claudecode-best-practices

```markdown
# Claude Code Best Practices Guide for Python Data Pipeline Project

## 1. Environment Setup
- Install Python 3.11+ and pip in a virtual environment:
```bash
python -m venv .venv && source .venv/bin/activate
pip install --upgrade pip
```
- Add required tools to `.env`:
```ini
CLAUDE_API_KEY=your_api_key_here
PYTHONPATH=/path/to/project
```

## 2. Prompt Optimization
For API endpoint generation, use structured prompts:
```
"Generate a FastAPI endpoint for user profile updates that:
1. Validates JWT tokens from [AUTH_SERVICE]
2. Accepts PATCH requests with {user_id, email, bio}
3. Returns 400 for invalid email formats
4. Logs changes to [LOG_SERVICE]"
```

## 3. Debugging Common Errors
**Circular Import Error:**
```python
# In .clauderc
{
  "code_execution": {
    "allowed_modules": ["fastapi", "pydantic", "sqlalchemy"],
    "disallowed_modules": ["os", "sys"]
  }
}
```

## 4. Performance Tuning
For large datasets (>100k records):
- Use batch processing with chunk_size=1000
- Enable parallel execution:
```python
# In your prompt
"Process the data in parallel using multiprocessing with 4 workers"
```

## 5. Security Considerations
- Never include API keys in prompts
- Use placeholder variables:
```
"Connect to database using credentials from [DB_CREDENTIALS]"
```
- Validate all generated code with:
```bash
python -m pyflakes generated_code.py
```
```