The Agent Skills Directory

[SAFE]: The skill provides instructional content on prompt engineering best practices, emphasizing rational and minimal prompting to improve model performance and token efficiency.
[PROMPT_INJECTION]: The skill uses strong directives such as 'MUST' and 'NEVER' to enforce a workflow where the agent applies these guidelines before editing files. This is an operational procedure for consistency and does not target safety filter bypasses.
[EXTERNAL_DOWNLOADS]: The skill includes links to OpenAI's documentation. Although the text references futuristic model versions, the URLs are to a well-known and trusted technology service domain and do not facilitate malicious activity.

writing-prompt