXTERNAL_DOWNLOADS]: The skill defines a mechanism for managing dependencies (Ansible Galaxy collections and Python packages) within the script header. These dependencies are resolved through the platform's standard metadata generation process.
[INDIRECT_PROMPT_INJECTION]: The skill describes how to ingest user parameters via extra_vars, which are then interpolated into Ansible playbooks. While this constitutes an attack surface typical of Ansible, the skill itself does not contain malicious instructions, and the platform provides mechanisms (like .script.yaml schemas) to define expected input types.
Ingestion points: extra_vars defined in the YAML header of Ansible scripts.
Boundary markers: None explicitly mentioned in the playbook templates.
Capability inventory: The resulting Ansible scripts can execute shell commands, manage files, and perform network operations as part of their intended infrastructure automation role.
Sanitization: The skill encourages providing type information for arguments, which allows the platform to validate inputs.

write-script-ansible