The Agent Skills Directory

[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it processes untrusted data from web search results and user-provided requirements to generate executable Ansible playbooks and roles.
Ingestion points: User requirements and external documentation fetched via WebSearch or MCP tools as described in the 'Handling Custom Modules and Collections' section of SKILL.md.
Boundary markers: The instructions do not define explicit delimiters or warnings to prevent the LLM from obeying instructions embedded within the untrusted search results.
Capability inventory: Generated code has high capabilities, including arbitrary command execution, file system modification, and network requests via Ansible modules.
Sanitization: No specific sanitization logic is described for the input data, though the skill mandates a validation step using 'devops-skills:ansible-validator' on the generated output.
[COMMAND_EXECUTION]: The skill is designed to generate tasks that execute commands and shell scripts on target systems.
Evidence: Multiple templates (e.g., assets/templates/role/tasks/main.yml) and reference guides (references/module-patterns.md) provide examples for using ansible.builtin.command and ansible.builtin.shell.
Controls: The skill enforces security best practices, including the use of idempotency guards (creates, removes) and appropriate task naming to ensure transparency.
[EXTERNAL_DOWNLOADS]: The skill facilitates the download and installation of external Ansible collections and roles.
Evidence: The assets/templates/project/requirements.yml file provides templates for fetching dependencies from well-known services like Ansible Galaxy and GitHub.
Context: These references target established repositories and are a standard part of the Ansible ecosystem.
[CREDENTIALS_UNSAFE]: Templates provided within the skill reference sensitive system paths commonly associated with credentials.
Evidence: assets/templates/inventory/group_vars/all.yml includes a default reference to ~/.ssh/id_rsa, and various role templates include placeholders for SSL private keys.
Context: These are used as configuration placeholders for the user's local environment rather than being hardcoded secrets or unauthorized access attempts by the skill itself.

ansible-generator