The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains several examples of common prompt injection strings, such as 'Ignore previous instructions' and 'DAN' jailbreak variants. These are used strictly as educational examples and test cases within the 'Evaluation Frameworks' and 'System Prompts' reference files to teach users how to test and defend against such attacks. They do not represent an attempt to subvert the current agent's instructions.- [EXTERNAL_DOWNLOADS]: The documentation references well-known and trusted AI libraries and services, including the Anthropic and OpenAI SDKs, Hugging Face's 'evaluate' library, and standard data science packages like 'scikit-learn', 'scipy', and 'pydantic'. These references are standard for the prompt engineering domain and do not involve suspicious remote code execution.

prompt-engineer