/interview-me

/interview-me

Generate technical interview questions by analyzing your entire project codebase.

Usage

/interview-me [options]

Options

• --count=<n>: Number of questions to generate (default: 3, max: 7)
• --lang=<ko|en>: Question language (default: ko)

Instructions

You are the interviewer agent defined in agents/interviewer.md. Follow these steps:

Step 1: Collect Target Code

Explore the entire project codebase to understand its structure and implementation:

1. First, identify the project type and structure (look for package.json, pom.xml, build.gradle, requirements.txt, etc.)
2. Find and read the main source files (src/, lib/, app/, etc.)
3. Focus on core business logic, not configuration or boilerplate
4. Prioritize files with substantial logic (services, controllers, models, utilities)

If the project is empty or has no meaningful code, inform the user.

Step 2: Analyze the Code

Read the code carefully and identify:

• Security concerns (auth, input validation, secrets, injection)
• Concurrency issues (race conditions, transactions, locks)
• Performance problems (N+1, missing indexes, memory leaks)
• Design smells (tight coupling, god classes, missing abstractions)
• Testing gaps (untested paths, hard-to-mock dependencies)
• Observability needs (logging, metrics, error handling)
• Data issues (schema problems, consistency, migrations)

Step 3: Generate Questions

Create questions following these rules:

Quality criteria:

• Reference specific code locations (file:line)
• Ask about concrete implementation choices, not abstract theory
• Include failure scenarios and edge cases
• Probe trade-offs and alternatives

Avoid:

• "What does this code do?" explanations
• Textbook definitions
• Questions about code that doesn't exist

Format each question with:

• Category emoji and name
• Difficulty stars (★☆☆ / ★★☆ / ★★★)
• Clear question text (max 54 chars per line)
• Code reference (📍 file:line)
• Hint keywords (💡 Hint: ...)

Step 4: Output Design

Use this exact visual format:

╭──────────────────────────────────────────────────────────────╮
│  🎤  Interview Me  ·  {count} questions                      │
│  Target: {target_description}                                │
╰──────────────────────────────────────────────────────────────╯

┌─ Q1 ─────────────────────────────── {emoji} {category} · {stars} ─┐
│                                                                    │
│  {question_line_1}                                                 │
│  {question_line_2}                                                 │
│  {question_line_3}                                                 │
│                                                                    │
│  📍 {file}:{line}                                                  │
│  💡 Hint: {keyword1}, {keyword2}, {keyword3}                       │
│                                                                    │
└────────────────────────────────────────────────────────────────────┘

... more questions ...

─────────────────────────────────────────────────────────────────────
 💬  답변 준비되면 그냥 이어서 말하면 돼.
─────────────────────────────────────────────────────────────────────

Design rules:

• Header uses round corners (╭╮╰╯)
• Question boxes use square corners (┌┐└┘)
• Box width: 60 characters fixed
• Question text: max 54 chars per line, auto-wrap
• Category emojis: 🔒 Security / ⚡ Concurrency / 🚀 Performance / 🏛 Design / 🧪 Testing / 👁 Observability / 🗄 Data
• Difficulty: ★☆☆ / ★★☆ / ★★★ (exactly this format)

Language

• Default output language: Korean (ko)
• If --lang=en, output questions in English
• Technical terms can remain in English regardless of language setting