The Agent Skills Directory

[SAFE]: The skill is entirely text-based and instructional, functioning as a knowledge base for prompt design. It does not request any specific tools in its frontmatter or execute shell commands in its workflow.- [PROMPT_INJECTION]: The skill includes dedicated sections on jailbreak mitigation and harmlessness screening in references/quality_improvement.md, providing defense-in-depth strategies for the prompts it helps generate. It contains no instructions intended to bypass or override the agent's own safety guardrails.- [EXTERNAL_DOWNLOADS]: The skill does not reference or download any external URLs, packages, or remote scripts. All referenced materials are located within the local references/ directory.- [DATA_EXFILTRATION]: No network operations, sensitive file access patterns, or credential harvesting attempts were detected. The skill focuses solely on prompt construction and optimization.

prompt-engineer