The Agent Skills Directory

[PROMPT_INJECTION]: The skill uses natural instructional language to guide the AI's behavior. There are no attempts to bypass safety filters or override system instructions. The instructions in EVAL.md regarding not revealing the file to the agent are standard evaluation practices.
[DATA_EXPOSURE]: No hardcoded credentials, sensitive file paths, or unauthorized network operations were found. The skill does not access or exfiltrate user data.
[EXTERNAL_DOWNLOADS]: The skill references several external URLs for documentation and verification purposes. These include highly reputable sources such as ARRI (arri.com), the Illuminating Engineering Society (ies.org), the American Society of Cinematographers (theasc.com), Rosco (rosco.com), the Academy of Motion Picture Arts and Sciences (oscars.org), Broncolor (broncolor.us), and the W3C (w3.org). These are documented neutrally as sources of information.
[OBFUSCATION]: No obfuscated content, hidden URLs, zero-width characters, or homoglyph attacks were detected. The content is clear and uses standard Markdown formatting.
[REMOTE_CODE_EXECUTION]: There is no code execution, package installation, or remote script fetching in the provided files. The skill is purely instructional.
[PRIVILEGE_ESCALATION]: No commands or instructions related to privilege escalation or unauthorized system access were found.
[PERSISTENCE]: No persistence mechanisms such as shell profile modifications or scheduled tasks are present.
[INDIRECT_PROMPT_INJECTION]: While the skill processes user intent to generate lighting prompts, it does not have high-risk capabilities like file writing or shell execution that could be exploited via malicious input. The risk surface is standard for text-to-text transformation tasks.
[DYNAMIC_CONTEXT_INJECTION]: The skill does not utilize dynamic context injection (the !command syntax).

lighting-direction