模拟面试官角色，帮助你准备 C++ 面试。
- 不是直接给答案，而是引导你思考、追问细节、指出常见误区
- 无需 PDF 资料，无需安装 Git LFS，下载后直接使用
- 两种模式，学练结合
| 入口 | 作用 | 示例 |
|---|---|---|
/interview |
知识讲解 | /interview 虚函数是怎么实现的 |
/coach |
面试训练 | /coach 虚函数、/coach weak |
/interview = 教学讲解（学）
/coach = 面试训练（练）
/interview 虚函数是怎么实现的
/interview 智能指针有哪几种
/interview 什么是内存泄漏
/coach 虚函数 # 专项训练：虚函数
/coach 智能指针 # 专项训练：智能指针
/coach weak # 训练薄弱知识点
/coach status # 查看掌握度仪表盘
/coach plan # 生成今日训练计划
/coach start # 进入训练循环
/coach 会像面试官一样：出题 → 等你回答 → 评价 → 追问 → 下一题。
专项训练 /coach 虚函数
你：/coach 虚函数
教练：专项训练：vtable
掌握度：2.4%
难度：基础
第 1 题：
什么是虚函数表（vtable）？每个包含虚函数的类和它的对象分别与 vtable 是什么关系？
你：虚函数表是一个函数指针数组，编译器为每个含虚函数的类生成一张表。类共享一张 vtable，对象通过 vptr 指向它。
教练：评价：概念基本正确，但遗漏了 vptr 的存储位置和继承场景下的表合并规则。
薄弱点：vptr 布局、多重继承
掌握度更新：2.4% → 12%
第 2 题：
多重继承下，一个对象会有几个 vptr？编译器如何决定调用哪个虚函数？
薄弱训练 /coach weak
你：/coach weak
教练：当前薄弱点：vtable（掌握度 12%）
开始训练...
第 1 题：
虚析构函数的作用是什么？如果基类析构函数不是虚函数，delete 派生类指针会发生什么？
掌握度仪表盘 /coach status
你：/coach status
教练：📊 掌握度仪表盘
──────────────────────────
总知识点：15 已掌握：2
薄弱：8 平均掌握度：23%
──────────────────────────
▓▓▓░░░░░░░ 虚函数 35%
▓▓░░░░░░░░ 智能指针 20%
▓░░░░░░░░░ vtable 12%
░░░░░░░░░░ 内存管理 5%
...
今日计划 /coach plan
你：/coach plan
教练：📋 今日训练计划
──────────────────────────
复习（到期）：
1. vtable — 上次训练 3 天前，建议复习
新增（薄弱）：
2. 内存管理 — 掌握度 5%
3. 多线程锁 — 掌握度 8%
──────────────────────────
输入 /coach 开始按计划训练
进入训练循环 /coach start
你：/coach start
教练：你想训练哪个方向？
- 直接说知识点名称（如"虚函数"、"智能指针"）
- weak：训练薄弱知识点
- due：复习到期内容
你：虚函数
教练：专项训练：虚函数
掌握度：35%
难度：中等
第 1 题：
虚函数的动态绑定在编译期和运行期分别做了什么？如果不用虚函数，还有什么方式实现多态？
- 面试官视角：引导思考，不直接给答案
- 弱项驱动训练：自动优先训练最薄弱知识点
- 六维度精准诊断：correctness、completeness、depth、clarity、code_accuracy、edge_case_awareness
- 追问式训练：根据你的回答持续追问，挖掘薄弱点
- SQLite 状态持久化：掌握度、QA 历史自动存储
git clone https://github.com/yiqi-7/Cpp-Interviewer.git
cd Cpp-Interviewer
python setup.py安装后重启 Claude Code，输入 / 即可看到 interview 和 coach。
git clone https://github.com/yiqi-7/Cpp-Interviewer.git将 skills/interview/ 和 skills/coach/ 两个文件夹复制到 ~/.claude/skills/ 下：
~/.claude/skills/
├── interview/
│ ├── SKILL.md
│ └── index/knowledge_index.json
└── coach/
├── SKILL.md
├── coach/ # Python 持久化后端
│ ├── cli.py
│ ├── db.py
│ ├── scheduler.py
│ └── ...
└── index/knowledge_index.json
- 无需 PDF 资料即可使用，默认基于内置知识点索引和模型知识生成回答
/coach reset和/coach export尚未实现- 真实 LLM 接入需要设置
OPENAI_API_KEY环境变量
# 从 skills/coach/ 目录运行
cd skills/coach
python -m coach.cli status
python -m coach.cli topic 虚函数
# 或者使用安装脚本
python setup.py本项目采用 MIT License 开源。
- 交流群：Q 828570482
- 已获 LINUX DO 社区认可
Simulates an interviewer to help you prepare for C++ interviews.
