The Agent Skills Directory

[COMMAND_EXECUTION]: The skill manages the execution of Ansible playbooks through the /playbook-executor sub-skill to remediate vulnerabilities on remote systems.
Evidence: Step 5 (Execute Playbook) triggers the execution phase using the Ansible Automation Platform (AAP) integrations.
[REMOTE_CODE_EXECUTION]: The skill workflow involves the dynamic generation of code (Ansible playbooks) by one component and its execution by another.
Evidence: The instructions describe generating a playbook in Step 4 and executing it in Step 5.
Mitigation: The skill requires mandatory human verification of the playbook content and remediation plan before any execution occurs.
[PROMPT_INJECTION]: The skill processes user-supplied data such as CVE IDs and system lists that could potentially contain malicious instructions intended to influence the generated output.
Ingestion points: User-provided CVE identifiers and system context gathered in SKILL.md.
Boundary markers: No explicit delimiter markers are specified in the prompt templates.
Capability inventory: The skill can trigger remote system modifications and file writes via its orchestration of sub-skills.
Sanitization: Employs a dedicated validation step (Step 2) and relies on manual user review of generated content.
[EXTERNAL_DOWNLOADS]: The skill relies on external MCP servers for data retrieval and task execution.
Evidence: The skill requires lightspeed-mcp, aap-mcp-job-management, and aap-mcp-inventory-management servers.
Context: These are trusted vendor-managed resources necessary for the skill's intended operation.

remediation