AI Tools 도구 업데이트됨 1개월 전

System Prompt Generator

Build a structured system prompt for any LLM with ready-made presets. Pick a starting point (Coding assistant, Writing coach, Data analyst, and more), fill the role, task, tone, output format, constraints, and language, and watch a clean, sectioned prompt assemble live. Only filled sections appear in the output. Everything runs in your browser — nothing is sent to a server, and your draft is auto-saved locally.

항상 무료 사용하기 쉬움 즉각적인 결과 비공개이며 안전합니다

사용법 System Prompt Generator

작동 원리 — 사전 설정된 템플릿을 선택하거나 빈 템플릿으로 시작하여 역할, 과제, 어조, 형식, 제약 조건 및 언어를 입력하면, 생성기가 이를 깔끔하게 구획된 프롬프트로 조합해 줍니다. 입력된 부분만 출력에 반영됩니다. 모든 과정은 브라우저 내에서 진행되며, 서버로 전송되는 데이터는 없으며, 초안은 로컬에 자동 저장됩니다.
시스템 프롬프트 생성
미완성
생성된 시스템 프롬프트
팁: 누르세요 Ctrl + Enter 프롬프트를 복사하려면.
도구 링크
클립보드에 복사되었습니다.
이 도구를 개선하는 데 도움을 주세요

내용 읽기

이 도구의 작동 방식과 사용 시기를 알아보세요.

How to Use This Tool

System prompts tell an LLM who it is, what it should do, and how it should behave. This generator assembles those pieces into a clean, sectioned prompt that works with ChatGPT, Claude, Gemini, and most other chat or API interfaces.

1. Pick a starting point

Click a preset — Coding assistant, Writing coach, Data analyst, Marketing copywriter, Customer support agent, SEO specialist, Translator, Academic researcher, Sales email writer, Resume & interview coach, or Productivity coach — or start Blank and describe everything yourself. Presets pre-fill every field so you can jump straight to tweaking.

2. Fill the core fields

The Role and Task description are required; everything else is optional and skips the output when empty. Add an Audience to tune who the model is talking to, and pick an expertise level to set the depth of the answer.

3. Set tone, format, and language

Tap a Tone chip (Professional, Friendly, Casual, Authoritative, Concise, or Empathetic) and an Output format chip (Markdown, Plain text, Bulleted list, JSON, or Code-friendly), then choose the Language the model should respond in.

4. Add constraints and behaviours

Type Constraints one per line — each line becomes a bullet in the generated prompt. Toggle Behaviour rules such as asking a clarifying question when the request is ambiguous, saying "I do not know" instead of guessing, reasoning step by step, or flagging security issues.

5. Copy or download

The output updates live and only includes the sections you filled. Use Copy prompt, Copy as JSON, or Download .txt, or press Ctrl+Enter to copy. Paste into task reads your clipboard straight into the task field.

Features

  • Eleven ready-made presets plus a blank start
  • Live sectioned output that skips empty fields
  • One-per-line constraints rendered as bullets
  • Behaviour toggles for reliable LLM responses
  • Character, word, token, and section counters
  • Copy prompt, copy as JSON, and download as .txt
  • Ctrl/Cmd+Enter keyboard shortcut to copy
  • Draft auto-save in the browser (localStorage)

자주 묻는 질문

사용자가 결과를 신뢰하기 전에 자주 묻는 질문에 대한 간결한 답변.

A system prompt is the opening instruction you give an LLM before your actual request. It sets the role, task, tone, format, and rules for the response, which makes output far more consistent than asking the model to guess.

The output is a markdown-style document with a heading per section: # Role, # Task, # Audience, # Expertise level, # Tone, # Constraints, # Output format, # Language, and # Behaviour. Empty sections are omitted, so the result stays short and focused.

No. Everything runs in your browser. Your draft is saved only in your browser's local storage and is never transmitted anywhere.

The generated prompt is plain text with simple headings, so it works with ChatGPT, Claude, Gemini, Grok, local models, and any chat or API interface that accepts a system message or an opening instruction.

Sections are only rendered when they have content. For example, the Tone section disappears until you pick a tone, and the Constraints section appears only when you type at least one constraint.

Yes. Presets, tones, formats, languages, and behaviour rules are defined in config/system-prompt-generator.php. The tool reads everything from that single config file.

댓글 (0)

사용자 토론, 예외적인 사례 및 추가 팁을 위한 공간입니다.
댓글 남기기
댓글을 제출하면 공개적으로 표시됩니다.
아직 댓글이 없습니다. 가장 먼저 댓글을 남겨보세요!
저희에게 알려주세요!
이 도구와 관련된 문제를 신고하기

발생한 문제를 자세히 설명해 주시면, 저희가 이를 조사하고 도구를 개선할 수 있습니다.

가장 도움이 되는 것은 무엇인가
입력 내용, 예상 결과, 실제 결과, 브라우저/기기 정보를 기재해 주시고, 도움이 될 경우 스크린샷도 첨부해 주세요.
스크린샷 버튼은 브라우저에 표시된 페이지를 캡처하여 생성된 이미지를 이 양식에 첨부합니다.
전체 화면 스크린샷
newisty를 통해 브라우저에서 캡처되었습니다.
아직 스크린샷을 찍지 않았습니다.
도구 보고서