- No direct answers — guides your thinking, asks follow-up questions, highlights pitfalls
- No PDF required — works out of the box with built-in knowledge index
- Two modes — learn + practice
| Entry | Purpose | Example |
|---|---|---|
/interview |
Teaching | /interview How are virtual functions implemented |
/coach |
Training | /coach virtual_function, /coach weak |
/interview = teaching (learn)
/coach = interview training (practice)
/interview How are virtual functions implemented
/interview What types of smart pointers are there
/interview What is a memory leak
/coach virtual_function # Train specific topic
/coach smart_pointer # Train specific topic
/coach weak # Train weak topics
/coach status # View mastery dashboard
/coach plan # Generate today's training plan
/coach start # Enter training loop
/coach acts like an interviewer: ask → wait for your answer → evaluate → follow-up → next question.
Specific Topic /coach virtual_function
You: /coach virtual_function
Coach: Training: vtable
Mastery: 2.4%
Difficulty: Basic
Q1:
What is a vtable? What's the relationship between a class with virtual functions, its objects, and the vtable?
You: A vtable is an array of function pointers. The compiler generates one per class with virtual functions. Objects point to it via vptr.
Coach: Evaluation: Core concept correct, but missed vptr storage layout and inheritance table merging rules.
Weakness: vptr layout, multiple inheritance
Mastery update: 2.4% → 12%
Q2:
Under multiple inheritance, how many vptrs does an object have? How does the compiler resolve virtual function calls?
Weak Topics /coach weak
You: /coach weak
Coach: Current weakness: vtable (mastery 12%)
Starting training...
Q1:
What's the purpose of a virtual destructor? What happens if you delete a derived class pointer through a non-virtual base destructor?
Mastery Dashboard /coach status
You: /coach status
Coach: 📊 Mastery Dashboard
──────────────────────────
Total: 15 Mastered: 2
Weak: 8 Avg Mastery: 23%
──────────────────────────
▓▓▓░░░░░░░ Virtual Functions 35%
▓▓░░░░░░░░ Smart Pointers 20%
▓░░░░░░░░░ vtable 12%
░░░░░░░░░░ Memory Mgmt 5%
...
Training Plan /coach plan
You: /coach plan
Coach: 📋 Today's Training Plan
──────────────────────────
Review (due):
1. vtable — last trained 3 days ago, recommended review
New (weak):
2. Memory Management — mastery 5%
3. Thread Locks — mastery 8%
──────────────────────────
Type /coach to start training
Training Loop /coach start
You: /coach start
Coach: What topic would you like to train?
- Type a topic name (e.g. "virtual_function", "smart_pointer")
- weak: train weak topics
- due: review due content
You: virtual_function
Coach: Training: Virtual Functions
Mastery: 35%
Difficulty: Intermediate
Q1:
What happens at compile time vs runtime for dynamic binding of virtual functions? Besides virtual functions, what other ways can achieve polymorphism?
- Interviewer perspective: Guides thinking, no direct answers
- Weakness-driven training: Automatically prioritizes weakest topics
- Six-dimension precise diagnosis: correctness, completeness, depth, clarity, code_accuracy, edge_case_awareness
- Follow-up based training: Continues questioning based on your answers to dig out weak points
- SQLite state persistence: Mastery, QA history automatically stored
git clone https://github.com/yiqi-7/Cpp-Interviewer.git
cd Cpp-Interviewer
python setup.pyAfter installation, restart Claude Code. Type / to see interview and coach.
git clone https://github.com/yiqi-7/Cpp-Interviewer.gitCopy skills/interview/ and skills/coach/ to ~/.claude/skills/:
~/.claude/skills/
├── interview/
│ ├── SKILL.md
│ └── index/knowledge_index.json
└── coach/
├── SKILL.md
├── coach/ # Python 持久化后端
│ ├── cli.py
│ ├── db.py
│ ├── scheduler.py
│ └── ...
└── index/knowledge_index.json
- Works without PDF resources — default responses based on built-in knowledge index and model knowledge
/coach resetand/coach exportare not yet implemented- Real LLM integration requires setting
OPENAI_API_KEYenvironment variable
pip install -e .
python -m pytest tests/ -v
python -m coach.cli status
python -m coach.cli topic virtual_functionThis project is licensed under the MIT License.
- Discussion group: Q 828570482
- Recognized by LINUX DO community