12 Practical AI Prompt Templates for Everyday Work (with Examples)
This guide presents twelve ready‑to‑use AI prompt templates covering single‑task queries, business writing, multi‑step projects, creative branding, logical reasoning, structured outputs, code editing, autonomous agents, image generation, and more, each illustrated with concrete examples.
Hello, this is the AI Open‑Source Efficiency Guide sharing twelve commonly used prompt templates for work, each with a concrete example.
1. Simple Single‑Task Prompt
Role, Task, Format – suitable for clear, quick single‑step requests.
角色:[用一句话定义 AI 的身份]
任务:[精确的动词 + 要生成的内容]
格式:[确切的输出格式和长度]Example:
Role: You are a senior technical writer. Task: Write a brief description of what a REST API is. Format: Plain text, up to three sentences, no jargon, suitable for a non‑technical audience.
2. Professional Documents & Business Writing
Context, Objective, Style, Tone, Audience, Response – for full‑background control of reports, marketing copy, etc.
上下文:[AI 理解当前情况所需的背景信息]
目标:[确切的目标 —— 成功的标准是什么]
风格:[写作风格:正式 / 对话式 / 技术性 / 叙事性]
语气:[情感基调:权威 / 共情 / 紧急 / 中立]
受众:[谁会阅读此内容 —— 他们的知识水平和期望]
响应:[输出的格式、长度和结构]Example:
Context: I am the founder of a B2B SaaS tool that automates expense reimbursement for mid‑size companies. Objective: Write a cold email that gets a reply from the CFO. Style: Direct and conversational, without a salesy feel. Tone: Confident but not pushy. Audience: CFO of a 200‑person company, busy and skeptical of vendor emails. Response: No more than five sentences, include a subject line, avoid bullet points.
3. Complex Multi‑Step Projects
Role, Instructions, Steps, End Goal, Narrowing – for projects requiring clear action sequences.
角色:[AI 应采用的专家身份]
指令:[用平实语言描述的总任务]
步骤:
1. [第一步行动]
2. [第二步行动]
3. [根据需要继续]
最终目标:[最终输出必须达到的效果]
范围缩小:[约束条件、范围限制、需要排除的内容]Example:
Role: You are a product manager with 10 years of mobile app experience. Instruction: Write a product requirement document (PRD) for a habit‑tracking feature. Steps: Define the problem statement in one paragraph. List user stories in the format “As a [user], I want [goal] so that [reason]”. Define acceptance criteria for each story. Explicitly list out‑of‑scope items. End Goal: A PRD that the engineering team can immediately use for sprint planning. Narrowing: No technical implementation details, no mockups, total length under 600 words.
4. Creative Work & Brand Voice
Capacity, Role, Insight, Statement, Personality, Experiment – for branding, tone, and iterative creativity.
能力:[AI 应具备的能力或专业知识]
角色:[要采用的具体人格]
洞察:[塑造响应的关键背景洞察]
陈述:[核心任务或问题]
个性:[语气和风格 —— 幽默 / 权威 / 随性 / 犀利]
实验:[请求变体或替代方案进行探索]Example:
Capacity: Senior copywriter experienced in SaaS product launches. Role: Brand spokesperson for a developer productivity tool. Insight: Developers dislike marketing fluff and prefer honest, concrete language. Statement: Write a headline and sub‑headline for a landing page. Personality: Sharp, cold, confident – no adjectives, no exclamation marks. Experiment: Provide three variants ranging from minimalistic to bold.
5. Logic, Math, Analysis, Debugging
For tasks that require careful reasoning before answering.
Important tip: Use chain‑of‑thought (CoT) only with standard reasoning models (Claude, GPT‑4o, Gemini). Do not add CoT to models that already perform internal reasoning (o1, o3, Claude with “deep thinking”).
[任务陈述]
在回答之前,请仔细思考以下过程:
<thinking>
1. 被询问的实际问题是什么?
2. 解决方案必须遵守哪些约束?
3. 有哪些可能的处理方法?
4. 哪种方法最好,为什么?
</thinking>
请仅在 <answer> 标签中给出最终答案。6. Consistent Structured Output & Pattern Replication
When the desired format is easier to show by example than by description.
[任务指令]
以下是所需确切格式的示例:
<examples>
<example>
<input>[示例输入 1]</input>
<output>[示例输出 1]</output>
</example>
<example>
<input>[示例输入 2]</input>
<output>[示例输出 2]</output>
</example>
</examples>
现在将此确切模式应用于:[实际输入]Rules:
2 to 5 examples is the sweet spot.
Examples must include edge cases, not just simple cases.
Wrap examples with XML‑style tags.
7. Code‑Editing AIs (Cursor, Windsurf, Copilot, etc.)
Prevents common failures such as editing the wrong file or breaking existing logic.
