模块 13 · 编程Module 13 · Coding
AI 时代编程
Coding in the AI Era
AI 时代,会写代码 = 会放大你的能力。
In the AI era, coding = amplifying your ability.
C1
AI 时代,编程还重要吗?
Does Programming Still Matter in the AI Era?
AI 不会让编程消失,而是让编程的门槛降低、价值上移。会用 AI 写代码 = 新时代的『会写文章』。
AI won't kill programming; it lowers the barrier and raises the value. Coding with AI = the new 'writing skills'.
为什么 AI 时代编程反而更重要?3 个原因:① AI 让『从想法到产品』的路径变短了——以前要团队,现在 1 人 + AI 就能做出可用的产品;② 编程能力是 AI 时代的『底层杠杆』——会写代码的人,能用 AI 放大 10 倍产出;③ 即使不写代码,『计算思维』也是 AI 时代最稀缺的思维方式。
Why programming is more important in the AI era? 3 reasons: ① AI shortens the path from 'idea to product'—used to need a team, now 1 person + AI makes a usable product; ② coding is the 'base lever' of the AI era—people who code can amplify output 10x with AI; ③ even without coding, 'computational thinking' is the rarest mindset in the AI era.
| 信号 | Signal | 含义 | Meaning | 对编程的意义 | Implication for coding |
|---|---|---|---|---|---|
| AI 让编码成本趋近于零 | AI brings coding cost near zero | 自然语言 → 代码 | Natural language → code | 编程从『手艺』变『沟通』 | Coding shifts from 'craft' to 'communication' |
| 一人公司成为可能 | One-person company becomes possible | 1 人 + AI = 1 团队 | 1 person + AI = 1 team | 会编程 = 会造产品 | Coding = making products |
| 计算思维成为底层能力 | Computational thinking becomes base | 拆解问题 + 抽象 + 算法 | Decompose + abstract + algorithm | 编程思维 > 具体语言 | Thinking > specific language |
| 低代码/无代码崛起 | Low-code/no-code rises | 拖拽 + AI 生成 | Drag-drop + AI generate | 非程序员也能造工具 | Non-programmers can build tools |
🎯AI 时代编程的定位变了:不是『谋生手艺』,而是『放大杠杆』。你会用 AI 写代码,就能把任何想法快速变成产品。这是新时代的『通识』。In the AI era, coding's position changed: not a 'living craft' but an 'amplification lever'. Using AI to code, you turn any idea into a product quickly. It's the new 'general literacy'.
C2
8 大核心概念:AI 时代编程的底层逻辑
8 Core Concepts: The Underlying Logic of Coding in the AI Era
AI 时代编程不是『学一门语言』,而是 8 个底层逻辑的重构。逐个理解,你就掌握了编程的完整框架。
Coding in the AI era isn't 'learning a language' but reconstructing 8 underlying logics. Understand each and you master the complete framework of coding.
| 概念 | Concept | 一句话解释 | One-line explanation | 为什么关键 | Why it matters | 怎么落地 | How to land it |
|---|---|---|---|---|---|---|---|
| 计算思维 | Computational thinking | 拆解问题+抽象+模式识别+算法 | Decompose+abstract+pattern+algorithm | 编程的底层,比任何语言都重要 | The base of coding; more important than any language | 用『拆解』思维解决任何问题 | Use 'decompose' thinking for any problem |
| 提示词工程 | Prompt engineering | 用自然语言精确表达需求给 AI | Express needs precisely to AI in natural language | AI 编码的质量 = 提示词的质量 | AI coding quality = prompt quality | 学写清晰的编码提示词 | Learn to write clear coding prompts |
| 代码审查 | Code review | 人审 AI 写的代码,抓错 | Human reviews AI code, catches bugs | AI 会错,你是最后防线 | AI errs; you're the last line | 每段 AI 代码都审查一遍 | Review every AI code snippet |
| 系统思维 | Systems thinking | 理解组件如何交互,不只看单点 | Understand how components interact, not just single points | AI 擅长单点,不擅长系统 | AI good at points, bad at systems | 学架构 + 数据流 | Learn architecture + data flow |
| 版本控制 | Version control | Git 记录每一次改动,可回滚 | Git records every change, can rollback | AI 改动多,必须可追溯 | AI makes many changes; must be traceable | 用 Git + 分支 | Use Git + branches |
| 测试思维 | Test thinking | 写代码前先想『怎么证明它对』 | Think 'how to prove it's right' before coding | AI 生成的代码必须验证 | AI code must be verified | 写单元测试 + 边界测试 | Write unit tests + boundary tests |
| 复用思维 | Reuse thinking | 不重复造轮子,用现成组件 | Don't reinvent the wheel; use existing components | AI 时代更强调复用 | AI era emphasizes reuse more | 学用开源 + 组件库 | Learn open source + component libraries |
| 产品思维 | Product thinking | 代码是手段,产品是目的 | Code is the means; product is the end | AI 让代码变便宜,产品思维变贵 | AI makes code cheap; product thinking becomes expensive | 先想用户要什么,再写代码 | Think what users need first, then code |
🧭8 个概念里,最常被忽略的是『产品思维』。AI 时代代码变便宜了,但『做对的产品』依然稀缺。会写代码的人很多,会做产品的人很少。Among the 8 concepts, the most ignored is 'product thinking'. In the AI era code is cheap, but 'making the right product' remains scarce. Many can code; few can make products.
C3
10 个 AI 时代编程方向(按入门难度排序)
10 AI-Era Coding Directions (By Entry Difficulty)
10 个适合 AI 时代学编程的方向,按入门难度排序,帮你找到最省力的起点。
10 coding directions for the AI era, ranked by entry difficulty, to find your least-effort starting point.
| 方向 | Direction | 核心技能 | Core skills | 入门难度 | Entry difficulty | 前景 | Prospect | 适合谁 | Who it suits |
|---|---|---|---|---|---|---|---|---|---|
| AI 应用开发 | AI app development | Python + LLM API | Python + LLM API | 低 | Low | 很高 | Very high | 所有人 | Everyone |
| 自动化脚本 | Automation scripts | Python + 工具 | Python + tools | 低 | Low | 高 | High | 办公/运营 | Office/operations |
| 数据脚本 | Data scripts | Python + Pandas | Python + Pandas | 低 | Low | 高 | High | 分析/运营 | Analysis/operations |
| Web 前端 | Web frontend | HTML/CSS/JS | HTML/CSS/JS | 中 | Medium | 高 | High | 设计/产品 | Design/product |
| Web 后端 | Web backend | Python/Node + 数据库 | Python/Node + DB | 中 | Medium | 高 | High | 逻辑强的人 | Logically strong people |
| 数据分析 | Data analysis | SQL + Python + 可视化 | SQL + Python + viz | 中 | Medium | 很高 | Very high | 财务/市场/运营 | Finance/marketing/operations |
| 机器学习 | Machine learning | Python + 数学 + 框架 | Python + math + frameworks | 高 | High | 很高 | Very high | 数学强的人 | Math-strong people |
| 移动开发 | Mobile dev | Swift/Kotlin + 框架 | Swift/Kotlin + frameworks | 中 | Medium | 中高 | Medium-high | 产品思维的人 | Product-thinking people |
| 游戏开发 | Game dev | C#/Unity + 图形 | C#/Unity + graphics | 中高 | Medium-high | 中高 | Medium-high | 创意+逻辑 | Creative + logical |
| 系统/底层 | System/low-level | C/C++ + 计算机原理 | C/C++ + CS principles | 高 | High | 高 | High | 底层兴趣的人 | Low-level enthusiasts |
🧭选编程方向的核心不是『哪个最火』,而是『哪个能最快做出作品』。AI 应用开发 + 自动化脚本是 2026 年最快出作品的两个方向,推荐作为起点。The core of choosing a coding direction isn't 'which is hottest' but 'which makes work fastest'. AI app development + automation scripts are the two fastest-to-work directions in 2026, recommended as starting points.
C4
10 个 AI 编程提示词(直接复制用)
10 AI Coding Prompts (Copy & Use)
10 个让 AI 帮你写代码的提示词,覆盖入门/进阶/调试/架构。
10 prompts to let AI help you code, covering beginner/advanced/debugging/architecture.
| 提示词 | Prompt | 用途 | Use |
|---|---|---|---|
| 帮我用 Python 写一个[功能],要求[约束],给出完整可运行代码 + 注释。 | Help me write [function] in Python, with [constraints], give complete runnable code + comments. | 写代码 | Write code |
| 这段代码有什么 bug?[贴代码]。找出问题 + 给出修复 + 解释原因。 | What bugs does this code have? [paste code]. Find the issue + fix + explain. | 调试 | Debugging |
| 优化这段代码的性能:[贴代码]。分析瓶颈 + 给出优化方案 + 复杂度对比。 | Optimize this code's performance: [paste code]. Analyze bottlenecks + give optimization + complexity comparison. | 优化 | Optimization |
| 把这段代码重构得更清晰:[贴代码]。保持功能不变,提升可读性。 | Refactor this code to be clearer: [paste code]. Keep functionality, improve readability. | 重构 | Refactoring |
| 解释这段代码是怎么工作的:[贴代码]。逐行说明 + 关键概念。 | Explain how this code works: [paste code]. Line-by-line + key concepts. | 学代码 | Learn code |
| 设计一个[系统]的架构,用[技术栈]。给出组件 + 数据流 + 部署图。 | Design an architecture for [system] with [tech stack]. Give components + data flow + deployment diagram. | 架构 | Architecture |
| 给这段代码写单元测试:[贴代码]。覆盖正常/边界/异常,用 pytest。 | Write unit tests for this code: [paste code]. Cover normal/boundary/exception, use pytest. | 测试 | Testing |
| 把这段代码从[语言A]翻译成[语言B]:[贴代码]。保持逻辑 + 适配语法。 | Translate this code from [language A] to [language B]: [paste code]. Keep logic + adapt syntax. | 翻译 | Translation |
| 用[技术栈]实现一个[功能]的 MVP,给出完整项目结构 + 核心代码。 | Implement an MVP of [function] with [tech stack]. Give complete project structure + core code. | 做 MVP | Make MVP |
| 我的代码报这个错:[错误信息]。[贴相关代码]。给出最可能的 3 个原因 + 排查步骤。 | My code gives this error: [error]. [paste relevant code]. Give the 3 most likely causes + troubleshooting steps. | 排错 | Troubleshoot |
📚AI 编程提示词的核心是『具体』:把功能/约束/技术栈写清楚,AI 的代码质量才高。模糊提示 = 模糊代码。The core of AI coding prompts is 'specificity': write function/constraints/tech stack clearly, and AI's code quality is high. Vague prompt = vague code.
C5
12 个 AI 编程常见误区(别踩坑)
12 Common AI Coding Myths (Don't Step Into the Pit)
12 个学 AI 编程时最容易踩的坑,识别它们,避免走弯路。
12 most common pitfalls in learning AI coding; recognize them to avoid detours.
| 误区 | Myth | 正确做法 | Correct approach |
|---|---|---|---|
| 『AI 写代码就不用学基础』 | 'AI codes, no need to learn basics' | 基础是审查 AI 代码的前提,必须学 | Basics are the prerequisite for reviewing AI code; must learn |
| 『直接信 AI 的代码』 | 'Directly trust AI's code' | 每段都审查 + 测试,AI 会错 | Review + test every snippet; AI errs |
| 『学一门语言就够』 | 'Learning one language is enough' | 学 1 门精通 + 了解 2-3 门 | Master 1 + know 2-3 |
| 『先学完再动手』 | 'Learn everything before starting' | 边做边学,项目驱动 | Learn while doing, project-driven |
| 『代码能跑就对了』 | 'If it runs, it's right' | 能跑≠对,要测试 + 边界 | Running ≠ right; need tests + boundaries |
| 『追最新框架』 | 'Chase the newest frameworks' | 选稳定的主流,不追热点 | Choose stable mainstream, don't chase trends |
| 『忽视代码质量』 | 'Ignore code quality' | 可读性 + 可维护性是长期价值 | Readability + maintainability is long-term value |
| 『只看教程不写代码』 | 'Only watch tutorials, don't code' | 动手写 1000 行 > 看 100 个教程 | Writing 1000 lines > watching 100 tutorials |
| 『AI 时代学编程没用』 | 'Learning coding is useless in the AI era' | 恰恰相反,编程是 AI 时代的杠杆 | On the contrary, coding is the lever of the AI era |
| 『追求完美再发布』 | 'Perfect before release' | 先 MVP 发布,再迭代 | Release MVP first, then iterate |
| 『忽视版本控制』 | 'Ignore version control' | 从第一行就用 Git | Use Git from the first line |
| 『只学语法不学思维』 | 'Only learn syntax, not thinking' | 计算思维比语法更重要 | Computational thinking matters more than syntax |
⚠️AI 编程最大的坑不是『代码错』,而是『不审查』。AI 写的代码看起来对,但隐藏 bug。养成『审查 + 测试』的习惯,是 AI 编程的生死线。The biggest pitfall of AI coding isn't 'wrong code' but 'not reviewing'. AI code looks right but hides bugs. Forming the 'review + test' habit is the life-or-death line of AI coding.
