The Agent Skills Directory

[Prompt Injection] (SAFE): The skill contains many persona-based prompts (e.g., 'Act as an expert software developer'). These are standard role-playing techniques for LLMs and do not contain instructions to bypass safety filters, ignore system prompts, or extract internal instructions.
[Indirect Prompt Injection] (SAFE): The prompt templates include placeholders for user data (e.g., [CONCEPT], [PRODUCT], [SOURCE]). While these represent a surface for indirect injection, the skill is a static documentation file with no underlying code that executes or processes this data in a privileged way.
[External Downloads] (SAFE): The skill references trusted external resources like github.com/f/awesome-chatgpt-prompts for informational purposes. No automated downloads or script executions are performed.
[Data Exposure & Exfiltration] (SAFE): No hardcoded credentials, sensitive file paths, or network exfiltration patterns were found.
[Remote Code Execution] (SAFE): The skill consists entirely of markdown text. There are no functions, scripts, or commands that could lead to code execution.

prompt-library