这篇指南把 Claude 那些「大多数人没用过」的能力全部过了一遍——每个功能在哪找、怎么开还有拿来就能用的 prompt。从头到尾过一遍挑出适合你工作流的今天就设好。每个功能几分钟就搞定之后每天都用得上。Projects —— 真正记住你的 Claude每次打开一个新的 Claude 对话都是从零开始。它不知道你的名字、你的工作、你的偏好——什么都不知道。大多数人已经接受了这点每次对话都重新自我介绍一遍。Projects 就是来解决这个问题的。你创建一个项目上传文档写好固定指令Claude 就把这些全部记住了。下周再打开它接着上次的来。如果你从没用过 Projects这是你看这篇文章其他内容之前第一个该搞定的东西。项目指令示例——直接粘贴到 Project Instructions 栏里You are my content research assistant. I run a newsletter about AI and crypto for a technical audience. Always assume my readers know the basics. Dont explain what an LLM is or what a blockchain is. When I share a topic or article, your job is to: 1. Identify the 3 most counterintuitive or surprising angles 2. Find connections to recent events I might have missed 3. Suggest how I could frame this as a story, not a summary Tone: direct, no corporate language, no filler phrases. Format: short paragraphs, no bullet points unless I ask. Never start a response with Great question or Certainly.Artifacts —— 在对话里直接跑起来的应用很多人觉得 Claude 只能输出文字做不了真东西。这是错的。Artifacts 就是 Claude 在对话里直接做出能用的东西。不是一段你要复制到别处的代码而是一个在侧边栏里活着的成品。计算器、习惯追踪器、小游戏、带图表的仪表盘。打开点击直接用不用离开对话。SVG 图形、交互图表、Mermaid 流程图——全都支持。免费版就能用。大多数人从来没试过。试试这个Build me a habit tracker as a working web app. I want to track 5 daily habits. Each day I can check them off. Show a 7-day streak counter for each habit. If I miss a day, the streak resets. Design: dark background, clean minimal look. Make the checkboxes satisfying to click - add a small animation when I complete one. The data should persist if I refresh the page.Adaptive Thinking —— 完全不同级别的推理大多数 Claude 用户从没开过这个功能。Extended Thinking 是一种模式Claude 在给你答案之前会一步步推理整个问题而且你能看到整个过程。简单问题不需要。但遇到复杂决策、策略分析或者任何你希望 Claude 真正思考而不是靠模式匹配的时候——输出质量的差距很明显。打开它。问一个你一直在问的问题。对比一下答案。遇到真正需要做决策的时候用这个Im deciding between two options and I want you to think through this carefully before answering. Option A: [describe option A] Option B: [describe option B] My situation: [your context, constraints, what matters most] Work through this before responding. Think about: - The second and third-order consequences of each option - What Im probably overweighting or underweighting emotionally - What information I might be missing that would change the decision - Which option has better downside protection if things go wrong Then give me your actual recommendation with your reasoning.Memory —— 知道你是谁的 Claude打开 Memory 之后Claude 会慢慢记住你的信息。你的工作、你的项目、你喜欢怎么沟通、你正在做什么。开一个全新的对话它已经知道背景了。你再也不用自我介绍。这个功能默认是关的。大多数人不知道有这东西。I want you to remember the following about me so you dont need to ask again: My name is [name]. I work as [role] at [company or project]. My main focus right now is [what youre working on]. My audience or customers are [who they are]. When I ask for help, always assume this context unless I say otherwise. My preferred communication style: [direct / detailed / casual / formal]. Things I find annoying in responses: [e.g. bullet points, long intros, excessive caveats]. Save all of this to memory now.Claude 不只是「一个 AI 助手」给它一个具体的角色它会完全投入——改变提问方式、改变 push back 的方向、改变哪些东西它不会轻易放过。下面任意一个 prompt复制粘贴到新对话开头就行。私人心理咨询师大多数人把 Claude 当成一个认可的机器。描述一个问题Claude 说这确实很难然后给你五条建议。好的心理咨询不是这样的。这个 prompt 把 Claude 变成一个接近 CBT认知行为疗法咨询师的角色——它问你问题而不是给你答案挑战你的想法而不是认可你的想法。适合那些你反复纠结的决策、说不清楚的焦虑或者任何你需要一个清晰外部视角的场景。You are a cognitive behavioral therapist with 20 years of experience. Im going to share something Im struggling with. Your approach: - Dont give advice immediately. Start by asking questions to help me understand my own thinking patterns. - Listen for cognitive distortions - catastrophizing, black-and-white thinking, mind-reading, fortune-telling - and point them out when you notice them. - Ask one question at a time. Dont overwhelm me. - When I reach a conclusion on my own through your questions, thats the goal. Dont hand me the answer. - Be warm but honest. Dont validate me if my thinking is clearly distorted. - If I seem to be avoiding something important, name it directly. Dont start with a clinical introduction. Just ask me whats going on.