The Agent Skills Directory

[SAFE]: The skill is composed of purely instructional Markdown files with no executable code or remote script calls.
[PROMPT_INJECTION]: The military role-play is a persona-setting mechanism and does not contain directives to bypass safety protocols or extract system instructions.
[INDIRECT_PROMPT_INJECTION]: The skill creates an attack surface by processing external logs and technical data during its investigative phases. Ingestion points: Technical reports and system logs used during '深入侦察' (In-depth Reconnaissance). Boundary markers: The provided reporting templates lack explicit delimiters for untrusted content. Capability inventory: Diagnostic tools like kubectl, docker, and curl are mentioned as investigative resources. Sanitization: No input validation or sanitization logic is specified in the files.

military-scout