An agent skill that generates complete data science interview questions — exactly as they appear on the StrataScratch platform. Title, question, and dataset preview, ready to solve.
Compatible with Claude Code, OpenAI Codex, Gemini CLI, Cursor, and other agents that support the open Agent Skills standard.
The Interview Grinder creates original, interview-ready questions with:
- Title — Short, descriptive name (e.g., "Top Salaries by Department")
- Question — Business-framed problem with clear output requirements
- Dataset Preview — Table schemas with sample rows, exactly as shown on the platform
Questions are calibrated against a bank of 100 real StrataScratch questions for style, difficulty, and structure — but every output is 100% original.
# npx (any agent)
npx skills add stratascratch/interview-grinderOr clone manually:
git clone https://github.com/stratascratch/interview-grinder.git
cp -r skills/strata-interview-grinder ~/.claude/skills/Just ask naturally:
Generate an interview question
Give me a hard SQL question about customer churn
Grind a question about window functions
The skill activates automatically when it detects interview question creation intent.
After each question, use these shortcuts:
| Key | Action |
|---|---|
| Q 🔄 | Generate a brand new question |
| R ✏️ | Revise a specific section |
| D 📊 | Regenerate dataset with different data |
This skill operates in interview mode — it presents the question and data, you solve it. Solutions, hints, and edge case explanations are never shown. Edge cases are silently embedded in the dataset for you to discover.
Each skill follows the open Agent Skills standard:
skills/strata-interview-grinder/
├── SKILL.md # Instructions and metadata
├── README.md # Documentation
└── assets/ # Reference data (question bank CSV)
Released under the Apache 2.0 License.