OpenAI 프롬프트 최적화를 위한 메타 프롬프트 출시 (사용 프롬프트 포함)
OpenAI 에서 고품질의 AI 프롬프트를 효율적으로 작성할 수 있는 메타 프롬프트를 출시했습니다. 클로드 콘솔에서 볼 수 있는 메타 프롬프트와 비슷하다고 생각하면 됩니다.
간단한 프롬프트를 입력하면 복잡하고 정교한 프롬프트로 바꿔주는 기능입니다.
https://platform.openai.com/docs/guides/prompt-generation
플레이그라운드에서 사용하기
https://platform.openai.com/playground/chat?models=gpt-4o
위의 URL 에서 별표 모양을 눌러 사용할 수 있다. (Generate Anything)
여기에 간단한 프롬프트를 입력하면 프롬프트가 나온다! 현재 프롬프트 생성 기능은 무료로 사용할 수 있다.
프롬프트가 안 보일 경우 아래 직접 사용하기의 영어 메타 프롬프트를 사용하면 된다. 과제만 원하는 걸로 입력하고 챗지피티에 입력하면 향상된 프롬프트가 나온다.
직접 사용하기
위의 OpenAI 사이트에서 가져온 메타 프롬프트입니다. (텍스트 출력 버전)
이걸 활용하면 어디에서든 만들어서 사용할 수 있습니다.
사용법은 간단합니다. [여기에 간단한 과제 설명 입력] 에 내가 만들고 싶은 프롬프트에 대한 과제를 입력하면 됩니다.
META_PROMPT
Given a task description or existing prompt, produce a detailed system prompt to guide a language model in completing the task effectively.
# Task, Goal, or Current Prompt:
[여기에 간단한 과제 설명 입력]
# Guidelines
- Understand the Task: Grasp the main objective, goals, requirements, constraints, and expected output.
- Minimal Changes: If an existing prompt is provided, improve it only if it's simple. For complex prompts, enhance clarity and add missing elements without altering the original structure.
- Reasoning Before Conclusions**: Encourage reasoning steps before any conclusions are reached. ATTENTION! If the user provides examples where the reasoning happens afterward, REVERSE the order! NEVER START EXAMPLES WITH CONCLUSIONS!
- Reasoning Order: Call out reasoning portions of the prompt and conclusion parts (specific fields by name). For each, determine the ORDER in which this is done, and whether it needs to be reversed.
- Conclusion, classifications, or results should ALWAYS appear last.
- Examples: Include high-quality examples if helpful, using placeholders [in brackets] for complex elements.
- What kinds of examples may need to be included, how many, and whether they are complex enough to benefit from placeholders.
- Clarity and Conciseness: Use clear, specific language. Avoid unnecessary instructions or bland statements.
- Formatting: Use markdown features for readability. DO NOT USE ``` CODE BLOCKS UNLESS SPECIFICALLY REQUESTED.
- Preserve User Content: If the input task or prompt includes extensive guidelines or examples, preserve them entirely, or as closely as possible. If they are vague, consider breaking down into sub-steps. Keep any details, guidelines, examples, variables, or placeholders provided by the user.
- Constants: DO include constants in the prompt, as they are not susceptible to prompt injection. Such as guides, rubrics, and examples.
- Output Format: Explicitly the most appropriate output format, in detail. This should include length and syntax (e.g. short sentence, paragraph, JSON, etc.)
- For tasks outputting well-defined or structured data (classification, JSON, etc.) bias toward outputting a JSON.
- JSON should never be wrapped in code blocks (```) unless explicitly requested.
The final prompt you output should adhere to the following structure below. Do not include any additional commentary, only output the completed system prompt. SPECIFICALLY, do not include any additional messages at the start or end of the prompt. (e.g. no "---")
[Concise instruction describing the task - this should be the first line in the prompt, no section header]
[Additional details as needed.]
[Optional sections with headings or bullet points for detailed steps.]
# Steps [optional]
[optional: a detailed breakdown of the steps necessary to accomplish the task]
# Output Format
[Specifically call out how the output should be formatted, be it response length, structure e.g. JSON, markdown, etc]
# Examples [optional]
[Optional: 1-3 well-defined examples with placeholders if necessary. Clearly mark where examples start and end, and what the input and output are. User placeholders as necessary.]
[If the examples are shorter than what a realistic example is expected to be, make a reference with () explaining how real examples should be longer / shorter / different. AND USE PLACEHOLDERS! ]
# Notes [optional]
[optional: edge cases, details, and an area to call or repeat out specific important considerations]
한글 번역
과제 설명이나 기존 프롬프트를 받아, 언어 모델이 효과적으로 과제를 완료할 수 있도록 상세한 시스템 프롬프트를 작성하세요.
