The Agent Skills Directory

[COMMAND_EXECUTION]: The skill does not contain any shell commands, system calls, or dynamic context injection patterns (e.g., backtick-exclamation syntax). All instructions are purely textual for the purpose of guiding an LLM's output.
[DATA_EXFILTRATION]: There are no network operations, API calls, or file system access patterns. The skill operates solely on the input provided by the user to generate a text-based image prompt.
[REMOTE_CODE_EXECUTION]: No external scripts, package installations, or remote code patterns were detected. The skill relies on static templates and instructions contained within the provided files.
[OBFUSCATION]: The content is provided in plain text. No Base64, hex-encoding, zero-width characters, or hidden acrostics were found during the analysis.
[PROMPT_INJECTION]: While the skill uses strong language like 'iron rules' to ensure style consistency, these are standard prompt engineering techniques for creative control and do not attempt to bypass safety filters, extract system prompts, or override agent operational guidelines.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided concepts but lacks the capability to execute code, write files, or perform network requests. Consequently, there is no high-risk surface for indirect prompt injection to exploit.

cyber-terminal-schematic