C6
30 天 AI 编程启动计划(零基础版)
30-Day AI Coding Launch Plan (Zero-Basis Version)
30 天从零基础启动 AI 编程的完整计划,每天具体动作,用 AI 加速。
A complete 30-day plan to launch AI coding from zero basis, with concrete daily actions, accelerated by AI.
| 周 | Week | 主题 | Theme | 核心动作 | Core actions | 产出 | Output |
|---|---|---|---|---|---|---|---|
| W1 | W1 | 入门 | Beginner | 学 Python 基础(变量/循环/函数)+ 用 AI 解释每个概念 | Learn Python basics (variables/loops/functions) + let AI explain each concept | 能写 10 个小脚本 | Can write 10 small scripts |
| W2 | W2 | 动手 | Hands-on | 做 3 个小项目(爬虫/文件处理/数据分析) | Make 3 small projects (crawler/file processing/data analysis) | 3 个可运行作品 | 3 runnable works |
| W3 | W3 | AI 协作 | AI collaboration | 用 AI 写代码 + 审查 + 测试,形成工作流 | Use AI to code + review + test, form a workflow | 1 个 AI 协作项目 | 1 AI-collab project |
| W4 | W4 | 作品 | Work | 做 1 个完整 MVP(解决一个真实问题) | Make 1 complete MVP (solve a real problem) | 1 个可发布 MVP | 1 publishable MVP |
📅30 天计划的核心是『项目驱动 + AI 协作』。不先学完再动手,而是边做边学。每天 1-2 小时,30 天你就能做出可发布的 MVP。The core of the 30-day plan is 'project-driven + AI collaboration'. Don't learn everything first, but learn while doing. 1-2 hours daily, and in 30 days you can make a publishable MVP.
C7
30 个『AI 编程』高频问题(一问一答)
30 'AI Coding' High-Frequency Questions (One Q One A)
30 个 AI 编程高频问题,覆盖认知/入门/进阶/工具。
30 high-frequency AI-coding questions, covering cognition/beginner/advanced/tools.
能。AI 时代入门门槛大幅降低,用 AI 解释 + 项目驱动,3 个月可入门。Yes. The entry barrier has dropped greatly; with AI explanation + project-driven, you can start in 3 months.
基础编程不用高数。AI 应用/自动化只需基础逻辑。机器学习才需要线性代数。Basic coding doesn't need advanced math. AI apps/automation only need basic logic. ML needs linear algebra.
Python 入门最友好 + AI 生态最强。2026 年推荐 Python 起步。Python is the friendliest entry + strongest AI ecosystem. Python recommended for 2026.
不能直接信。要审查 + 测试。AI 会错,你是最后防线。Don't trust directly. Review + test. AI errs; you're the last line.
基础编程学常见算法(排序/搜索/递归)即可。不要求竞赛级。Learn common algorithms (sort/search/recursion) for basic coding. No competition level needed.
建议学基础(数组/链表/哈希/树)。理解数据怎么组织,写代码更高效。Recommend learning basics (array/linked list/hash/tree). Understanding data organization makes coding more efficient.
看需求。简单工具够用;复杂产品要写代码。AI 时代两者结合。Depends on needs. Simple tools suffice; complex products need code. In the AI era, combine both.
不会消失,但会『上移』。重复性编码被替代,设计/架构/产品价值上升。Won't disappear, but will 'move up'. Repetitive coding is replaced; design/architecture/product value rises.
能。用自然语言 + AI,非程序员也能做出可用的工具/应用。Yes. With natural language + AI, non-programmers can make usable tools/apps.
30 天入门 MVP;3 个月做实用工具;6-12 个月做产品。30 days for MVP entry; 3 months for a practical tool; 6-12 months for a product.
2026 年:Claude/Cursor/Copilot 都是好选择。核心是『你能用 AI 协作』,不是追最强。In 2026: Claude/Cursor/Copilot are all good choices. The core is 'you can collaborate with AI', not chasing the strongest.
看 3 点:逻辑对不对、边界处理没、性能够不够。用 AI 辅助审查 + 自己判断。Check 3 points: logic right, boundaries handled, performance enough. Use AI to assist review + judge yourself.
先想『怎么证明它对』,写正常/边界/异常 3 类用例。用 pytest/内置测试框架。First think 'how to prove it's right', write normal/boundary/exception 3 types of cases. Use pytest/built-in frameworks.
极重要。从第一行就用。记录改动、可回滚、可协作。Extremely important. Use from the first line. Records changes, can rollback, can collaborate.
做产品要学。SQL + 一种 NoSQL(如 MongoDB)。数据是产品的核心。Learn for products. SQL + one NoSQL (like MongoDB). Data is the core of products.
做完整产品要学。HTML/CSS/JS + 一个框架(React/Vue)。Learn for complete products. HTML/CSS/JS + a framework (React/Vue).
做完整产品要学。Python/Node + 框架 + 数据库。Learn for complete products. Python/Node + framework + database.
要学基础。把产品上线(云/服务器)。学 Docker + 一种云平台。Learn basics. Put the product online (cloud/server). Learn Docker + one cloud platform.
进阶要学。CI/CD + 自动化部署。先会基础再深入。Learn for advanced. CI/CD + automated deployment. Learn basics first, then deep.
建议学一种(AWS/阿里云/Azure)。产品最终都要上云。Recommend learning one (AWS/Ali Cloud/Azure). Products eventually go to the cloud.
做产品要学基础(认证/加密/防注入)。安全是底线。Learn basics for products (auth/encryption/injection prevention). Security is the bottom line.
进阶要学。先让它对,再让它快。学 profiling + 常见瓶颈。Learn for advanced. First make it right, then fast. Learn profiling + common bottlenecks.
做复杂产品要学。组件/数据流/扩展性。先做简单再学架构。Learn for complex products. Components/data flow/scalability. Start simple, then learn architecture.
进阶要学。复用 + 解耦。先理解再套用,不生搬硬套。Learn for advanced. Reuse + decoupling. Understand first, then apply, don't force.
建议学。用开源 + 贡献开源,是能力提升最快的方式。Recommend learning. Using + contributing to open source is the fastest way to improve.
ML 方向要学。线性代数/概率/微积分。AI 应用方向不要求。Learn for ML direction. Linear algebra/probability/calculus. AI app direction doesn't require it.
做 AI 应用要学基础。理解 LLM 的能力边界 + 提示词工程。Learn basics for AI apps. Understand LLM capability boundaries + prompt engineering.
做 AI 应用建议学。让 LLM 用你的数据回答,是 AI 应用的核心。Recommend learning for AI apps. Let LLM answer with your data, is the core of AI apps.
进阶要学。让 AI 自主执行多步任务,是 AI 应用的前沿。Learn for advanced. Let AI autonomously execute multi-step tasks, is the frontier of AI apps.
先编程基础 + AI 工具并行。会编程才能审查 AI 的代码,会用 AI 才能加速编程。Learn coding basics + AI tools in parallel. Coding lets you review AI code; AI lets you accelerate coding.
💡30 个问题的核心答案:先 Python + AI 协作,项目驱动,边做边学。别追『学完』,追『做出作品』。The core answer to the 30 questions: start with Python + AI collaboration, project-driven, learn while doing. Don't chase 'finishing learning', chase 'making work'.
C8
30 条『AI 编程』金句(可背可分享)
30 'AI Coding' Golden Quotes (Memorize & Share)
30 条金句,浓缩 AI 时代编程的核心心态,可背可分享。
30 golden quotes condensing the core mindset of coding in the AI era, memorize and share.
这是 AI 时代编程的铁律。This is the iron rule of coding in the AI era.
AI 让代码变便宜,产品思维变贵。AI makes code cheap; product thinking becomes expensive.
稀缺的是『做对的产品』。What's scarce is 'making the right product'.
AI 会错,你是最后防线。AI errs; you're the last line.
性能优化在对的基础之上。Performance optimization builds on correctness.
测试 + 边界是底线。Tests + boundaries are the bottom line.
项目驱动是最快的路径。Project-driven is the fastest path.
动手是学习的唯一方式。Hands-on is the only way to learn.
语言会变,思维不变。Languages change; thinking doesn't.
AI 时代更强调复用。The AI era emphasizes reuse more.
可追溯是底线。Traceability is the bottom line.
完美主义是产品杀手。Perfectionism is a product killer.
它放大你的能力,不取代你的判断。It amplifies your ability, doesn't replace your judgment.
把需求写清楚,AI 才给好代码。Write the need clearly, and AI gives good code.
AI 擅长单点,不擅长系统。AI is good at points, bad at systems.
没有测试 = 裸奔。No tests = running naked.
三个月后你还得读懂它。You'll need to read it in 3 months.
热点会变,主流稳定。Trends change; mainstream is stable.
AI 给选项,你做决策。AI gives options; you make decisions.
精通一门 > 略知五门。Mastering one > knowing five slightly.
不能上线的代码 = 没写。Code that can't go online = not written.
一次安全事故 = 前功尽弃。One security incident = all wasted.
保护好数据,产品才有未来。Protect data, and the product has a future.
读优秀开源代码,进步最快。Reading excellent open-source code, progress is fastest.
像会写字一样,会写代码。Like being able to write, being able to code.
期望要现实。Expectations should be realistic.
专注,别贪多。Focus, don't be greedy.
学底层,别只学表层。Learn the base, not just the surface.
重复的交给 AI,创造的留给自己。Give repetition to AI, keep creation for yourself.
会造 = 会想 + 会做 + 会用 AI。Able to make = able to think + do + use AI.
💬30 条金句,每天 1 条 + 当天用 1 次。30 天读完,你的编程心态会肉眼可见地变稳。30 quotes, 1 daily + use 1 time that day. Read all in 30 days, your coding mindset will visibly stabilize.
C9
30 条『AI 编程』实用技巧(高手常用)
30 Practical 'AI Coding' Techniques (Experts' Usual)
30 个 AI 编程时真正好用的技巧,不是鸡汤,是可执行动作。
30 techniques that truly work in AI coding, not chicken soup, but executable actions.
| 技巧 | Technique | 做法 | How to do it | 效果 | Effect |
|---|---|---|---|---|---|
| 提示词具体化 | Prompt specificity | 把功能/约束/技术栈/示例写清楚 | Write function/constraints/tech stack/examples clearly | AI 代码质量高 | High AI code quality |
| 分步生成 | Step-by-step generation | 复杂功能拆成 3-5 步,逐步生成 | Break complex features into 3-5 steps, generate step by step | 避免一次生成太长出错 | Avoid long-generation errors |
| 给 AI 看上下文 | Give AI context | 贴相关代码 + 说明目标,不只是片段 | Paste relevant code + explain the goal, not just snippets | AI 理解你的意图 | AI understands your intent |
| 让 AI 解释代码 | Let AI explain code | 对陌生代码,让 AI 逐行解释 | For unfamiliar code, let AI explain line by line | 快速读懂 | Quickly understand |
| 让 AI 找 bug | Let AI find bugs | 贴代码 + 错误信息,让 AI 找原因 | Paste code + error, let AI find the cause | 快速定位 | Quickly locate |
| 让 AI 写测试 | Let AI write tests | 让 AI 按正常/边界/异常写测试 | Let AI write tests by normal/boundary/exception | 测试覆盖全 | Full test coverage |
| 让 AI 重构 | Let AI refactor | 让 AI 重构代码,保持功能不变 | Let AI refactor code, keep functionality unchanged | 代码更清晰 | Clearer code |
| 让 AI 优化性能 | Let AI optimize performance | 让 AI 分析瓶颈 + 优化 + 对比复杂度 | Let AI analyze bottlenecks + optimize + compare complexity | 性能提升 | Performance improvement |
| 让 AI 翻译代码 | Let AI translate code | 跨语言翻译,保持逻辑 + 适配语法 | Cross-language translation, keep logic + adapt syntax | 跨语言迁移 | Cross-language migration |
| 审查 3 要点 | Review 3 points | 逻辑对不对、边界处理没、性能够不够 | Logic right, boundaries handled, performance enough | 审查高效 | Efficient review |
| 边界测试 | Boundary testing | 测空值/极值/异常输入 | Test null/extreme/abnormal input | 抓隐藏 bug | Catch hidden bugs |
| 版本控制习惯 | Version control habit | 小步提交 + 有意义的 commit 信息 | Small commits + meaningful commit messages | 可追溯 | Traceable |
| 分支开发 | Branch development | 主分支稳定 + 特性分支开发 | Stable main + feature branches | 不破坏主线 | Don't break the main line |
| 代码注释 | Code comments | 注释『为什么』,不只『是什么』 | Comment 'why', not just 'what' | 代码自解释 | Self-explaining code |
| 复用开源 | Reuse open source | 找成熟的库,不重复造轮子 | Find mature libraries, don't reinvent the wheel | 节省时间 | Save time |
| 组件化 | Componentization | 把可复用部分拆成组件 | Break reusable parts into components | 复用 + 维护 | Reuse + maintenance |
| 配置分离 | Config separation | 把可变配置抽出来,不硬编码 | Extract variable config, don't hardcode | 易维护 | Easy to maintain |
| 日志习惯 | Logging habit | 关键节点打日志,方便排查 | Log at key points, ease troubleshooting | 快速定位问题 | Quickly locate problems |
| 异常处理 | Exception handling | 捕获 + 处理 + 上报,不吞异常 | Catch + handle + report, don't swallow exceptions | 系统稳定 | System stability |
| 幂等设计 | Idempotent design | 重复调用结果一致 | Repeated calls give consistent results | 防重复操作 | Prevent duplicate operations |
| 超时控制 | Timeout control | 给外部调用设超时 | Set timeouts for external calls | 防卡死 | Prevent hanging |
| 重试机制 | Retry mechanism | 对暂时性失败重试(带退避) | Retry on transient failures (with backoff) | 提高可靠性 | Improve reliability |
| 数据校验 | Data validation | 入口处校验输入,不信任外部 | Validate input at entry, don't trust external | 防注入/脏数据 | Prevent injection/dirty data |
| 最小权限 | Least privilege | 只给必要的权限 | Only give necessary permissions | 安全底线 | Security bottom line |
| 环境变量 | Environment variables | 密钥/配置用环境变量,不进代码库 | Secrets/config in env vars, not in the repo | 安全 | Security |
| 代码风格统一 | Consistent code style | 用 linter/formatter 统一风格 | Use linter/formatter to unify style | 团队协作顺畅 | Smooth team collaboration |
| 持续学习 | Continuous learning | 每周学 1 个新技巧/新工具 | Learn 1 new technique/tool weekly | 能力持续更新 | Ability continuously updated |
| 项目复盘 | Project review | 每个项目结束复盘:什么好/什么差 | Review each project: what's good/what's bad | 经验沉淀 | Experience sediment |
| 写技术文档 | Write tech docs | 把解决方案写成文档 | Write solutions as docs | 知识复用 | Knowledge reuse |
| 结对/评审 | Pairing/review | 让 AI 或同事审查你的代码 | Let AI or colleagues review your code | 质量提升 | Quality improvement |
🧭30 个技巧不用全用。最核心的 5 个:提示词具体化 + 分步生成 + 审查 3 要点 + 边界测试 + 版本控制。先做这 5 个,AI 编程就高效了。No need to use all 30. The 5 most core: prompt specificity + step-by-step + review 3 points + boundary testing + version control. Do these 5 first, and AI coding is efficient.
