prompt-engineering

Advanced prompt engineering techniques to maximize LLM performance, reliability, and controllability.

Teach the model by showing examples instead of explaining rules. Include 2-5 input-output pairs that demonstrate the desired behavior. Use when you need consistent formatting, specific reasoning patterns, or handling of edge cases. More examples improve accuracy but consume tokens—balance based on task complexity.

Request step-by-step reasoning before the final answer. Add "Let's think step by step" (zero-shot) or include example reasoning traces (few-shot). Use for complex problems requiring multi-step logic, mathematical reasoning, or when you need to verify the model's thought process. Improves accuracy on analytical tasks by 30-50%.

دليل الخبراء حول الأنماط الهندسية السريعة وأفضل الممارسات وتقنيات التحسين. يُستخدم عندما يريد المستخدم تحسين المطالبات، أو تعلم إستراتيجيات المطالبة، أو تصحيح سلوك الوكيل. المصدر: sebas-aikon-intelligence/antigravity-awesome-skills.