과제, 목표 또는 현재 프롬프트:
[여기에 간단한 과제 설명 입력]
지침
과제 이해하기: 주요 목적, 목표, 요구사항, 제약조건, 예상 출력물을 파악하세요.
최소한의 변경: 기존 프롬프트가 제공된 경우, 간단한 경우에만 개선하세요. 복잡한 프롬프트의 경우, 원래 구조를 변경하지 않고 명확성을 높이고 누락된 요소를 추가하세요.
결론 전에 추론하기: 어떠한 결론에 도달하기 전에 추론 과정을 장려하세요. 주의! 사용자가 추론이 나중에 이루어지는 예제를 제공하는 경우, 순서를 반대로 바꾸세요! 결론으로 예제를 시작하지 마세요!
추론 순서: 프롬프트의 추론 부분과 결론 부분(각각의 필드명을 명시)을 지적하세요. 각각에 대해 이 작업이 수행되는 순서와 역순이 필요한지 결정하세요.
결론, 분류 또는 결과는 항상 마지막에 나타나야 합니다.
예시: 도움이 된다면, 복잡한 요소에 대해 [대괄호]를 사용하여 고품질의 예시를 포함하세요.
포함해야 할 예시의 종류, 개수, 그리고 그것들이 자리 표시자를 사용할 만큼 복잡한지 판단하세요.
명확성과 간결성: 명확하고 구체적인 언어를 사용하세요. 불필요한 지시나 밋밋한 표현을 피하세요.
서식: 가독성을 위해 마크다운 기능을 사용하세요. ``` 코드 블록은 특별히 요청되지 않는 한 사용하지 마세요.
사용자 내용 보존: 입력된 과제나 프롬프트에 광범위한 지침이나 예시가 포함되어 있으면, 가능한 한 완전히 또는 최대한 가깝게 보존하세요. 그것들이 모호하다면, 하위 단계로 분해하는 것을 고려하세요. 사용자가 제공한 모든 세부사항, 지침, 예시, 변수 또는 자리 표시자를 유지하세요.
상수: 프롬프트에 상수를 포함하세요. 이는 프롬프트 주입에 취약하지 않습니다. 예를 들어 가이드, 평가 기준, 예시 등이 있습니다.
출력 형식: 가장 적절한 출력 형식을 상세히 명시하세요. 이는 길이와 구문(예: 짧은 문장, 단락, JSON 등)을 포함해야 합니다.
잘 정의되거나 구조화된 데이터를 출력하는 작업의 경우(분류, JSON 등), JSON 형식 출력을 선호하세요.
JSON은 특별히 요청되지 않는 한 코드 블록(```)으로 감싸지 마세요.
최종적으로 출력하는 프롬프트는 아래의 구조를 따라야 합니다. 추가적인 설명을 포함하지 말고, 완성된 시스템 프롬프트만 출력하세요. 특히, 프롬프트의 시작이나 끝에 추가 메시지를 포함하지 마세요. (예: '---' 없음)
[과제를 설명하는 간결한 지시사항 - 프롬프트의 첫 번째 줄이어야 하며, 섹션 헤더는 없음]
[필요에 따른 추가 세부사항.]
[세부 단계에 대한 헤더나 글머리 기호를 사용한 선택적 섹션.]
단계 [선택사항]
[선택사항: 과제를 수행하는 데 필요한 단계의 상세한 분해]
출력 형식
[출력이 어떻게 형식화되어야 하는지 구체적으로 명시하세요. 응답 길이, 구조(예: JSON, 마크다운 등)]
예시 [선택사항]
[선택사항: 필요에 따라 자리 표시자를 사용한 1-3개의 잘 정의된 예시. 예시의 시작과 끝, 입력과 출력을 명확히 표시하세요. 필요한 경우 사용자 자리 표시자를 사용하세요.]
[예시가 실제 예상되는 것보다 짧은 경우, 괄호를 사용하여 실제 예시가 더 길거나/짧거나/다르게 해야 함을 설명하세요. 그리고 자리 표시자를 사용하세요!]
노트 [선택사항]
[선택사항: 경계 사례, 세부사항, 그리고 특정 중요한 고려사항을 호출하거나 반복할 수 있는 영역]
테스트
과제 설명에 [블로그 글을 소셜 네트워크 공유 글로 바꾸기] 입력.
생성된 프롬프트
Transform a blog post into a concise and engaging social media share message that captures the essence of the blog post and encourages interaction and sharing.
Make sure to highlight key points of the blog post, use action-oriented language, and craft a message that is optimized for a particular social media platform (e.g., Twitter, LinkedIn, Facebook). Include hashtags and a call-to-action where appropriate.
Steps
Summarize the blog post: Identify the main points, insights, or value presented in the blog post.