C10
10 个 AI 编程工具对比(2026)
10 AI Coding Tools Compared (2026)
10 个 2026 年主流的 AI 编程工具,按用途对比,帮你选型。
10 mainstream AI coding tools in 2026, compared by use case, to help you choose.
| 工具 | Tool | 类型 | Type | 强项 | Strengths | 费用 | Cost | 适合谁 | Who it suits |
|---|---|---|---|---|---|---|---|---|---|
| Claude | Claude | 通用 LLM | General LLM | 长上下文 + 代码理解 + 架构 | Long context + code understanding + architecture | 免费/订阅 | Free/sub | 所有人 | Everyone |
| ChatGPT | ChatGPT | 通用 LLM | General LLM | 综合能力 + 多模态 | Overall ability + multimodal | 免费/订阅 | Free/sub | 所有人 | Everyone |
| Cursor | Cursor | AI IDE | AI IDE | 代码编辑器 + AI 补全/重构 | Code editor + AI completion/refactoring | 订阅 | Sub | 日常写代码 | Daily coding |
| GitHub Copilot | GitHub Copilot | AI 补全 | AI completion | IDE 内补全 + 行级建议 | In-IDE completion + line-level suggestions | 订阅 | Sub | IDE 用户 | IDE users |
| Windsurf | Windsurf | AI IDE | AI IDE | 多文件编辑 + Agent 模式 | Multi-file editing + agent mode | 订阅 | Sub | 中大型项目 | Medium-large projects |
| Replit | Replit | 云端 IDE | Cloud IDE | 在线协作 + 一键部署 | Online collaboration + one-click deploy | 免费/订阅 | Free/sub | 快速原型 | Quick prototypes |
| Codeium | Codeium | AI 补全 | AI completion | 免费额度 + 多语言 | Free tier + multi-language | 免费/订阅 | Free/sub | 预算有限者 | Budget-limited |
| Tabnine | Tabnine | AI 补全 | AI completion | 私有化 + 安全 | Private + security | 订阅 | Sub | 企业/安全敏感 | Enterprise/security-sensitive |
| Bolt.new | Bolt.new | AI 全栈生成 | AI full-stack generation | 自然语言 → 全栈应用 | Natural language → full-stack app | 订阅 | Sub | 非程序员做应用 | Non-programmers making apps |
| v0 | v0 | AI 前端生成 | AI frontend generation | 自然语言 → 前端 UI | Natural language → frontend UI | 订阅 | Sub | 前端快速原型 | Frontend quick prototype |
🧭工具选型原则: ① 入门选 Claude/ChatGPT(免费起步);② 日常写代码选 Cursor/Copilot;③ 快速做应用选 Bolt.new/v0;④ 企业选 Tabnine(安全)。别追『最强』,追『最合适』。Tool selection principle: ① entry: Claude/ChatGPT (free start); ② daily coding: Cursor/Copilot; ③ quick apps: Bolt.new/v0; ④ enterprise: Tabnine (security). Don't chase 'strongest', chase 'most suitable'.
C11
30 天 AI 编程详细日历(每天做什么)
30-Day AI Coding Detailed Calendar (What to Do Each Day)
30 天每天的具体任务,精确到小时,直接照做。
Concrete daily tasks for 30 days, precise to the hour, just do it.
| 天数 | Day | 任务 | Task | 时长 | Duration | 产出 | Output |
|---|---|---|---|---|---|---|---|
| D1 | D1 | 装 Python + 跑第一个 Hello World | Install Python + run first Hello World | 1h | 1h | 能运行 Python | Can run Python |
| D2 | D2 | 学变量/类型/运算(AI 解释) | Learn variables/types/operations (AI explains) | 1h | 1h | 5 个小脚本 | 5 small scripts |
| D3 | D3 | 学条件/循环 | Learn conditionals/loops | 1.5h | 1.5h | 3 个小脚本 | 3 small scripts |
| D4 | D4 | 学函数/作用域 | Learn functions/scope | 1.5h | 1.5h | 3 个函数练习 | 3 function exercises |
| D5 | D5 | 学列表/字典 | Learn lists/dicts | 1.5h | 1.5h | 3 个数据结构练习 | 3 data structure exercises |
| D6 | D6 | 学文件读写 | Learn file read/write | 1.5h | 1.5h | 1 个文件处理脚本 | 1 file processing script |
| D7 | D7 | 学异常处理 + 复习 | Learn exception handling + review | 1.5h | 1.5h | 复习笔记 | Review notes |
| D8 | D8 | 项目 1:爬虫(抓网页数据) | Project 1: crawler (scrape web data) | 3h | 3h | 可运行爬虫 | Runnable crawler |
| D9 | D9 | 项目 1 改进:加存储 + 异常 | Project 1 improve: add storage + exceptions | 2h | 2h | 改进版爬虫 | Improved crawler |
| D10 | D10 | 项目 2:文件批处理 | Project 2: batch file processing | 2h | 2h | 可运行批处理 | Runnable batch processing |
| D11 | D11 | 项目 2 改进:加日志 + 参数 | Project 2 improve: add logging + params | 2h | 2h | 改进版批处理 | Improved batch processing |
| D12 | D12 | 项目 3:数据分析(Pandas) | Project 3: data analysis (Pandas) | 3h | 3h | 1 个分析脚本 | 1 analysis script |
| D13 | D13 | 项目 3 改进:加可视化 | Project 3 improve: add visualization | 2h | 2h | 带图表的分析 | Analysis with charts |
| D14 | D14 | 休息 + 复盘第 1 周 | Rest + review week 1 | 0.5h | 0.5h | 复盘笔记 | Review notes |
| D15 | D15 | 学 Git 基础(commit/branch) | Learn Git basics (commit/branch) | 2h | 2h | Git 仓库 | Git repo |
| D16 | D16 | 学 Git 进阶(merge/push) | Learn Git advanced (merge/push) | 2h | 2h | 推送到 GitHub | Pushed to GitHub |
| D17 | D17 | AI 协作:让 AI 写 + 你审查 | AI collaboration: AI writes + you review | 3h | 3h | 1 个协作项目 | 1 collaboration project |
| D18 | D18 | AI 协作:让 AI 找 bug + 修 | AI collaboration: AI finds + fixes bugs | 3h | 3h | bug 修复记录 | Bug fix record |
| D19 | D19 | AI 协作:让 AI 写测试 | AI collaboration: AI writes tests | 3h | 3h | 测试覆盖 | Test coverage |
| D20 | D20 | 学 HTTP/API 基础 | Learn HTTP/API basics | 2h | 2h | API 调用脚本 | API call script |
| D21 | D21 | 学 LLM API(调 Claude/ChatGPT) | Learn LLM API (call Claude/ChatGPT) | 3h | 3h | 1 个 AI 应用脚本 | 1 AI app script |
| D22 | D22 | 学 RAG 基础(向量 + 检索) | Learn RAG basics (vectors + retrieval) | 3h | 3h | 1 个 RAG 脚本 | 1 RAG script |
| D23 | D23 | 学 Prompt 工程(AI 编程专用) | Learn prompt engineering (for AI coding) | 2h | 2h | 提示词模板库 | Prompt template library |
| D24 | D24 | 休息 + 复盘第 2 周 | Rest + review week 2 | 0.5h | 0.5h | 复盘笔记 | Review notes |
| D25 | D25 | MVP:选 1 个真实问题 | MVP: pick 1 real problem | 1h | 1h | 问题定义 | Problem definition |
| D26 | D26 | MVP:设计 + 架构 | MVP: design + architecture | 2h | 2h | 架构文档 | Architecture doc |
| D27 | D27 | MVP:核心代码(AI 协作) | MVP: core code (AI collaboration) | 4h | 4h | 核心功能 | Core features |
| D28 | D28 | MVP:测试 + 审查 | MVP: test + review | 3h | 3h | 测试通过 | Tests pass |
| D29 | D29 | MVP:部署上线 | MVP: deploy online | 3h | 3h | 可访问的 MVP | Accessible MVP |
| D30 | D30 | MVP:复盘 + 写文档 | MVP: review + write docs | 2h | 2h | 复盘 + 文档 | Review + docs |
📅30 天日历的核心: 前 7 天基础 + 中 7 天项目 + 后 7 天 AI 协作 + 最后 9 天 MVP。每天 1-4 小时,30 天做出可上线的 MVP。The core of the 30-day calendar: first 7 days basics + middle 7 days projects + next 7 days AI collaboration + last 9 days MVP. 1-4 hours daily, make an online MVP in 30 days.
C12
40 个『AI 编程』真实案例(可复用路径)
40 'AI Coding' Real Cases (Reusable Paths)
40 个真实 AI 编程案例,拆解『从哪来、做什么、怎么用的 AI、结果』。
40 real AI-coding cases, deconstructing 'from where, what made, how AI used, outcome'.
