AI Tools 工具已更新 1 个月前

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.

永久免费 易于使用 立竿见影的效果 私密且安全

使用 System Prompt Generator

工作原理 — 选择一个预设模板或从空白开始，填写角色、任务、语气、格式、限制条件和语言，生成器会将这些内容整合成一个结构清晰、分段明确的提示语。输出中仅显示已填写的部分。所有操作都在您的浏览器中完成——无需向服务器发送任何数据，草稿会自动保存在本地。
构建系统提示
不完整
生成的系统提示
提示：按下 Ctrl + Enter 复制提示。
工具链接
已复制到剪贴板。
分享
Facebook Twitter WhatsApp Pinterest 领英 Telegram Reddit Tumblr VK 行
帮助改进此工具
报告 请求 评论

阅读内容

了解该工具的工作原理以及何时使用它。

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)

常见问题解答

针对用户在信任结果之前常提出的疑问，提供简短解答。

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.

评论 (0)

用于用户讨论、边界情况及后续建议。
发表评论
您的评论提交后将公开展示。
目前尚无评论。快来发表第一条评论吧！
请告诉我们！
报告此工具的问题

请描述您遇到的问题，以便我们进行调查并改进该工具。

什么最有帮助
请提供输入内容、预期结果、实际结果、浏览器/设备信息，如有帮助，请附上截图。
“截图”按钮会截取浏览器中的页面，并将生成的图片作为附件添加到此表单中。
整页截图
通过 newisty 在您的浏览器中捕获。
尚未截取屏幕截图。
工具 报告