文件:[确切路径/文件名.ext]
函数/组件:[确切名称]
当前行为:
[代码目前的功能 —— 请具体描述]
预期变更:
[编辑后应实现的功能 —— 请具体描述]
范围:
仅修改 [函数 / 组件 / 章节]。
不要触动:[列出所有保持不变的内容]
约束:
- 语言/框架:[指定版本]
- 不要添加不在 [package.json / requirements.txt] 中的依赖
- 保留现有的 [类型签名 / API 契约 / 变量命名]
完成标准:
[确认更改已正确生效的确切条件]8. Autonomous Agents (Claude Code, Devin, AutoGPT, …)
Stops runaway loops and scope explosion by requiring explicit stop conditions.
目标:
[用一句话描述单一、明确的目标]
起始状态:
[当前文件结构 / 代码库状态 / 环境]
目标状态:
[代理完成时应存在的内容]
允许的操作:
- [代理可以采取的具体操作]
- 仅安装 [requirements.txt / package.json] 中列出的包
禁止的操作:
- 不要修改 [目录/范围] 之外的文件
- 不要运行开发服务器或部署
- 不要推送到 git
- 未显示 diff 前不要删除文件
- 未经人工批准不要做架构决策
停止条件:
在以下情况请暂停并请求人工复核:
- 文件将被永久删除
- 需要集成新的外部服务或 API
- 存在两条有效实现路径且选择会影响架构
- 错误尝试 2 次仍无法解决
- 任务需要修改声明范围之外的内容
检查点:
在每个主要步骤后,输出:✅ [已完成内容]
最后,输出所有更改文件的完整摘要。9. Image & Video Generation (Midjourney, DALL‑E, Stable Diffusion, Sora, …)
Prompt structure for visual generation tools.
主体 (Subject):[主要主体 —— 具体,不要模糊]
动作/姿态 (Action/Pose):[主体正在做什么]
场景 (Setting):[场景发生的地点]
风格 (Style):[写实 / 电影感 / 动漫 / 油画 / 矢量等]
情绪 (Mood):[戏剧性 / 宁静 / 诡异 / 欢乐等]
光照 (Lighting):[黄金时刻 / 影棚光 / 霓虹灯 / 阴天 / 烛光等]
配色方案 (Color Palette):[主导颜色或命名色板]
构图 (Composition):[全景 / 特写 / 航拍 / 荷兰式倾斜等]
宽高比 (Aspect Ratio):[16:9 / 1:1 / 9:16 / 4:3]
负面提示词 (Negative Prompts):[模糊、水印、多余手指、畸变、低质量]
风格参考 (Style Reference):[艺术家 / 电影 / 审美参考(如果适用)]10. Image Editing Based on Reference Image
Use when the user provides an existing image and only wants modifications.
Before writing the prompt, tell the user: “Attach your reference image to the tool before sending this prompt.”
参考图:[已附加 / URL]
保持完全一致的内容:[列出所有不得更改的内容]
需要更改的内容:[具体的编辑内容 —— 请精确]
更改幅度:[轻微 / 中等 / 显著]
风格一致性:保持参考图的确切风格、光照和情绪
负面提示词:[要避免引入的内容]11. ComfyUI Node‑Based Image Workflow
Prompt format for ComfyUI, separating positive and negative prompts.
正面提示词 (POSITIVE PROMPT):
[主体], [风格], [情绪], [光照], [构图], [质量增强词: highly detailed, sharp focus, 8k]
负面提示词 (NEGATIVE PROMPT):
[需要排除的内容: blurry, low quality, watermark, extra limbs, bad anatomy, distorted, oversaturated]
检查点 (CHECKPOINT):[模型名称]
采样器 (SAMPLER):Euler a (推荐起点)
CFG SCALE:7
步数 (STEPS):20-30
分辨率 (RESOLUTION):[宽 x 高 —— 必须能被 64 整除]12. Decompose, Adapt, or Split Existing Prompts
When a user wants to break down, adapt, or simplify an existing prompt.
Decomposition Output Format:
原始 Prompt:[粘贴内容]
结构分析:
- 角色/身份:[分配了什么角色及其原因]
- 任务:[请求了什么行动]
- 约束:[设置了哪些限制]
- 格式:[预期的输出形状]
- 弱点:[缺失了什么或可能导致错误输出的地方]
推荐修复:[填补空白后的重写版本]Adaptation Output Format:
原始 ([来源工具]):[原始 Prompt]
适配于 [目标工具]:
[使用目标工具语法和最佳实践重写的 Prompt]
关键更改:
- [更改 1 及其原因]
- [更改 2 及其原因]AI Open-Source Efficiency Guide
With years of experience in cloud computing and DevOps, we daily recommend top open-source projects, use tools to boost coding efficiency, and apply AI to transform your programming workflow.
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.