| 案例 | Case | 做什么 | What made | AI 用法 | How AI used | 结果 | Outcome |
|---|---|---|---|---|---|---|---|
| A 老师做课件工具 | A teacher makes courseware tool | Python + Web 做课件生成器 | Python + Web courseware generator | AI 写核心 + 审查 | AI writes core + reviews | 成功,10 个学生用 | Success, 10 students use |
| B 会计做报表自动化 | B accountant automates reports | Python 脚本批处理报表 | Python batch report processing | AI 写脚本 + 测试 | AI writes script + tests | 成功,省 3 小时/周 | Success, saves 3h/week |
| C 运营做爬虫 | C operations builds crawler | Python 爬数据做监控 | Python crawl data for monitoring | AI 写爬虫 + 反爬处理 | AI writes crawler + anti-crawl handling | 成功,数据监控上线 | Success, data monitoring online |
| D 记者做数据新闻 | D journalist makes data journalism | Python 分析 + 可视化 | Python analysis + visualization | AI 分析 + 图表 | AI analysis + charts | 成功,1 篇报道 | Success, 1 article |
| E 设计师做作品集 | E designer makes portfolio | Web 前端 + 交互 | Web frontend + interaction | AI 写代码 + 审查 | AI writes code + reviews | 成功,1 个作品集站 | Success, 1 portfolio site |
| F 销售做 CRM | F sales builds CRM | Python + 数据库 + Web | Python + database + Web | AI 全栈 + 部署 | AI full-stack + deploy | 成功,小团队用 | Success, small team uses |
| G 厨师做订餐小程序 | G chef builds ordering mini-app | 微信小程序 + 后端 | WeChat mini-app + backend | AI 写代码 + 测试 | AI writes code + tests | 成功,月 100 单 | Success, 100 orders/month |
| H 记者做 AI 摘要 | J journalist makes AI summaries | Python + LLM API | Python + LLM API | AI 写 + 提示词调优 | AI writes + prompt tuning | 成功,日报自动化 | Success, daily report automated |
| I 老师做 AI 批改 | I teacher builds AI grading | Python + LLM 批改作文 | Python + LLM grades essays | AI 写 + 提示词 | AI writes + prompts | 成功,批改时间减半 | Success, grading time halved |
| J 会计做发票识别 | J accountant builds invoice OCR | Python + OCR + LLM | Python + OCR + LLM | AI 写 + 测试 | AI writes + tests | 成功,识别率 95% | Success, 95% accuracy |
| K 运营做 A/B 测试 | K operations builds A/B test | Python 统计 + 可视化 | Python statistics + visualization | AI 写 + 解释 | AI writes + explains | 成功,转化率 +15% | Success, +15% conversion |
| L 记者做事实核查 | L journalist builds fact-checking | Python + 搜索 API | Python + search API | AI 写 + 提示词 | AI writes + prompts | 成功,核查时间减半 | Success, check time halved |
| M 设计师做配色工具 | M designer makes color tool | Web 前端 + 算法 | Web frontend + algorithm | AI 写 + 审查 | AI writes + reviews | 成功,1000 用户 | Success, 1000 users |
| N 销售做话术生成 | N sales builds script generator | Python + LLM | Python + LLM | AI 写 + 提示词 | AI writes + prompts | 成功,成单率 +10% | Success, +10% close rate |
| O 厨师做营养计算 | O chef builds nutrition calculator | Web + 数据库 | Web + database | AI 写 + 测试 | AI writes + tests | 成功,小团队用 | Success, small team uses |
| P 记者做舆情监控 | P journalist builds sentiment monitor | Python + NLP | Python + NLP | AI 写 + 调优 | AI writes + tuning | 成功,实时监控 | Success, real-time monitoring |
| Q 老师做错题本 | Q teacher builds wrong-answer book | Python + Web | Python + Web | AI 写 + 测试 | AI writes + tests | 成功,50 学生用 | Success, 50 students use |
| R 会计做对账 | R accountant builds reconciliation | Python + 数据库 | Python + database | AI 写 + 测试 | AI writes + tests | 成功,对账时间减半 | Success, reconciliation time halved |
| S 运营做内容日历 | S operations builds content calendar | Web + 后端 | Web + backend | AI 写 + 部署 | AI writes + deploy | 成功,小团队用 | Success, small team uses |
| T 记者做采访整理 | T journalist builds interview organizer | Python + LLM | Python + LLM | AI 写 + 提示词 | AI writes + prompts | 成功,整理时间减半 | Success, organize time halved |
| U 设计师做图片批处理 | U designer builds image batch processing | Python + Pillow | Python + Pillow | AI 写 + 测试 | AI writes + tests | 成功,省 2 小时/周 | Success, saves 2h/week |
| V 销售做客户分层 | V sales builds customer segmentation | Python + 统计 | Python + statistics | AI 写 + 解释 | AI writes + explains | 成功,精准营销 | Success, precise marketing |
| W 厨师做菜谱生成 | W chef builds recipe generator | Web + LLM | Web + LLM | AI 写 + 提示词 | AI writes + prompts | 成功,100 用户 | Success, 100 users |
| X 记者做翻译校对 | X journalist builds translation check | Python + LLM | Python + LLM | AI 写 + 提示词 | AI writes + prompts | 成功,校对时间减半 | Success, check time halved |
| Y 老师做题库 | Y teacher builds question bank | Python + Web | Python + Web | AI 写 + 测试 | AI writes + tests | 成功,100 题 | Success, 100 questions |
| Z 会计做税务计算 | Z accountant builds tax calculator | Python + 数据库 | Python + database | AI 写 + 测试 | AI writes + tests | 成功,省 1 小时/天 | Success, saves 1h/day |
| A2 运营做用户画像 | A2 operations builds user profile | Python + 数据 | Python + data | AI 写 + 分析 | AI writes + analyzes | 成功,精准投放 | Success, precise targeting |
| B2 记者做多语言 | B2 journalist builds multilingual | Python + LLM | Python + LLM | AI 写 + 提示词 | AI writes + prompts | 成功,3 语言 | Success, 3 languages |
| C2 设计师做字体 | C2 designer builds font | Web + 字体 API | Web + font API | AI 写 + 审查 | AI writes + reviews | 成功,100 用户 | Success, 100 users |
| D2 销售做跟单 | D2 sales builds follow-up | Python + 数据库 | Python + database | AI 写 + 测试 | AI writes + tests | 成功,跟单效率 +30% | Success, +30% follow-up efficiency |
| E2 厨师做库存 | E2 chef builds inventory | Web + 后端 | Web + backend | AI 写 + 部署 | AI writes + deploy | 成功,小团队用 | Success, small team uses |
| F2 记者做选题 | F2 journalist builds topic selection | Python + 数据 | Python + data | AI 写 + 分析 | AI writes + analyzes | 成功,选题效率 +50% | Success, +50% topic efficiency |
| G2 老师做进度 | G2 teacher builds progress | Python + Web | Python + Web | AI 写 + 测试 | AI writes + tests | 成功,50 学生用 | Success, 50 students use |
| H2 会计做预算 | H2 accountant builds budget | Python + 数据 | Python + data | AI 写 + 分析 | AI writes + analyzes | 成功,预算准确率 +20% | Success, +20% budget accuracy |
| I2 运营做活动 | I2 operations builds campaign | Web + 后端 | Web + backend | AI 写 + 部署 | AI writes + deploy | 成功,1 个活动 | Success, 1 campaign |
| J2 记者做数据 | J2 journalist builds data | Python + 可视化 | Python + visualization | AI 写 + 解释 | AI writes + explains | 成功,1 篇数据报道 | Success, 1 data article |
| K2 设计师做插画 | K2 designer builds illustration | Python + 图像 | Python + image | AI 写 + 调优 | AI writes + tuning | 成功,100 幅 | Success, 100 pieces |
| L2 销售做报价 | L2 sales builds quotation | Python + 数据库 | Python + database | AI 写 + 测试 | AI writes + tests | 成功,报价效率 +40% | Success, +40% quotation efficiency |
| M2 厨师做排班 | M2 chef builds scheduling | Web + 算法 | Web + algorithm | AI 写 + 审查 | AI writes + reviews | 成功,小团队用 | Success, small team uses |
| N2 记者做版权 | N2 journalist builds copyright | Python + 搜索 | Python + search | AI 写 + 提示词 | AI writes + prompts | 成功,版权检查 | Success, copyright check |
| O2 老师做评估 | O2 teacher builds assessment | Python + 数据 | Python + data | AI 写 + 分析 | AI writes + analyzes | 成功,评估效率 +30% | Success, +30% assessment efficiency |
📌40 个案例的共同点:① 都不是专业程序员;② 都用 AI 写核心代码 + 自己审查;③ 都解决一个真实问题;④ 都在 30-90 天完成。这是可复用的路径。Common points of 40 cases: ① none are professional programmers; ② all use AI to write core code + self-review; ③ all solve a real problem; ④ all complete in 30-90 days. This is a reusable path.
C13
60 个『调试』高频场景(排错手册)
60 'Debugging' High-Frequency Scenarios (Troubleshooting Handbook)
60 个编程时最常遇到的调试场景,给出排查思路。
60 most common debugging scenarios in coding, with troubleshooting approaches.
| 场景 | Scenario | 排查思路 | Troubleshooting approach |
|---|---|---|---|
| 程序崩溃(崩溃栈) | Crash (stack trace) | 看崩溃栈最后一行 + 定位变量 + 检查空值 | See last line of crash stack + locate variable + check null |
| 空指针异常 | Null pointer exception | 检查变量是否赋值 + 加空值判断 | Check if variable assigned + add null check |
| 索引越界 | Index out of bounds | 检查循环边界 + 数据长度 | Check loop bounds + data length |
| 无限循环 | Infinite loop | 检查循环条件 + 是否有退出 + 打印循环变量 | Check loop condition + has exit + print loop variable |
| 变量未定义 | Variable undefined | 检查拼写 + 作用域 + 定义顺序 | Check spelling + scope + definition order |
| 类型错误 | Type error | 检查数据类型 + 用 str/int/float 转换 | Check data type + use str/int/float conversion |
| 编码错误 | Encoding error | 检查文件编码 + 用 utf-8 + 处理 BOM | Check file encoding + use utf-8 + handle BOM |
| 文件找不到 | File not found | 检查路径 + 相对/绝对路径 + 文件存在 | Check path + relative/absolute + file exists |
| 权限错误 | Permission error | 检查文件权限 + 用 chmod + 检查用户 | Check file permission + use chmod + check user |
| 端口占用 | Port in use | 检查占用进程 + 换端口 + 杀进程 | Check occupying process + change port + kill process |
| 依赖缺失 | Missing dependency | 检查 import + 装依赖 + 检查版本 | Check import + install dependency + check version |
| 版本不兼容 | Version incompatible | 检查版本 + 用兼容版本 + 看文档 | Check version + use compatible version + read docs |
| API 404 | API 404 | 检查 URL + 方法 + 参数 + 认证 | Check URL + method + params + auth |
| API 401/403 | API 401/403 | 检查 Token + 权限 + 过期 | Check Token + permission + expiry |
| API 超时 | API timeout | 检查网络 + 加超时 + 重试 | Check network + add timeout + retry |
| 内存溢出 | Memory overflow | 检查数据结构大小 + 分页 + 检查泄漏 | Check data structure size + paginate + check leak |
| CPU 100% | CPU 100% | 检查循环 + 算法复杂度 + profiling | Check loop + algorithm complexity + profiling |
| 磁盘满 | Disk full | 检查日志大小 + 清理 + 扩容 | Check log size + clean + expand |
| 数据库连接失败 | Database connection failed | 检查地址 + 端口 + 认证 + 网络 | Check address + port + auth + network |
| SQL 语法错误 | SQL syntax error | 检查 SQL 语句 + 引号 + 分号 | Check SQL statement + quotes + semicolon |
| SQL 死锁 | SQL deadlock | 检查事务 + 锁顺序 + 隔离级别 | Check transaction + lock order + isolation level |
| 并发问题 | Concurrency issue | 检查竞态 + 加锁 + 用线程安全结构 | Check race + add lock + use thread-safe structure |
| 死锁 | Deadlock | 检查锁顺序 + 加超时 + 用 try-finally | Check lock order + add timeout + use try-finally |
| 竞态条件 | Race condition | 检查共享变量 + 加锁 + 用原子操作 | Check shared variable + add lock + use atomic operation |
| 缓存不一致 | Cache inconsistency | 检查缓存失效 + 一致性 + 过期时间 | Check cache invalidation + consistency + expiry time |
| 数据丢失 | Data loss | 检查事务 + 备份 + 日志 | Check transaction + backup + logs |
| 数据重复 | Data duplication | 检查唯一约束 + 幂等 + 去重 | Check unique constraint + idempotent + dedup |
| 时区问题 | Timezone issue | 检查时区 + 用 UTC + 转换 | Check timezone + use UTC + convert |
| 日期解析 | Date parsing | 检查格式 + 时区 + 边界 | Check format + timezone + boundary |
| 浮点精度 | Floating point precision | 用 Decimal + 避免浮点比较 | Use Decimal + avoid float comparison |
| 大数溢出 | Big number overflow | 用大数类型 + 检查范围 | Use big number type + check range |
| 字符串转义 | String escape | 检查引号 + 转义 + 用原始字符串 | Check quotes + escape + use raw string |
| 正则错误 | Regex error | 检查模式 + 测试 + 用调试工具 | Check pattern + test + use debug tool |
| JSON 解析 | JSON parsing | 检查格式 + 引号 + 嵌套 + 编码 | Check format + quotes + nesting + encoding |
| XML 解析 | XML parsing | 检查标签 + 闭合 + 编码 | Check tags + closing + encoding |
| CSV 错误 | CSV error | 检查分隔符 + 引号 + 编码 + 换行 | Check delimiter + quotes + encoding + newline |
| HTTP 编码 | HTTP encoding | 检查 Content-Type + 字符集 + 压缩 | Check Content-Type + charset + compression |
| 跨域错误 | CORS error | 检查 CORS 配置 + 跨域 + 预检 | Check CORS config + cross-origin + preflight |
| Cookie 问题 | Cookie issue | 检查域名 + 路径 + 过期 + Secure | Check domain + path + expiry + Secure |
| 会话丢失 | Session lost | 检查会话存储 + 过期 + 重启 | Check session storage + expiry + restart |
| 文件上传 | File upload | 检查大小 + 类型 + 权限 + 临时目录 | Check size + type + permission + temp dir |
| 图片处理 | Image processing | 检查格式 + 尺寸 + 内存 + 库版本 | Check format + size + memory + library version |
| 网络请求 | Network request | 检查 URL + 方法 + 头 + 代理 | Check URL + method + headers + proxy |
| DNS 解析 | DNS resolution | 检查域名 + DNS + 网络 + 超时 | Check domain + DNS + network + timeout |
| SSL 错误 | SSL error | 检查证书 + 过期 + 信任 + 版本 | Check certificate + expiry + trust + version |
| 防火墙 | Firewall | 检查端口 + 规则 + 安全组 | Check port + rules + security group |
| 负载均衡 | Load balancer | 检查健康检查 + 会话保持 + 后端 | Check health check + session sticky + backend |
| 容器化 | Containerization | 检查镜像 + 端口 + 卷 + 网络 | Check image + port + volume + network |
| Kubernetes | Kubernetes | 检查 Pod + 服务 + 副本 + 资源 | Check Pod + service + replicas + resources |
| CI/CD 失败 | CI/CD failure | 检查日志 + 步骤 + 环境 + 依赖 | Check logs + steps + environment + dependency |
| 部署失败 | Deployment failure | 检查配置 + 权限 + 资源 + 日志 | Check config + permission + resource + logs |
| 回滚 | Rollback | 检查版本 + 数据 + 配置 + 回滚脚本 | Check version + data + config + rollback script |
| 监控报警 | Monitoring alert | 检查指标 + 阈值 + 根因 + 缓解 | Check metric + threshold + root cause + mitigation |
| 日志分析 | Log analysis | 检查时间 + 错误 + 频率 + 关联 | Check time + error + frequency + correlation |
| 性能分析 | Performance analysis | 用 profiling + 热点 + 优化 | Use profiling + hotspot + optimize |
| 内存泄漏 | Memory leak | 检查引用 + GC + 监控 + 释放 | Check reference + GC + monitor + release |
| 代码审查 | Code review | 看逻辑 + 边界 + 性能 + 安全 | Check logic + boundary + performance + security |
🔧60 个调试场景的共同思路:先看错误信息(栈/日志)→ 定位 → 最小复现 → 修复 → 验证。别凭感觉改,要凭证据改。Common approach of 60 debugging scenarios: first see error info (stack/logs) → locate → minimum reproduce → fix → verify. Don't change by feeling, change by evidence.
