The Agent Skills Directory

[PROMPT_INJECTION]: The skill uses detailed system prompts and operational principles to define a professional persona. No evidence of override instructions, safety bypasses, or jailbreak patterns (e.g., 'ignore previous instructions') was found.
[DATA_EXFILTRATION]: No hardcoded credentials, sensitive file paths, or network transmission patterns (like curl or fetch) were identified. The author and contact information are provided as standard metadata.
[OBFUSCATION]: All content is provided in plain text. Analysis confirmed the absence of Base64 encoding, zero-width characters, homoglyphs, or other techniques used to hide malicious content.
[REMOTE_CODE_EXECUTION]: The skill contains no scripts, package installations, or remote code download commands. All reference material is hosted locally within the skill's file structure.
[COMMAND_EXECUTION]: There are no shell commands or subprocess calls present in the skill instructions or reference documents.
[SAFE]: The skill follows best practices for a knowledge-based AI agent extension, focusing on high-quality informational content without requiring dangerous capabilities.

regeneron