The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-defined playbooks, inventories, and templates which represents a vulnerability surface for indirect prompt injection. The documentation provides mitigation strategies such as boundary markers and testing frameworks like Molecule to validate configurations. Evidence found in references/playbook-patterns.md and references/testing-guide.md.
[COMMAND_EXECUTION]: The skill includes instructions for executing arbitrary shell commands on remote systems using Ansible modules like command and shell. The guide mitigates risks by recommending state-based modules and idempotency controls. Evidence found in references/idempotency-guide.md and references/playbook-patterns.md.
[PRIVILEGE_ESCALATION]: The skill utilizes privilege escalation mechanisms (become: yes, sudo) to manage system-level configurations on target hosts. This is the primary intended functionality of the configuration management skill. Evidence found in SKILL.md and references/role-structure.md.
[EXTERNAL_DOWNLOADS]: The skill documents the use of ansible-galaxy and pip to download and install dependencies from official registries. These operations target well-known services and organizations. Evidence found in SKILL.md and references/testing-guide.md.
[CREDENTIALS_UNSAFE]: The documentation references sensitive file paths such as ~/.ssh/id_rsa and ~/.aws/credentials as part of standard configuration workflows. It correctly promotes the use of ansible-vault and HashiCorp Vault for managing actual secrets. Evidence found in SKILL.md and references/secrets-management.md.

managing-configuration