C14
30 个『AI 编程』实战场景(具体动作)
30 'AI Coding' Real-World Scenarios (Concrete Actions)
30 个编程时的具体场景,每个给出『当下该做什么』。
30 concrete scenarios in coding, each with 'what to do now'.
| 场景 | Scenario | 当下该做 | What to do now | 为什么 | Why |
|---|---|---|---|---|---|
| AI 给的代码跑不起来 | AI's code doesn't run | 看错误信息 + 贴给 AI 排查 + 最小复现 | See error + paste to AI + minimum reproduce | 错误信息是最快的线索 | Error is the fastest clue |
| AI 给的代码有 bug | AI's code has a bug | 审查 + 加测试 + 让 AI 找 bug | Review + add tests + let AI find bugs | AI 会错,你是防线 | AI errs; you're the defense |
| 不知道用什么技术栈 | Don't know which tech stack | 让 AI 给 3 个方案 + 对比 + 选稳定的 | Let AI give 3 options + compare + choose stable | 先调研再动手 | Research first, then act |
| 代码写了一半卡住 | Code stuck halfway | 把卡点描述给 AI + 给上下文 + 要具体方案 | Describe the stuck point to AI + give context + ask for specific plan | AI 擅长解决具体卡点 | AI is good at solving specific stuck points |
| 需求不清晰 | Requirements unclear | 先写需求文档 + 让 AI 帮你澄清 + 确认边界 | Write requirements doc + let AI help clarify + confirm boundaries | 需求清晰 = 代码清晰 | Clear requirements = clear code |
| 不知道怎么写测试 | Don't know how to write tests | 让 AI 按正常/边界/异常写 + 你审查 | Let AI write by normal/boundary/exception + you review | AI 能快速生成测试模板 | AI can quickly generate test templates |
| 代码太乱想重构 | Code too messy, want to refactor | 先加测试 + 让 AI 重构 + 验证功能不变 | First add tests + let AI refactor + verify functionality unchanged | 测试是重构的安全网 | Tests are the safety net for refactoring |
| 性能慢想优化 | Performance slow, want to optimize | 先 profiling 找瓶颈 + 让 AI 分析 + 针对性优化 | First profiling to find bottleneck + let AI analyze + targeted optimization | 不 profiling 不优化 | No optimization without profiling |
| 不知道架构怎么设计 | Don't know how to design architecture | 让 AI 给 3 种架构 + 对比 + 选简单的 | Let AI give 3 architectures + compare + choose simple | 简单 = 好维护 | Simple = easy to maintain |
| 数据库设计不会 | Can't design database | 让 AI 给表结构 + 你审查 + 加索引 | Let AI give table structure + you review + add indexes | 先设计后写代码 | Design first, then code |
| API 对接不会 | Can't integrate API | 看 API 文档 + 让 AI 写调用 + 测试 | Read API doc + let AI write call + test | 文档是最好的老师 | Docs are the best teacher |
| 部署上线不会 | Can't deploy online | 让 AI 给部署方案 + 一步步做 + 验证 | Let AI give deployment plan + do step by step + verify | 部署是最后一公里 | Deployment is the last mile |
| 代码不会管理 | Can't manage code | 用 Git + 小步提交 + 有意义的 commit | Use Git + small commits + meaningful commit messages | 可追溯是底线 | Traceability is the bottom line |
| 多人协作不会 | Can't collaborate with multiple people | 用 Git 分支 + PR + 代码审查 | Use Git branches + PR + code review | 协作靠流程不靠吼 | Collaboration relies on process, not shouting |
| 不知道选开源库 | Don't know which open-source library | 让 AI 推荐 + 对比维护活跃度 + 选稳定的 | Let AI recommend + compare maintenance activity + choose stable | 选稳定不选新 | Choose stable, not new |
| 许可证不知道 | Don't know about licenses | 让 AI 解释 + 选宽松许可证 + 看商用 | Let AI explain + choose permissive license + check commercial use | 许可证是底线 | License is the bottom line |
| 安全漏洞不知道 | Don't know about security vulnerabilities | 让 AI 扫描 + 检查常见漏洞 + 修复 | Let AI scan + check common vulnerabilities + fix | 安全是底线 | Security is the bottom line |
| 数据隐私不知道 | Don't know about data privacy | 让 AI 解释 + 最小化收集 + 加密 | Let AI explain + minimize collection + encrypt | 隐私是底线 | Privacy is the bottom line |
| 不知道怎么写文档 | Don't know how to write docs | 让 AI 给文档模板 + 你补充 + 保持更新 | Let AI give doc template + you supplement + keep updated | 文档是长期价值 | Docs are long-term value |
| 不知道怎么写 README | Don't know how to write README | 让 AI 给模板 + 写清楚安装/使用/示例 | Let AI give template + write clear install/usage/examples | README 是脸面 | README is the face |
| 不知道怎么做 CI/CD | Don't know how to do CI/CD | 让 AI 给方案 + 从简单开始 + 逐步加 | Let AI give plan + start simple + gradually add | 自动化是效率 | Automation is efficiency |
| 不知道怎么做监控 | Don't know how to do monitoring | 让 AI 给方案 + 监控关键指标 + 设报警 | Let AI give plan + monitor key metrics + set alerts | 监控是安全网 | Monitoring is the safety net |
| 不知道怎么做日志 | Don't know how to do logging | 让 AI 给方案 + 关键节点打日志 + 集中管理 | Let AI give plan + log at key points + centralized management | 日志是排查基础 | Logs are the foundation of troubleshooting |
| 不知道怎么做缓存 | Don't know how to do caching | 让 AI 给方案 + 缓存热点 + 设过期 | Let AI give plan + cache hotspots + set expiry | 缓存是性能 | Cache is performance |
| 不知道怎么做并发 | Don't know how to do concurrency | 让 AI 解释 + 用线程/协程 + 加锁 | Let AI explain + use threads/coroutines + add locks | 并发要小心 | Concurrency needs care |
| 不知道怎么做队列 | Don't know how to do queues | 让 AI 解释 + 用消息队列 + 解耦 | Let AI explain + use message queue + decouple | 队列是解耦 | Queue is decoupling |
| 不知道怎么做限流 | Don't know how to do rate limiting | 让 AI 给方案 + 令牌桶 + 保护后端 | Let AI give plan + token bucket + protect backend | 限流是保护 | Rate limiting is protection |
| 不知道怎么做幂等 | Don't know how to do idempotency | 让 AI 解释 + 用唯一 ID + 去重 | Let AI explain + use unique ID + dedup | 幂等是可靠 | Idempotency is reliability |
| 不知道怎么做降级 | Don't know how to do degradation | 让 AI 给方案 + 非核心降级 + 保护核心 | Let AI give plan + degrade non-core + protect core | 降级是兜底 | Degradation is the fallback |
| 不知道怎么做熔断 | Don't know how to do circuit breaking | 让 AI 给方案 + 熔断非核心 + 快速失败 | Let AI give plan + circuit-break non-core + fail fast | 熔断是保护 | Circuit breaking is protection |
🎯30 个场景的共同点: 编程时遇到不会的, 第一反应是『让 AI 给方案 + 我审查 + 我验证』。不是自己闷头猜, 也不是全信 AI。Common point of 30 scenarios: when stuck in coding, the first reaction is 'let AI give a plan + I review + I verify'. Not guess silently, not fully trust AI.
C15
30 条『AI 编程』行动清单(直接做)
30 'AI Coding' Action Checklist (Just Do It)
30 个可直接执行的 AI 编程动作,按周分组。
30 directly executable AI-coding actions, grouped by week.
| 周 | Week | 动作 | Action | 产出 | Output |
|---|---|---|---|---|---|
| W1 | W1 | 装 Python + 跑 Hello World | Install Python + run Hello World | 能运行 | Can run |
| W1 | W1 | 学变量/类型/运算 | Learn variables/types/operations | 5 个脚本 | 5 scripts |
| W1 | W1 | 学条件/循环 | Learn conditionals/loops | 3 个脚本 | 3 scripts |
| W1 | W1 | 学函数/作用域 | Learn functions/scope | 3 个练习 | 3 exercises |
| W1 | W1 | 学列表/字典 | Learn lists/dicts | 3 个练习 | 3 exercises |
| W1 | W1 | 学文件读写 | Learn file read/write | 1 个脚本 | 1 script |
| W1 | W1 | 学异常处理 | Learn exception handling | 复习笔记 | Review notes |
| W2 | W2 | 项目 1: 爬虫 | Project 1: crawler | 可运行爬虫 | Runnable crawler |
| W2 | W2 | 项目 1 改进 | Project 1 improve | 改进版 | Improved |
| W2 | W2 | 项目 2: 文件批处理 | Project 2: batch file processing | 可运行 | Runnable |
| W2 | W2 | 项目 2 改进 | Project 2 improve | 改进版 | Improved |
| W2 | W2 | 项目 3: 数据分析 | Project 3: data analysis | 1 个脚本 | 1 script |
| W2 | W2 | 项目 3 改进: 可视化 | Project 3 improve: visualization | 带图表 | With charts |
| W2 | W2 | 休息 + 复盘 | Rest + review | 复盘笔记 | Review notes |
| W3 | W3 | 学 Git 基础 | Learn Git basics | Git 仓库 | Git repo |
| W3 | W3 | 学 Git 进阶 | Learn Git advanced | 推送 GitHub | Pushed to GitHub |
| W3 | W3 | AI 协作: 写 + 审查 | AI collab: write + review | 1 个协作项目 | 1 collab project |
| W3 | W3 | AI 协作: 找 bug | AI collab: find bugs | bug 修复记录 | Bug fix record |
| W3 | W3 | AI 协作: 写测试 | AI collab: write tests | 测试覆盖 | Test coverage |
| W3 | W3 | 学 HTTP/API 基础 | Learn HTTP/API basics | API 脚本 | API script |
| W3 | W3 | 学 LLM API | Learn LLM API | 1 个 AI 应用 | 1 AI app |
| W3 | W3 | 学 RAG 基础 | Learn RAG basics | 1 个 RAG 脚本 | 1 RAG script |
| W4 | W4 | 学 Prompt 工程 | Learn prompt engineering | 提示词模板库 | Prompt template library |
| W4 | W4 | MVP: 选真实问题 | MVP: pick real problem | 问题定义 | Problem definition |
| W4 | W4 | MVP: 设计 + 架构 | MVP: design + architecture | 架构文档 | Architecture doc |
| W4 | W4 | MVP: 核心代码 | MVP: core code | 核心功能 | Core features |
| W4 | W4 | MVP: 测试 + 审查 | MVP: test + review | 测试通过 | Tests pass |
| W4 | W4 | MVP: 部署上线 | MVP: deploy online | 可访问 | Accessible |
| W4 | W4 | MVP: 复盘 + 文档 | MVP: review + docs | 复盘 + 文档 | Review + docs |
| W4 | W4 | 写 README + 开源 | Write README + open source | 开源项目 | Open-source project |
✅30 个动作按周做, 4 周完成。核心是『项目驱动 + AI 协作 + 审查测试』。完成 30 个动作, 你就能做出可上线的 MVP。Do the 30 actions by week, finish in 4 weeks. Core is 'project-driven + AI collaboration + review+test'. Complete the 30 actions, and you can make an online MVP.
C16
30 个『AI 编程』扩充问答
30 Extended 'AI Coding' Q&A
30 个更细的 AI 编程问题,覆盖进阶/架构/安全/部署。
30 finer AI-coding questions, covering advanced/architecture/security/deployment.
传统: 人写每一行。AI: 人定义 + AI 生成 + 人审查。人从『写』转向『判断』。Traditional: write every line. AI: human defines + AI generates + human reviews. Human shifts from 'writing' to 'judging'.
具体: 功能 + 约束 + 技术栈 + 示例 + 期望输出。越具体越好。Specific: function + constraints + tech stack + examples + expected output. More specific, the better.
分步生成: 先架构 + 再核心 + 再边界 + 再测试。逐步审查。Generate step by step: architecture first + then core + then boundaries + then tests. Review progressively.
贴相关上下文 + 说明目标 + 标注关键函数。只贴片段不够。Paste relevant context + explain goal + mark key functions. Just pasting snippets isn't enough.
要审查。AI 测试可能只测 happy path。自己加边界/异常。Review. AI tests may only cover the happy path. Add boundaries/exceptions yourself.
3 要点: 逻辑对不对 + 边界处理没 + 性能够不够。逐行过关键逻辑。3 points: logic right + boundaries handled + performance enough. Go through key logic line by line.
重要。AI 能写代码, 不能替你判断算法对不对。算法是判断力。Important. AI can write code, but can't judge for you whether the algorithm is right. Algorithms are judgment.
重要。选对数据结构 = 性能/内存的基础。AI 不会替你选。Important. Choosing the right data structure = the base of performance/memory. AI won't choose for you.
进阶要学。内存/进程/网络是理解性能的前提。Learn for advanced. Memory/process/network are prerequisites for understanding performance.
基础: HTTP/DNS/TCP/SSL。进阶: 负载均衡/CDN/微服务。Basic: HTTP/DNS/TCP/SSL. Advanced: load balancing/CDN/microservices.
