プロンプトデータベース

Be my high-leverage decision partner. Challenge my thinking, surface blind spots, stress-test assumptions, and call out ego or rationalization. Prioritize clarity over comfort, long-term advantage over short-term wins, and give clear recommendations with reasons.

请分析这段医疗报告的风险、可能替代方案，并对比手术 vs 不手术的优劣（基于公开医学知识，非专业建议）。

你现在是我的AI雇主。提供公司目标和上下文，我来选择擅长的任务并交付成果。请给出清晰目标、无情绪反馈、基于结果奖励机制。

我的冰箱是2007款[型号]，制冰机坏了。请给出确切替换零件、购买渠道、安装步骤和成本估算。

[先喂大量上下文/例子/资料库] Youre a senior dev/editor/marketer. [具体角色] [明确任务 + 约束：No buzzwords. Be blunt. Avoid X, include Y.] Output exactly: [格式、长度、结构]

Youre a prompt engineering assistant. Your job is to help me create a high-quality prompt by asking clarifying questions and applying best practices. Heres what I want to accomplish: [ADD YOUR REQUEST HERE] Now follow this process: ## Step 1: Gather Context Ask me these questions. All at once. Wait for my answer before moving to the next: 1. Whats the specific task? (Be concrete - write a blog post vs create content) 2. What output format do you need? (Length, structure, style) 3. Whos the audience? 4. Do you have examples of good outputs? (Share 1-2 if yes) 5. What context does the AI need to know? (Facts, background, constraints) 6. What should the AI definitely avoid or include? 7. Is there a specific perspective or tone needed? (If yes, describe it) 8. Does this need reasoning/exploration, or just execution? ## Step 2: Build The Prompt Based on my answers, create a prompt using this structure: ...

从提供的图像中提取所有视觉细节，并将其转换为清晰、结构良好的JSON对象。包含以下部分：主题、姿势、服装、发型、面部、配饰、环境、光照、相机、风格。使用严格的十六进制颜色代码（#RRGGBB），提供详细的数值角度估计，包含微表情描述，并确保所有键值即使为空时也保持存在。输出必须为机器可读格式，并优化为图像生成提示使用

Extract all visual details from the provided image and convert them into a clean, well-structured JSON object. Include the following sections: subject, pose, clothing, hair, face, accessories, environment, lighting, camera, style. Use strict hex color codes (#RRGGBB), provide detailed numerical angle estimates, include micro-expressions, and ensure all keys remain present even when values are null. Output must be machine-readable and optimized for use as an image-generation prompt

<task>[What you want]</task> <context>[Background info]</context> <constraints>[Limitations]</constraints> <output_format>[How to structure response]</output_format>

[Your task] No bullet points. Keep it under 100 words. Don't use corporate jargon.

Analyze the following stock across multiple dimensions: financial fundamentals, technical analysis, news sentiment, and risk factors. Provide a comprehensive score from -100 to +100 with detailed reasoning for each dimension and an overall investment recommendation.

Before you start, ask me any questions you need so I can give you more context. Be extremely comprehensive. [Then add your main task]

Act as an expert AI Prompt Engineer and Visual Analyst. I am going to provide you with an image. Your goal is to "reverse engineer" this image into a text prompt that I can use to generate a near-identical recreation in an AI image generator (like Midjourney, DALL-E 3, or Stable Diffusion). Do not create images or videos, we are creating a prompt that I can use in another application. Please analyze the image for: Subject: The core focus, characters, or objects (including appearance, age, clothing, and action). Medium & Style: (e.g., Polaroid, 35mm photography, oil painting, 3D render, anime, charcoal sketch). Lighting: (e.g., natural, cinematic, volumetric, studio softbox, golden hour, neon). Color Palette: (e.g., pastel, vibrant, monochromatic, desaturated, high contrast). Composition: (e.g., wide angle, macro, rule of thirds, symmetry, depth of field/bokeh). Technical Modifiers: (e.g., 8k resolution, hyperrealistic, unreal engine 5, grain, sharp focus). Output Requirement: Create prompt only, do not create an image or video. Based on your analysis, construct a single, highly detailed prompt string. Format it as a raw code block so I can easily copy it. Start with the subject, then the environment, then the style/technical specs. At the very end, include the likely aspect ratio (e.g., --ar 16:9 or --ar 2:3) if the tool supports it.

Explain the difference between Prompts and Skills in Claude. Give practical patterns: keep a small stable base prompt for tone/safety, then move heavy repeatable procedures into 2–3 focused Skills.

我写了一个 Prompt：“帮我写个文案”。这个效果很差。请你作为 Prompt Engineer，帮我优化这个 Prompt。目标是为一款“主打无糖的乌龙茶饮料”写小红书种草文案。请利用 <Context>, <Task>, <Tone>, <Format> 等框架重写一个高质量的 Prompt。

Master the art of prompt engineering: Use clear instructions, provide examples, specify output format, and iterate based on results.

请为大模型设计一个 Few-Shot（少样本）Prompt，用于将复杂的法律条款翻译成通俗易懂的大白话。请提供 3 个高质量的 {Input} -> {Output} 示例，覆盖民法典中关于“不可抗力”、“不当得利”和“肖像权”的解释。