The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions involve executing local management scripts (e.g., ./deploy) and standard Ansible tooling such as ansible-playbook and ansible-lint for automation tasks.
[PRIVILEGE_ESCALATION]: The skill describes operations requiring elevated permissions, specifically the use of Ansible's become module for task execution and the modification of SSH authorized_keys to manage deployment access.
[INDIRECT_PROMPT_INJECTION]: The skill acts on external project files which provides a surface for processing untrusted instructions.
Ingestion points: Project-provided playbooks, roles, inventories, and configuration files (ansible.cfg, requirements.yml) are read and processed by the agent.
Boundary markers: The skill mandates pre-execution validation using ansible-playbook --syntax-check and ansible-lint to verify file integrity and compliance.
Capability inventory: The agent can execute local management scripts and modify remote host states through Ansible modules.
Sanitization: The instructions emphasize the use of no_log: true for sensitive tasks and the enforcement of secret encryption via Ansible Vault.

specsfy-specialist-ansible