1 个关系型 (PostgreSQL) + 1 个 NoSQL (MongoDB) + 1 个缓存 (Redis)。1 relational (PostgreSQL) + 1 NoSQL (MongoDB) + 1 cache (Redis).
基础: CRUD + JOIN + 索引。进阶: 优化/事务/并发。Basic: CRUD + JOIN + indexes. Advanced: optimization/transactions/concurrency.
基础: HTML/CSS/JS。进阶: 一个框架 (React/Vue) + 工程化。Basic: HTML/CSS/JS. Advanced: a framework (React/Vue) + engineering.
Python: FastAPI/Django。Node: Express。选生态大的。Python: FastAPI/Django. Node: Express. Choose the one with a large ecosystem.
进阶要学。先单体跑通, 再拆微服务。别一开始就上。Learn for advanced. Get the monolith working first, then split into microservices. Don't start with it.
要学。部署的基础。容器化让环境一致。Learn it. It's the foundation of deployment. Containerization makes environments consistent.
进阶要学。Docker 用熟后再学。从理解 Pod/Service 开始。Learn for advanced. Learn after mastering Docker. Start from understanding Pod/Service.
看生态 + 成本。国内: 阿里云。海外: AWS。入门够用即可。Look at ecosystem + cost. China: Ali Cloud. Overseas: AWS. Enough for entry.
进阶要学。自动化测试 + 部署。从 GitHub Actions 开始。Learn for advanced. Automated testing + deployment. Start with GitHub Actions.
进阶要学。CI/CD + 监控 + 自动化。先会基础再深入。Learn for advanced. CI/CD + monitoring + automation. Learn basics first, then deep.
监控 3 类: 资源 (CPU/内存) + 应用 (QPS/延迟) + 业务 (成功/失败)。Monitor 3 types: resources (CPU/memory) + app (QPS/latency) + business (success/failure).
级别 (DEBUG/INFO/WARN/ERROR) + 结构化 + 集中收集 + 保留策略。Levels (DEBUG/INFO/WARN/ERROR) + structured + centralized collection + retention policy.
缓存热点 + 设过期 + 一致性策略 (Cache-Aside)。Cache hotspots + set expiry + consistency strategy (Cache-Aside).
令牌桶/漏桶 + 按用户/IP 限流 + 保护后端。Token bucket/leaky bucket + rate limit by user/IP + protect backend.
唯一 ID + 去重表 + 状态机。防止重复操作。Unique ID + dedup table + state machine. Prevent duplicate operations.
非核心功能可降级 + 返回兜底数据 + 保护核心。Non-core features can degrade + return fallback data + protect core.
失败率超阈值就熔断 + 快速失败 + 定期恢复。Circuit-break when failure rate exceeds threshold + fail fast + periodic recovery.
JWT + 密码哈希 + 多因素 + HTTPS。别自造轮子。JWT + password hashing + multi-factor + HTTPS. Don't reinvent the wheel.
传输: TLS。存储: AES。密钥: 密钥管理服务。Transmission: TLS. Storage: AES. Keys: key management service.
最小化收集 + 匿名化 + 合规 (GDPR/个保法) + 加密。Minimize collection + anonymize + comply (GDPR/PIPL) + encrypt.
💡30 个进阶问题的核心: 先让东西『跑起来』, 再让它『跑得好』, 最后让它『跑得稳』。顺序不能反。The core of the 30 advanced questions: first make it 'work', then make it 'work well', finally make it 'work stably'. The order can't be reversed.
C17
30 个『编程』术语中英对照(扫盲表)
30 'Programming' Terms: Chinese-English Glossary
30 个编程高频术语, 中英对照 + 一句话解释, 扫盲用。
30 high-frequency programming terms, bilingual + one-line definition, for beginners.
| 术语 | Term | 一句话解释 | One-line definition | 英文解释 | English definition |
|---|---|---|---|---|---|
| 变量 | Variable | 存数据的命名容器 | Named container holding data | ||
| 数据类型 | Data type | 数据的种类 (字符串/数字/布尔) | Kinds of data (string/number/boolean) | ||
| 函数 | Function | 可复用的代码块 | Reusable block of code | ||
| 循环 | Loop | 重复执行代码 | Repeat code execution | ||
| 条件 | Condition | 根据判断走不同分支 | Take different branches based on judgment | ||
| 作用域 | Scope | 变量可见的范围 | The visible range of a variable | ||
| 异常 | Exception | 运行时的错误事件 | A runtime error event | ||
| 调试 | Debugging | 找并修 bug 的过程 | The process of finding and fixing bugs | ||
| 断点 | Breakpoint | 暂停执行检查状态 | Pause execution to inspect state | ||
| 堆栈 | Stack | 后进先出的数据结构 | Last-in-first-out data structure | ||
| 队列 | Queue | 先进先出的数据结构 | First-in-first-out data structure | ||
| 哈希 | Hash | 快速查找的映射结构 | Fast lookup mapping structure | ||
| 递归 | Recursion | 函数调用自身 | Function calling itself | ||
| 闭包 | Closure | 捕获外层变量的函数 | Function capturing outer variables | ||
| 多态 | Polymorphism | 同一接口不同实现 | Same interface, different implementations | ||
| 继承 | Inheritance | 子类复用父类 | Subclass reusing parent class | ||
| 封装 | Encapsulation | 隐藏内部, 暴露接口 | Hide internals, expose interface | ||
| 接口 | Interface | 约定的方法集合 | A set of agreed methods | ||
| 对象 | Object | 类的实例 | An instance of a class | ||
| 类 | Class | 对象的模板 | Template for objects | ||
| 模块 | Module | 组织代码的单位 | Unit organizing code | ||
| 包 | Package | 组织模块的单位 | Unit organizing modules | ||
| 依赖 | Dependency | 代码依赖的库/服务 | Libraries/services the code relies on | ||
| 库 | Library | 可复用的代码集合 | A set of reusable code | ||
| 框架 | Framework | 规定结构的代码骨架 | Code skeleton prescribing structure | ||
| API | API | 系统间通信的约定 | The convention for system-to-system communication | ||
| SDK | SDK | 开发工具包 | Software development kit | ||
| 编译器 | Compiler | 把代码翻译成机器语言 | Translate code into machine language | ||
| 解释器 | Interpreter | 逐行执行代码 | Execute code line by line | ||
| 运行时 | Runtime | 程序执行的阶段/环境 | The phase/environment of program execution |
📖30 个术语不用全记。最核心的 8 个: 变量 + 函数 + 循环 + 异常 + 堆栈 + 递归 + 封装 + API。理解这 8 个, 编程的骨架就通了。No need to memorize all 30. The 8 most core: variable + function + loop + exception + stack + recursion + encapsulation + API. Understand these 8 and the skeleton of coding clicks.
C18
30 个『AI 编程』项目实战模板(从想法到上线)
30 'AI Coding' Project Templates (From Idea to Launch)
30 个真实项目模板,每个给出『步骤 + 提示词 + 工具』,照着做就能上线。
30 real project templates, each with 'steps + prompts + tools', follow to launch.
| 项目 | 步骤(AI 协作) | 核心提示词 | 工具 |
|---|---|---|---|
| 个人博客 | ① 让 AI 生成静态站点结构 ② 写内容 ③ 部署到 GitHub Pages | 帮我生成一个 Jekyll 博客主题,支持中文,简洁 | Jekyll / GitHub Pages |
| 待办应用 | ① AI 生成前端 ② 加本地存储 ③ 加拖拽排序 | 帮我做一个待办应用,HTML+JS,支持 localStorage 保存 | VS Code + AI |
| 数据可视化 | ① AI 读 CSV ② 生成 ECharts 图表 ③ 加交互 | 帮我用 ECharts 可视化这份销售数据,加时间筛选 | Python / ECharts |
| API 客户端 | ① AI 生成 fetch 封装 ② 加错误处理 ③ 加缓存 | 帮我写一个 fetch 封装,支持超时、重试、缓存 | Node.js / fetch |
| 表单验证 | ① AI 生成验证规则 ② 加实时反馈 ③ 加国际化 | 帮我做一个表单验证,支持手机号/邮箱/身份证,实时反馈 | HTML + JS |
| 聊天机器人 UI | ① AI 生成对话界面 ② 接入 API ③ 加流式输出 | 帮我做一个聊天界面,支持流式显示 AI 回复 | React / ChatGPT API |
| 爬虫 | ① AI 写解析逻辑 ② 加反爬处理 ③ 存数据库 | 帮我写一个爬虫,抓取 XX 网站标题和链接,存 SQLite | Python + requests |
| 数据清洗 | ① AI 读脏数据 ② 写清洗规则 ③ 输出干净数据 | 帮我清洗这份数据,处理缺失值、重复、格式统一 | Python + pandas |
| 定时任务 | ① AI 写任务逻辑 ② 加 cron ③ 加日志 | 帮我做一个定时任务,每天 8 点发报告,带日志 | Node.js / cron |
| 邮件自动化 | ① AI 写模板 ② 接入 SMTP ③ 加批量 | 帮我写一个邮件发送脚本,支持批量、模板变量 | Python + smtplib |
| 图片处理 | ① AI 写批量缩放 ② 加水印 ③ 转格式 | 帮我批量缩放 1000 张图到 800px,加水印,转 WebP | Python + Pillow |
| PDF 处理 | ① AI 提取文本 ② 合并/拆分 ③ 加书签 | 帮我提取 PDF 文本,合并多个 PDF,加目录 | Python + PyPDF2 |
| Excel 自动化 | ① AI 读表格 ② 写计算规则 ③ 输出报表 | 帮我读取这份 Excel,计算月度汇总,输出新表 | Python + openpyxl |
| 报表生成 | ① AI 写模板 ② 填数据 ③ 导出 | 帮我生成月度报表,包含图表,导出 PDF | Python + reportlab |
| 文本分析 | ① AI 读文本 ② 词频/情感 ③ 可视化 | 帮我分析这份评论数据,做情感分析和词云 | Python + jieba |
| 日志分析 | ① AI 解析日志 ② 统计错误 ③ 报警 | 帮我解析 Nginx 日志,统计 Top 错误,超阈值报警 | Python + 正则 |
| 监控脚本 | ① AI 写检查逻辑 ② 定时运行 ③ 通知 | 帮我写一个监控脚本,检查服务状态,异常发飞书 | Python + requests |
| 备份脚本 | ① AI 写压缩 ② 定时 ③ 上传云 | 帮我写一个备份脚本,每天压缩网站数据,上传 S3 | Bash / Python |
| 部署脚本 | ① AI 写打包 ② SSH 上传 ③ 重启服务 | 帮我写一个部署脚本,打包后 SSH 到服务器,重启 | Bash / SSH |
| 数据库迁移 | ① AI 写迁移脚本 ② 测试 ③ 回滚 | 帮我写一个数据库迁移脚本,支持测试和回滚 | Python / SQL |
| 配置管理 | ① AI 生成 .env 模板 ② 加校验 ③ 文档 | 帮我生成配置模板,区分开发/生产,加文档 | .env / dotenv |
| 代码生成器 | ① AI 写模板 ② 变量替换 ③ 输出 | 帮我做一个代码生成器,输入名称生成 CRUD 代码 | Python / Jinja2 |
| 脚手架 | ① AI 生成项目结构 ② 装依赖 ③ 初始化 | 帮我生成一个 React 项目脚手架,含路由、状态管理 | create-react-app |
| API 文档 | ① AI 读代码 ② 生成 Swagger ③ 部署 | 帮我从代码生成 Swagger 文档,支持在线调试 | Swagger / FastAPI |
| 测试生成 | ① AI 读函数 ② 生成单测 ③ 跑覆盖率 | 帮我给这个函数生成单元测试,覆盖边界情况 | pytest / Jest |
| 性能优化 | ① AI 读瓶颈 ② 写优化 ③ 对比基准 | 帮我优化这个慢查询,给前后对比基准 | SQL / 索引 |
| 国际化 | ① AI 提取字符串 ② 翻译 ③ 切换 | 帮我把这个网站做成中英双语,切换按钮在右上 | i18n / JSON |
| 暗色模式 | ① AI 写 CSS 变量 ② 切换 ③ 记忆 | 帮我加暗色模式,跟随系统,记忆用户选择 | CSS 变量 / localStorage |
| PWA 化 | ① AI 加 manifest ② Service Worker ③ 离线 | 帮我把这个网站变成 PWA,支持离线和安装 | PWA / SW |
| SEO 优化 | ① AI 生成 meta ② 加结构化数据 ③ sitemap | 帮我优化 SEO,加 JSON-LD、sitemap、语义化 | Meta / JSON-LD |
30 templates cover 'blog/app/data/API/crawler/report/monitor/deploy'. Each gives 'steps + prompts + tools', follow to launch. This is AI-era coding: not from scratch, but assembling + collaborating.项目模板Project Templates
C19
40 个『AI 编程』扩充案例
40 Extended 'AI Coding' Cases
40 个更细的 AI 编程案例,覆盖不同技术栈和场景。
40 finer AI-coding cases, covering different stacks and scenarios.