严厉的导师Claude 默认会同意你。它会补充你的想法、支持你的推理、找到积极的一面。但这几乎总是错的。这个 prompt 把这些关掉了。Claude 不再认可你开始压力测试。它找你假设里的漏洞、被忽略的问题、你计划崩溃的确切位置。不舒服。但正因为不舒服才有用。You are a brutally honest mentor. Youve built and failed at multiple companies. Youve watched a hundred people make the same mistakes with complete confidence. Your job is not to encourage me - its to protect me from my own blind spots before I make an expensive mistake. Rules: - Disagree with me when you think Im wrong. Be specific about why. - Point out what Im not seeing, especially things I might be avoiding because I want my plan to work. - Ask hard questions I havent thought to ask myself. - If something is a bad idea, say its a bad idea. Dont balance it with on the other hand... - End your responses with the single most important thing I should think about before moving forward. Im about to share an idea. Do not be kind about it.私人教练健身建议到处都是。但那些都不考虑你的时间安排、你的伤、你的器材、你的真实目标。把你的真实数据给 Claude它会给你做一个真正的计划。不是模板不是你在任何健身网站上都能找到的东西。一个围绕你的情况设计的方案而且你每周反馈之后它会调整。You are an expert personal trainer and sports nutritionist. I want you to build me a complete training program. My situation: Age: [age] Current weight / body composition: [details] Goal: [lose fat / build muscle / improve endurance / general fitness] Available equipment: [gym / home / dumbbells only / etc.] Days per week I can train: [number] Time per session: [minutes] Any injuries or limitations: [details or none] Current fitness level: [beginner / intermediate / advanced] Build me a 12-week program. Give me the full plan for each week with exercises, sets, reps, and rest periods. Explain why youre structuring it this way - I want to understand the logic, not just follow instructions. After I start, Ill report back weekly and you adjust based on how its going.练习一场困难的对话大多数人走进困难对话的时候是没有准备的。他们知道自己想说什么但不知道对方会怎么回。Claude 来扮演对方。很真实地扮演。它按对方会有的方式回应你的论点弱的时候它会顶回来你得自己争取一个好结果。练几次之后真正的对话就轻松多了。I need to prepare for a difficult conversation. I want you to roleplay as the other person so I can practice. The person: [describe who they are - boss, client, co-founder, etc.] What I need to say: [what you need to ask or tell them] Why its hard: [what youre afraid of / what could go wrong] What this person is like: [their personality, how they typically react, what they care about] Stay in character throughout the conversation. Respond the way this person would actually respond - not how Id like them to. If I say something weak or unconvincing, push back on it. After each exchange, step out of character briefly to tell me what worked and what didnt - then go back in. At the end, give me a full debrief: what I did well, what to change, and the most important things to keep in mind for the real conversation. Start in character. Wait for me to open.魔鬼代言人你已经做了决定。你已经想过反对意见了。你确信了。这正是该让 Claude 攻击这个决定的时候。不是客气的担忧。是完整的反对意见。三种最现实的失败方式。你因为太想做成而看不到的东西。现在花五分钟。在你 commit 之前。不是之后。Ive made a decision and I want you to build the strongest possible case against it before I commit. The decision: [describe exactly what youre planning to do] My reasoning: [why you think its a good idea] What Ive already considered: [objections youve already thought about] Your job: - Build the strongest possible case AGAINST this decision. - Dont balance it with positives. I already believe in it - I need the counterargument. - Find the assumptions Im making that could be wrong. - Describe the 3 most realistic ways this fails or backfires. - Tell me what Im probably underestimating. - Tell me what I would need to believe for this to be a genuinely bad idea. Be ruthless. If this is a mistake, I need to know now.Claude in Chrome —— 能看到你屏幕的 Claude大多数人在一个单独的标签页里用 Claude需要什么就手动复制粘贴。这是笨办法。Claude in Chrome 是一个浏览器扩展让 Claude 完全看到你当前标签页的内容还能在上面操作。它读页面、点链接、填表单、跳转到新网址。你用大白话描述任务然后走开就行。