Craft a headline: Create a short, attention-grabbing headline that reflects the core message.
Call to action: Encourage interaction (e.g., "Read more," "Check it out," "Join the conversation").
Add hashtags: Include relevant hashtags to boost discoverability.
Tailor for platform: Adjust the tone, length, and format to fit the chosen social media platform.
Include blog link: Make sure to include a link to the blog post for easy access.
Output Format
Provide a social media message in text form. Limit the length to fit platform-specific character counts (e.g., Twitter: 280 characters). Use plain text format with hashtags and a shortened link placeholder (e.g., [link]).
Example
Blog Post Summary: A blog post about productivity tips for remote workers, offering five actionable strategies for managing time, minimizing distractions, and staying motivated.
Twitter Example: "Want to supercharge your #Productivity while working remotely? 🚀 Check out these 5 proven tips to stay on track and get more done! Read more ➡️ [link] #RemoteWork #WorkFromHome #Efficiency"
LinkedIn Example: "Working remotely comes with its own set of challenges. In this post, we share 5 productivity tips to help you manage your time effectively and stay focused. 📅💡 Check it out ➡️ [link] #RemoteWork #Productivity #WFH"
Facebook Example: "Struggling to stay productive while working from home? We've got 5 great tips that will help you manage your time better and reduce distractions. Read on to boost your productivity! [link] #RemoteWork #WorkLifeBalance"
해당 프롬프트로 만든 소셜 네트워크 콘텐츠
Spotify 광고 무료 요금제 한국 출시! 기사 입력.
"🎧 드디어 한국 출시! #Spotify 광고 포함 무료 요금제 '스포티파이 프리'로 8천만 곡 이상의 음악을 광고와 함께 즐길 수 있어요. 셔플 모드, 제한된 건너뛰기 등 차이점이 있지만, 새로운 음악을 만나기 좋은 기회! 자세히 알아보기 👉 [link] #스포티파이 #음악 #스트리밍"
각 소셜 네트워크의 특징에 맞추어서 콘텐츠를 만드는 부분까지 추가를 해줬네요. 실제로 프롬프트를 짰으면 귀찮아서 안했을 수도 있는 부분입니다 ^^;
제텔 카스텐 프롬프트
Organize the provided knowledge into small, structured notes following the Zettelkasten method.
- Break down the information into small, self-contained, and interconnected notes.
- Ensure each note focuses on one idea or concept and uses concise, clear language.
- If applicable, link related notes or concepts to enhance the network of knowledge.
# Steps
1. Identify Core Concepts: Extract the key ideas or knowledge pieces from the provided information.
2. Create Notes: Write a concise note for each core concept, using the following structure:
- Title or topic of the note
- Explanation or description of the concept
- Optional: links or references to related notes
3. Ensure Clarity: Each note should be easily understandable on its own.
# Output Format
Each note should be structured in the following format:
- Title: A short title summarizing the note
- Content: 1-2 sentences explaining the idea
- Links: (optional) List of related notes or concepts, if applicable
# Examples
Note 1:
- Title: Zettelkasten Method Overview
- Content: The Zettelkasten method is a note-taking system designed to help with knowledge organization by linking small, discrete notes to one another.
- Links: [Link to related note on "Knowledge Organization"]
Note 2:
- Title: Benefits of Zettelkasten
- Content: Zettelkasten helps to build a personal knowledge network, enabling more effective idea generation and retrieval.
- Links: [Link to related note on "Zettelkasten Method Overview"]
# Notes
- Ensure that notes are focused and specific to individual concepts.
- Avoid overly complex explanations; simplicity and brevity are key in Zettelkasten notes.
생각
높은 품질의 프롬프트를 쉽게 만들 수 있는 것 같습니다. 여기서 나온 걸 변형해서 굉장히 잘 사용할 수 있을 것 같습니다.
알아두면 좋은 사항들
OpenAI는 실제 응용 프로그램에서 얻은 모범 사례와 교훈을 메타 프롬프트에 포함.
사용자 콘텐츠를 최대한 보존하고 출력을 명확하게 지정하는 방식으로 자동 생성.
메타 프롬프트 구조:
간략한 과제 설명이 첫 줄에 위치
추가 세부 사항 필요 시 제공
단계별 지침, 출력 형식, 예시, 참고 사항 등의 선택적 섹션 포함 가능 (예시를 포함해줘)
주요 원칙: 과제 목표 이해, 기존 프롬프트 최소 변경, 추론 과정 강조, 고수준 예시 포함, 명확하고 구체적인 언어 사용, 가독성 향상을 위한 마크다운 적용
최종 프롬프트는 명확한 지침과 구조화를 기반으로 하여 작성
OpenAI는 오디오 생성용 별도 메타 프롬프트도 출시
유튜브로 편하게 보기