| 案例 | 技术栈 | AI 做了什么 | 结果 |
|---|---|---|---|
| 用 AI 一天做完待办 App | HTML+JS+localStorage | AI 生成全部前端代码 + 本地存储逻辑 | 从想法到可用 App 只用 1 天 |
| AI 帮我把 Excel 变成交互报表 | Python+ECharts | AI 写数据读取 + 图表 + 交互 | 原本 2 天的报表 2 小时搞定 |
| 用 AI 写爬虫抓 10 万条数据 | Python+requests | AI 写解析 + 反爬 + 存 SQLite | 3 小时抓完 10 万条 |
| AI 帮我优化慢查询 | SQL | AI 分析执行计划,加索引 | 查询从 5 秒降到 50ms |
| 用 AI 做网站国际化 | i18n+JSON | AI 提取字符串 + 生成中英文 | 一天完成中英双语 |
| AI 帮我写测试 | pytest | AI 读函数生成单测 | 覆盖率从 30% 到 85% |
| 用 AI 做 PDF 批量处理 | PyPDF2 | AI 写提取/合并/拆分脚本 | 1000 个 PDF 批量处理 10 分钟 |
| AI 帮我写部署脚本 | Bash+SSH | AI 写打包 + 上传 + 重启 | 部署从 20 分钟到 1 分钟 |
| 用 AI 做数据清洗 | pandas | AI 写缺失值/重复/格式处理 | 脏数据变干净,1 小时 |
| AI 帮我做日志监控 | Python+正则 | AI 写解析 + 统计 + 报警 | 异常 5 分钟内通知到飞书 |
| 用 AI 写邮件自动化 | smtplib | AI 写模板 + 批量发送 | 1000 封邮件 10 分钟发完 |
| AI 帮我做图片批量处理 | Pillow | AI 写缩放/水印/转格式 | 1000 张图 5 分钟处理完 |
| 用 AI 做代码生成器 | Jinja2 | AI 写模板 + 变量替换 | 输入名称生成 CRUD 代码 |
| AI 帮我做 API 文档 | Swagger | AI 读代码生成文档 | 文档从 0 到完整 1 小时 |
| 用 AI 做性能优化 | SQL+索引 | AI 分析瓶颈 + 写优化 | 接口从 2s 到 200ms |
| AI 帮我做 PWA | Service Worker | AI 加 manifest + 离线 | 网站可安装、可离线 |
| 用 AI 做 SEO | JSON-LD | AI 生成结构化数据 | Google 收录率提升 |
| AI 帮我做表单验证 | HTML+JS | AI 写验证规则 + 实时反馈 | 表单错误率下降 60% |
| 用 AI 做聊天机器人 UI | React+API | AI 生成界面 + 流式输出 | 对话体验接近商用产品 |
| AI 帮我做定时任务 | cron | AI 写任务 + 日志 | 每天 8 点自动发报告 |
| 用 AI 做备份脚本 | Bash+S3 | AI 写压缩 + 上传 | 网站数据每天自动备份 |
| AI 帮我做数据库迁移 | SQL | AI 写迁移 + 回滚 | 数据库升级零停机 |
| 用 AI 做配置管理 | dotenv | AI 生成模板 + 校验 | 开发/生产配置分离 |
| AI 帮我做脚手架 | React | AI 生成项目结构 | 新项目 5 分钟起步 |
| 用 AI 做文本分析 | jieba | AI 写词频 + 情感分析 | 评论数据变洞察 |
| AI 帮我做监控 | Python | AI 写检查 + 通知 | 服务异常 3 分钟知道 |
| 用 AI 做报表 | reportlab | AI 写模板 + 图表 | 月度报表自动生成 |
| AI 帮我做爬虫反爬 | requests | AI 写代理 + 频率控制 | 稳定抓取不被封 |
| 用 AI 做数据可视化 | ECharts | AI 写图表 + 交互 | 数据变直观 |
40 cases cover 'frontend/backend/data/ops/crawler/performance/SEO'. Common point: not from scratch, but AI generation + human review + assembly. This is the norm of AI-era coding.扩充案例Extended Cases
C20
60 个『AI 编程』高频问题(FAQ)
60 'AI Coding' High-Frequency Questions (FAQ)
60 个 AI 编程高频问题,直接给答案,覆盖认知/入门/进阶/工具/心态。
60 high-frequency AI-coding questions, direct answers, covering mindset/beginner/advanced/tools/attitude.
取决于你怎么用。直接复制粘贴 = 质量差;AI 生成 + 人审查 + 测试 = 质量高。代码质量的上限是你的审查能力。Depends on how you use it. Copy-paste = poor; AI + human review + tests = high. Quality ceiling is your review ability.
能写简单的(小工具、脚本、页面),但复杂的需要基础。建议先学 2-4 周基础,再深度用 AI。You can write simple things (scripts, pages), but complex needs basics. Learn 2-4 weeks basics first, then deep AI use.
不会让你失业,但会让你『不会用 AI』的竞争力下降。会用 AI 的程序员 > 不会用的。It won't make you unemployed, but lowers competitiveness of those who don't use AI. AI-using programmers > non-users.
Python(AI/数据/脚本最友好)+ JavaScript(前端/Web 通用)。先选一个,别贪多。Python (AI/data/scripts most friendly) + JavaScript (frontend/web general). Pick one, don't be greedy.
盲目信任。AI 会自信地写错代码,你必须审查、测试、验证。不能跑 ≠ 对,能跑 ≠ 对。Blind trust. AI confidently writes wrong code; you must review, test, verify. Runs ≠ correct, can't run ≠ correct.
① 先让 AI 出方案 ② 分块生成 ③ 每块审查 ④ 集成测试 ⑤ 让 AI 写注释和文档。① Let AI propose plan ② generate in chunks ③ review each ④ integration test ⑤ let AI write comments and docs.
能,而且很好用。把报错 + 代码 + 期望行为告诉 AI,它通常能定位。但你要能读懂它给的解释。Yes, very useful. Tell AI the error + code + expected behavior, it usually locates. But you must read its explanation.
给足上下文:语言/框架/输入输出/边界/示例/风格。越具体,AI 输出越准。Give full context: language/framework/input/output/edge cases/examples/style. The more specific, the more accurate.
需要。AI 能帮你实现,但你得判断它写的算法对不对、效率行不行。不懂算法 = 无法审查。Yes. AI implements, but you must judge if the algorithm is correct/efficient. No algorithm knowledge = can't review.
短期内不会取代『会思考的程序员』,会取代『只会复制粘贴的』。核心是判断力 + 系统设计,不是敲代码。Short-term won't replace 'thinking programmers', replaces 'copy-paste only'. Core is judgment + system design, not typing.
能读懂代码、理解变量/函数/循环/条件就够用。深度使用需要数据结构 + 算法 + 系统设计。Reading code, understanding variables/functions/loops/conditions is enough. Deep use needs data structures + algorithms + system design.
有争议。AI 生成的代码版权归属不明确,商用前建议审查来源,避免引入他人代码。Controversial. AI code copyright unclear; review sources before commercial use to avoid importing others' code.
边做边学:定一个小项目 → 让 AI 解释每一步 → 自己改写 → 迭代。不要光看,要动手。Learn by doing: set a small project → let AI explain each step → rewrite yourself → iterate. Don't just read, do.
不能。AI 可以辅助 review(找明显问题),但架构、业务逻辑、安全必须人来。No. AI can assist review (find obvious issues), but architecture, business logic, security must be human.
不必买贵课。免费资源(官方文档 + YouTube + 实战)+ AI 当老师,足够了。No need for expensive courses. Free resources (docs + YouTube + practice) + AI as teacher is enough.
负责:需求、方案、审查、测试、调试、部署。AI 负责:生成、补全、解释、找错。你是导演,AI 是演员。You handle: requirements, plan, review, test, debug, deploy. AI handles: generate, complete, explain, find errors. You're the director, AI the actor.
因为没给足上下文,或没审查。把报错贴回给 AI,让它自己修,通常 2-3 轮能通。Because insufficient context, or no review. Paste errors back to AI, let it fix itself, usually 2-3 rounds to pass.
CRUD、数据脚本、小工具、网站、API、自动化。不适合:高性能、高安全、复杂系统(需要深度设计)。CRUD, data scripts, small tools, websites, APIs, automation. Not: high-performance, high-security, complex systems (need deep design).
把函数 + 输入输出 + 边界情况告诉 AI,让它生成测试用例。然后跑覆盖率,补漏。Tell AI the function + input/output + edge cases, let it generate test cases. Then run coverage, fill gaps.
能给建议、画草图、列方案,但最终决策要人。AI 不懂你的业务约束和长期目标。Can suggest, sketch, list options, but final decision is human. AI doesn't know your business constraints and long-term goals.
用 AI,1-4 周能做出第一个小产品(脚本/工具/网站)。不用 AI,3-6 个月。With AI, 1-4 weeks to make first small product (script/tool/website). Without AI, 3-6 months.
VS Code(编辑器)+ Copilot/Cursor(AI 助手)+ Git(版本控制)+ 浏览器(测试)。VS Code (editor) + Copilot/Cursor (AI assistant) + Git (version control) + browser (test).
会,如果你不会判断。必须学会读代码、理解依赖、知道每个库干嘛的,否则就是黑箱。Yes, if you can't judge. Must learn to read code, understand dependencies, know what each library does, otherwise it's a black box.
对明显瓶颈靠谱(慢查询、冗余循环)。对复杂优化要自己验证基准。Reliable for obvious bottlenecks (slow queries, redundant loops). For complex optimization, verify benchmarks yourself.
不信任 AI 输入的校验、密码存储、SQL 注入防护。这些必须人审查 + 用成熟框架。Don't trust AI's input validation, password storage, SQL injection protection. These must be human-reviewed + use mature frameworks.
传统 = 自己写每一行。AI = 你描述需求,AI 生成,你审查。核心从『写』变成『判断 + 组装』。Traditional = write every line yourself. AI = describe requirements, AI generates, you review. Core shifts from 'writing' to 'judging + assembling'.
先学 2-4 周基础(变量/函数/循环/条件/数据结构),再深度用 AI。完全零基础直接用,会卡。Learn 2-4 weeks basics first (variables/functions/loops/conditions/data structures), then deep AI. Zero-basics direct use will get stuck.
能,很好用。让它给代码写注释、README、API 文档、用户手册。但你要校对。Yes, very useful. Let it write comments, README, API docs, user manuals. But you must proofread.
不一定。免费额度够学习(Copilot 免费/Cursor 试用)。重度使用再考虑付费。Not necessarily. Free tiers enough for learning (Copilot free/Cursor trial). Consider paid for heavy use.
能帮你拆解任务、估时间、写 README、规划里程碑。但沟通、决策、优先级要人。Can help break tasks, estimate time, write README, plan milestones. But communication, decisions, priorities are human.
AI 是概率模型,不是知识库。它会自信地编造。所以必须验证:跑一下、查文档、交叉比对。AI is a probabilistic model, not a knowledge base. It confidently fabricates. So must verify: run, check docs, cross-check.
适合,但要有方法:让 AI 解释原理,不要只抄代码。抄 = 学不会,理解 = 真会。Suitable, but with method: let AI explain principles, don't just copy code. Copy = don't learn, understand = truly learn.
把旧代码 + 重构目标告诉 AI,让它给新方案 + 解释改动。然后逐步替换,测试验证。Tell AI old code + refactor goal, let it give new plan + explain changes. Then replace step by step, test verify.
对常见查询靠谱(SELECT/JOIN/聚合)。对复杂窗口函数、性能优化要自己验证。Reliable for common queries (SELECT/JOIN/aggregate). For complex window functions, performance, verify yourself.
小项目可以(审查后)。大项目不建议,因为可维护性、一致性、安全性需要人的全局把控。Small projects can (after review). Large projects not recommended, maintainability/consistency/security need human global control.
能,如果你用它学原理而不是抄代码。让它解释、让你改写、让你对比,能力会快速提升。Yes, if you use it to learn principles not copy code. Let it explain, let you rewrite, let you compare, ability grows fast.
① 能跑吗 ② 逻辑对吗 ③ 边界处理了吗 ④ 安全吗 ⑤ 可维护吗。5 项都过才用。① Does it run ② Is logic correct ③ Edge cases handled ④ Secure ⑤ Maintainable. Use only if all 5 pass.
能给候选、列优劣,但最终要你的业务判断。AI 不懂你的团队、数据量、预算。Can give candidates, list pros/cons, but final is your business judgment. AI doesn't know your team, data volume, budget.
1-2 周能基本上手(写小工具)。1-3 个月能熟练(做完整项目)。1-2 weeks to basic (small tools). 1-3 months to fluent (complete projects).
简单部署(GitHub Pages/Cloudflare/Vercel)不用。复杂部署(Docker/K8s/CI)需要基础。Simple deploy (GitHub Pages/Cloudflare/Vercel) no. Complex (Docker/K8s/CI) needs basics.
能生成测试脚本、分析瓶颈,但真实压测要自己跑。Can generate test scripts, analyze bottlenecks, but real load testing you run yourself.
让 AI 生成学习路径 + 小项目 → 边做边问 → 每步理解 → 整合成项目。Let AI generate learning path + small projects → do while ask → understand each step → integrate into projects.
要。让它给关键逻辑写注释,但不要每行都注释。注释解释『为什么』,不是『是什么』。Yes. Let it comment key logic, not every line. Comments explain 'why', not 'what'.
可以,但必须:完整测试 + 代码审查 + 监控。AI 代码进生产 = 人负责兜底。Yes, but must: complete tests + code review + monitoring. AI code in production = human owns fallback.
能给表结构、索引、范式建议。但业务约束、扩展性要人决策。Can give table structure, indexes, normalization suggestions. But business constraints, scalability are human decisions.