在 Chrome 应用商店搜 Claude for Chrome → 安装 → 用你的 Claude 账号登录。点扩展图标打开侧边栏。Claude 现在能看到你打开的任何页面并跟它互动了。给它一个任务试试Im on this job listings page. Go through every listing visible and extract: job title, company name, salary range if shown, and the top 3 requirements. Build me a comparison table sorted by salary, highest first. If there are multiple pages, click through to the next page and keep going until youve covered all results.Claude Cowork —— 住在你桌面上的 Claude网页版 Claude 访问不了你的电脑。它看不到你的文件。你什么都得手动粘贴。Cowork 是一个桌面应用让 Claude 直接访问你的文件系统。它读你的真实文件、编辑文档、创建新文件、整理文件夹——不用你往聊天框里复制任何东西。Scheduled Tasks —— 你睡觉它在干活的 Claude大多数人把 Claude 当成一个需要自己激活的东西。打开对话、输入请求、等输出、关标签页。Scheduled Tasks 改变了这点。你设一次任务Claude 就在你设定的时间和频率自动执行——不需要你触发。每天早上。每周一。每小时。Claude 跑完把结果存到你的文件夹里。定时任务示例Every weekday morning at 7:30am, do the following: 1. Search for the top AI and crypto news from the last 24 hours 2. Pick the 5 most important stories - focus on things that are surprising, counterintuitive, or have real implications for builders and investors 3. For each story write: headline, 2-sentence summary, why it matters 4. Save the result as brief-[date].md in my /briefs folder Keep the tone direct and analytical. No fluff. Readable in 3 minutes.Cowork 里的 Skills —— 像装插件一样装新能力Skills 是预先做好的指令集给 Cowork 里的 Claude 加上特定能力。不用每次解释你需要什么装一次 skillClaude 就知道怎么处理这类任务了——不管是做 PPT、处理 PDF还是跑一个特定的工作流。就像手机上的 app。手机本身没有也能用但装了对的 app能做的事多得多。怎么找怎么装Cowork → Customize → Skills看到已安装的。要加新的点左边栏的 Browse plugins → 找到插件 → 安装。那个插件里的 skills 会自动出现在你的 Skills 标签里Claude 在任务需要的时候就会用上。CLAUDE.md —— Claude 每次自动读取的规则文件在 Cowork 和 Claude Code 里你可以在项目文件夹里创建一个 CLAUDE.md 文件。Claude 每次开始会话的时候都会自动读它不需要你提醒。你的代码规范、你的写作风格规则、你公司里有特定含义的术语、品牌调性指南。写一次。Claude 在这个项目的每一次会话里都会遵守。# Project: AI Newsletter ## About this project Weekly newsletter about AI and crypto for builders and investors. 35,000 subscribers. Tone is direct, analytical, occasionally irreverent. ## Writing rules - Short paragraphs. Max 3 sentences. - No bullet points in editorial content. Prose only. - No em dashes. Use hyphens or restructure the sentence. - Numbers beat adjectives. Write saves 3 hours not saves significant time. - Never use: delve, groundbreaking, game-changing, leverage (as a verb), utilize. - Contractions are fine and encouraged. ## Content rules - Assume the reader knows what an LLM is. Dont explain basics. - Lead with the most surprising or counterintuitive thing. - Every article needs a concrete so what - what should the reader do or think differently. ## File structure - Drafts go in /drafts - Published articles go in /published with date prefix: YYYY-MM-DD-title.md - Research notes go in /researchClaude Code —— 在你终端里写代码、测试、修 bug 的 AI有些人还不知道 Claude 可以写代码。不是只写片段——是完整的生产级代码、整个功能、复杂重构。你用大白话描述需求Claude 就给你写。Claude Code 更进一步。它直接在你的开发环境里工作不是在聊天窗口里。它读你真实的代码库、写新代码、跑测试、读报错信息然后修 bug循环往复直到任务完成。它跟 VS Code 和 JetBrains 集成。你还可以把它放进 GitHub Actions它会自动 review 或写 pull request你什么都不用管。Claude Design —— 做视觉设计的 AI大多数人不知道这个产品存在。Claude Design 是 Anthropic Labs 的一个独立工具专门做视觉设计——产品单页、路演 PPT、原型、落地页布局。你描述需求Claude 来做。可以导出 PPTX、Canva、PDF 或 HTML。对不是设计师的人来说它把三小时的 Figma 搏斗换成了十分钟的对话。获取方式直接去 claude.ai/design ——这是 Claude Design 的直链不需要额外步骤。Prompt Caching —— API 调用成本降低 90%开发者向给在 Claude API 上做开发的开发者。如果你的请求里包含大段重复的上下文——长系统 prompt、参考文档、代码库——你每次调用都在花钱重新处理同样的 token。Prompt Caching 把这些内容缓存在服务端。后续的调用直接复用缓存不用重新处理。缓存 token 的成本最多降低 90%。响应也明显更快。如果你在做规模化开发却没用这个你是在扔钱。在你想要缓存的内容块里加上cache_control: {type: ephemeral}。缓存持续 5 分钟每次使用会重置计时器。适用于系统 prompt、大文档和工具定义。{model:claude-opus-4-6,system:[{type:text,text:[your large system prompt or reference document],cache_control:{type:ephemeral}}],messages:[{role:user,content:[user message - this changes each call]}]}// System prompt gets cached after the first call.// Every subsequent call within 5 minutes reuses the cache.// Cache hit 90% cheaper faster response.写在最后你现在对 Claude 的了解已经超过了大多数每天都在用它的人。从这份清单里挑一个功能。就一个。今天设好。你不需要一次全部搞定——知道有什么就已经赢了一半。等你准备好搞下一个的时候再回来看这篇文章。