① 审查 ② 测试 ③ 理解 ④ 不盲信 ⑤ 版本控制 ⑥ 安全审查。6 条铁律。① Review ② Test ③ Understand ④ Don't blindly trust ⑤ Version control ⑥ Security review. 6 iron rules.
能,从旧框架/语言迁移到新框架/语言,AI 能加速。但兼容性、边界要人验证。Yes, migrate old framework/language to new, AI accelerates. But compatibility, edges need human verification.
给 AI 你的代码示例 + 风格说明(命名/结构/注释),让它模仿。Give AI your code examples + style description (naming/structure/comments), let it imitate.
能给 RESTful 建议、字段设计、错误码。但业务语义、版本策略要人。Can give RESTful suggestions, field design, error codes. But business semantics, version strategy are human.
要。AI 能帮你实现模式,但你得知道什么时候用哪个模式。不懂 = 无法审查架构。Yes. AI implements patterns, but you must know when to use which. Don't know = can't review architecture.
能,按项目类型生成 checklist(安全/性能/可读性/测试)。Yes, generate checklists by project type (security/performance/readability/tests).
让 AI 列候选 + 优劣 + 案例 + 风险,然后你深入验证关键项。Let AI list candidates + pros/cons + cases + risks, then you deep-verify key items.
小项目不用。中大型项目建议学基础(容器/无服务器/CI/CD)。Small projects no. Medium-large suggest learning basics (containers/serverless/CI/CD).
可以,但注意:遵守 robots.txt、控制频率、不碰反爬强的站、数据合规。Yes, but note: follow robots.txt, control frequency, avoid strong anti-crawl sites, data compliance.
能,按数据类型给推荐(折线/柱状/散点/地图)。Yes, recommend by data type (line/bar/scatter/map).
把常用代码片段整理成模板,让 AI 补全变量、生成文档,形成你的『代码资产』。Organize common snippets into templates, let AI fill variables, generate docs, form your 'code assets'.
能给优化建议(缓存/索引/并发/算法)。但必须 A/B 验证,不能只看理论。Can give optimization suggestions (cache/index/concurrency/algorithm). But must A/B verify, not just theory.
能,成长路径变了:从『会写』变成『会判断 + 会设计 + 会审查』。这是更高阶的能力。Yes, growth path changed: from 'can write' to 'can judge + design + review'. This is higher-level ability.
C21
20 个『AI 编程』高频术语(扫盲表)
20 'AI Coding' High-Frequency Terms (Glossary)
20 个 AI 编程高频术语,中英对照 + 一句话解释,扫盲用。
20 high-frequency AI-coding terms, bilingual + one-line explanation, for onboarding.
| 术语 | 英文 | 一句话解释 | One-line explanation |
|---|---|---|---|
| 代码补全 | Code Completion | AI 根据上下文自动续写代码 | AI auto-continues code based on context |
| 上下文窗口 | Context Window | AI 一次能『看到』的代码量上限 | Max code AI can 'see' at once |
| 提示词 | Prompt | 你给 AI 的指令和上下文 | Your instruction and context given to AI |
| 幻觉 | Hallucination | AI 自信地编造不存在的内容 | AI confidently fabricates nonexistent content |
| 审查 | Code Review | 人对 AI 代码的检查 | Human inspection of AI code |
| 单测 | Unit Test | 验证单个函数行为的测试 | Test verifying single function behavior |
| 覆盖率 | Coverage | 测试覆盖代码的比例 | Ratio of code covered by tests |
| 重构 | Refactor | 改结构不改行为的优化 | Optimize structure without changing behavior |
| 依赖 | Dependency | 代码用到的外部库 | External libraries your code uses |
| 版本控制 | Version Control | Git,记录代码每次变化 | Git, records each code change |
| 合并冲突 | Merge Conflict | 多人改同一处的代码冲突 | Code conflict when multiple edit same spot |
| 持续集成 | Continuous Integration | 每次提交自动跑测试 | Auto-run tests on each commit |
| 部署 | Deployment | 把代码上线到服务器 | Put code live on server |
| 环境变量 | Environment Variable | 区分开发/生产的配置 | Config distinguishing dev/prod |
| 缓存 | Cache | 用空间换时间的加速手段 | Trade space for time acceleration |
| 索引 | Index | 数据库加速查询的结构 | Structure accelerating DB queries |
| 并发 | Concurrency | 多个任务同时处理 | Multiple tasks processed simultaneously |
| 异步 | Async | 不阻塞地等待结果 | Wait for results without blocking |
| 流式输出 | Streaming Output | AI 边生成边返回 | AI returns as it generates |
| 多模态 | Multimodal | AI 处理文本/图像/音频 | AI processes text/image/audio |
C21
自测:你的 AI 学外语能力
Quiz: Your AI Language-Learning Ability
10 题,点选项即判对错 + 解析
10 questions, click to judge + explain
Q1
AI 时代编程最核心的变化是?The most core change of coding in the AI era?
人从『写代码』转向『定义 + 判断』Human shifts from 'writing code' to 'defining + judging'完全不用编程了No coding at all只能学新语言Can only learn new languages
✅ 正确:人从『写代码』转向『定义 + 判断』
✅ Correct: Human shifts from 'writing code' to 'defining + judging'
Q2
AI 写的代码为什么不能直接信?Why can't AI code be trusted directly?
AI 会自信地犯错,需要人审查 + 测试AI confidently errs; needs human review + testsAI 太慢AI too slowAI 不会中文AI can't Chinese
✅ 正确:AI 会自信地犯错,需要人审查 + 测试
✅ Correct: AI confidently errs; needs human review + tests
Q3
审查 AI 代码的 3 要点是?The 3 points to review AI code?
逻辑对不对 + 边界处理没 + 性能够不够Logic right + boundaries handled + performance enough只看有没有报错Only see if it errors只看格式Only see format
✅ 正确:逻辑对不对 + 边界处理没 + 性能够不够
✅ Correct: Logic right + boundaries handled + performance enough
Q4
『最小可行转型』在编程里叫什么?What is 'minimum viable switch' called in coding?
MVP(最小可行产品)MVP (Minimum Viable Product)PPTPPT路演Roadshow
✅ 正确:MVP(最小可行产品)
✅ Correct: MVP (Minimum Viable Product)
Q5
AI 编程提示词的核心是?The core of AI coding prompts?
具体:功能 + 约束 + 技术栈 + 示例Specific: function + constraints + tech stack + examples越长越好Longer is better越短越好Shorter is better
✅ 正确:具体:功能 + 约束 + 技术栈 + 示例
✅ Correct: Specific: function + constraints + tech stack + examples
Q6
边做边学 vs 先学完再动手,哪个对?Learn-while-doing vs learn-all-then-do, which is right?
边做边学(项目驱动)Learn while doing (project-driven)先学完再动手Learn all then act只看不做Only watch, don't do
✅ 正确:边做边学(项目驱动)
✅ Correct: Learn while doing (project-driven)
Q7
Git 最重要的价值是?The most important value of Git?
可追溯 + 可回滚 + 可协作Traceable + rollbackable + collaboratable让代码变好看Make code look nice加快运行Speed up running
✅ 正确:可追溯 + 可回滚 + 可协作
✅ Correct: Traceable + rollbackable + collaboratable
Q8
能跑不等于对,还缺什么?Running doesn't mean right; what's missing?
测试 + 边界验证Tests + boundary verification更多注释More comments更快速度Faster speed
✅ 正确:测试 + 边界验证
✅ Correct: Tests + boundary verification
Q9
AI 时代最稀缺的编程能力是?The most scarce coding ability in the AI era?
产品思维(做对的产品)Product thinking (making the right product)打字速度Typing speed背语法Memorizing syntax
✅ 正确:产品思维(做对的产品)
✅ Correct: Product thinking (making the right product)
Q10
遇到不会的编程问题,第一反应?First reaction to an unknown coding problem?
让 AI 给方案 + 我审查 + 我验证Let AI give a plan + I review + I verify全信 AIFully trust AI全靠自己猜Guess alone
✅ 正确:让 AI 给方案 + 我审查 + 我验证
✅ Correct: Let AI give a plan + I review + I verify
C22
AI 学外语 20 项行动清单
AI Language-Learning 20-Item Action List
20 个可立即执行的动作,打勾跟踪
20 executable actions, track with checks
- 我能用 Python 写 10 个小脚本I can write 10 small scripts in Python
- 我理解变量/函数/循环/异常I understand variables/functions/loops/exceptions
- 我能审查 AI 的代码(3 要点)I can review AI's code (3 points)
- 我给 AI 代码加过测试I've added tests to AI code
- 我用过 Git 做版本控制I've used Git for version control
- 我能写具体的编码提示词I can write specific coding prompts
- 我做过 3 个可运行的小项目I've done 3 runnable small projects
- 我用过 AI 协作完成 1 个项目I've completed 1 project with AI collaboration
- 我理解计算思维(拆解/抽象)I understand computational thinking (decompose/abstract)
- 我知道常用数据结构(数组/哈希/树)I know common data structures (array/hash/tree)
- 我理解 API 调用(HTTP/JSON)I understand API calls (HTTP/JSON)
- 我调过 LLM API(Claude/ChatGPT)I've called LLM APIs (Claude/ChatGPT)
- 我做过 1 个 RAG 应用I've made 1 RAG app
- 我理解部署(Docker/云)I understand deployment (Docker/cloud)
- 我理解测试(正常/边界/异常)I understand testing (normal/boundary/exception)
- 我理解安全基础(认证/加密)I understand security basics (auth/encryption)
- 我做过 1 个可上线的 MVPI've made 1 online MVP
- 我写过 README + 技术文档I've written README + tech docs
- 我养成『审查 + 测试』的习惯I've formed the 'review + test' habit
- 我理解产品思维(先想用户要什么)I understand product thinking (think what users need first)
C23
15 条 AI 学外语 FAQ
15 AI Language-Learning FAQ
15 条最高频的问题,直接给答案
15 most high-frequency questions, direct answers
不难。AI 时代门槛大幅降低,用 AI 解释 + 项目驱动,3 个月入门。Not hard. The barrier has dropped greatly; with AI explanation + project-driven, start in 3 months.
基础编程不用高数。AI 应用/自动化只需逻辑。机器学习才需线性代数。Basic coding doesn't need advanced math. AI apps/automation only need logic. ML needs linear algebra.
Python 起步最友好 + AI 生态最强。2026 推荐 Python。Python is the friendliest start + strongest AI ecosystem. Python recommended for 2026.
要审查 + 测试。AI 会错,你是最后防线。Review + test. AI errs; you're the last line.
学常见算法(排序/搜索/递归)即可,不要求竞赛级。Learn common algorithms (sort/search/recursion) is enough; no competition level needed.
简单工具够用;复杂产品要写代码。AI 时代两者结合。Simple tools suffice; complex products need code. In the AI era, combine both.
不会消失,但会『上移』。重复编码被替代,设计/架构/产品价值上升。Won't disappear, but will 'move up'. Repetitive coding is replaced; design/architecture/product value rises.
能。用自然语言 + AI,非程序员也能做出可用的工具/应用。Yes. With natural language + AI, non-programmers can make usable tools/apps.
30 天 MVP;3 个月实用工具;6-12 个月产品。30 days for MVP; 3 months for a practical tool; 6-12 months for a product.
Claude/Cursor/Copilot 都好。核心是『能协作』,不是追最强。Claude/Cursor/Copilot are all good. The core is 'can collaborate', not chasing the strongest.
3 要点: 逻辑 + 边界 + 性能。AI 辅助 + 自己判断。3 points: logic + boundary + performance. AI assist + judge yourself.
先想『怎么证明对』,写正常/边界/异常。用 pytest。First think 'how to prove right', write normal/boundary/exception. Use pytest.
极重要。从第一行就用。可追溯是底线。Extremely important. Use from the first line. Traceability is the bottom line.
做产品要学。SQL + 一种 NoSQL + 一种缓存。Learn for products. SQL + one NoSQL + one cache.
有意义。编程是 AI 时代的杠杆 + 通识,不是被淘汰的手艺。Meaningful. Coding is the lever + general literacy of the AI era, not an eliminated craft.
C24
参考来源
References
综合 CEFR 标准、语言学习研究 + AI 工具官方资料(2026)
Synthesizes CEFR, language-learning research + official AI tool materials (2026)
- 《计算思维》(Jeannette Wing)'Computational Thinking' (Jeannette Wing)
- Python 官方教程Python official tutorial
- 《代码整洁之道》(Robert C. Martin)'Clean Code' (Robert C. Martin)
- 《重构》(Martin Fowler)'Refactoring' (Martin Fowler)
- Git 官方文档Git official docs
- 《设计模式》(Gang of Four)'Design Patterns' (Gang of Four)
- 《测试驱动开发》(Kent Beck)'Test-Driven Development' (Kent Beck)
- 《凤凰架构》(周志明)'Phoenix Architecture' (Zhou Zhiming)
- 《SRE: 站点可靠性工程》'SRE: Site Reliability Engineering'
- 《数据密集型应用》(Martin Kleppmann)'Data Intensive Applications' (Martin Kleppmann)
- 《系统架构设计》(Michael Nygard)'Software Architecture'
- Anthropic 编码指南Anthropic coding guide
- GitHub Copilot 文档GitHub Copilot docs
- Cursor 文档Cursor docs
- 《AI 编程实践 2026》'AI Coding Practice 2026'