Nov17

Getting Started with This Workspace

Initial Setup

1. Initialize Git Repository

git init
git add .
git commit -m "Initial commit: Claude Code experimentation workspace"

2. Create GitHub Repository

1. Go to https://github.com/new
2. Name your repository (e.g., claude-code-experiments)
3. Keep it public to share with others
4. Don’t initialize with README, .gitignore, or license (already included)
5. Click “Create repository”

3. Push to GitHub

# Replace YOUR_USERNAME and REPO_NAME with your values
git remote add origin https://github.com/YOUR_USERNAME/REPO_NAME.git
git branch -M main
git push -u origin main

Using This Workspace with Claude Code

Quick Start

1. Open this workspace in VS Code with Claude Code extension
2. Use sandbox/ for quick experiments
3. Use templates from templates/ as starting points
4. Move successful experiments to experiments/ with dates

Example Workflow

Testing an idea:

# 1. Create a test file in sandbox
cd sandbox
touch test_idea.py

# 2. Ask Claude Code to help implement
# "Help me implement a function to parse JSON data"

# 3. Test it
python test_idea.py

# 4. If successful, move to experiments with date
mv test_idea.py ../experiments/2025-11-17_json_parser.py

Using a template:

# Copy a template to sandbox
cp templates/data_analysis.py sandbox/my_analysis.py

# Ask Claude Code to customize it
# "Modify my_analysis.py to analyze CSV data about sales"

Utility Usage Examples

Timer for performance testing:

from utils.timer import timer, Timer

@timer
def my_function():
    # Your code here
    pass

# Or use context manager
with Timer("My operation"):
    # Your code here
    pass

Logger for debugging:

from utils.logger import setup_logger

log = setup_logger("my_experiment")
log.info("Starting analysis")
log.debug("Debug details here")

Best Practices with Claude Code

1. Clear Communication

• Be specific about what you want to test
• Reference files: “modify sandbox/test.py to…”
• Ask for explanations: “explain how this algorithm works”

2. Iterative Development

• Start simple in sandbox/
• Test frequently with python your_file.py
• Use utilities to benchmark and log
• Refine before moving to experiments/

3. File Organization

sandbox/           → Quick tests, temporary work
experiments/       → Documented, working code
templates/         → Reusable starting points
utils/            → Helper functions (rarely modified)

4. Naming Conventions

• Experiments: YYYY-MM-DD_description.py
  • 2025-11-17_api_performance_test.py
  • 2025-11-17_sorting_algorithm_comparison.py
• Sandbox: Anything goes, but descriptive names help
  • test_api.py
  • quick_sort_test.py

Example Claude Code Prompts

Starting an Experiment

"Create a script in sandbox/ that tests different sorting algorithms 
using the timer utility to compare performance"

Using Templates

"Copy the algorithm_test.py template and modify it to compare 
three different approaches to finding prime numbers"

Debugging

"Help me debug sandbox/my_test.py - it's throwing an IndexError"

Code Review

"Review experiments/2025-11-17_data_parser.py and suggest improvements 
for performance and readability"

Documentation

"Add docstrings and comments to sandbox/working_code.py, then move it 
to experiments/ with today's date"

Git Workflow

Daily Commits

# After successful experiments
git add experiments/2025-11-17_*.py
git commit -m "Add experiment: [description]"
git push

Clean Sandbox (Optional)

# Sandbox is git-ignored, so you can clean anytime
rm -rf sandbox/*

# Or keep a backup first
tar -czf sandbox_backup_$(date +%Y%m%d).tar.gz sandbox/
rm -rf sandbox/*

Update Documentation

# After adding new utilities or templates
git add utils/ templates/ README.md
git commit -m "Update utilities and documentation"
git push

Sharing Your Work

Make a Great README

Update README.md with:

• What you’re experimenting with
• Interesting findings from experiments
• Custom utilities you’ve added
• Links to specific experiments

Add Topics to GitHub

On your GitHub repo page, add topics:

• python
• experimentation
• claude-code
• prototyping
• Add domain-specific tags (e.g., data-science, algorithms)

Share Experiments

Point people to specific experiments:

Check out my performance comparison:
github.com/YOUR_USERNAME/REPO_NAME/blob/main/experiments/2025-11-17_string_concat_perf.py

Maintenance

Weekly Cleanup

# Review sandbox contents
ls -la sandbox/

# Remove temporary files
rm sandbox/temp_*.py

# Commit any new experiments
git add experiments/
git commit -m "Add week's experiments"
git push

Monthly Review

• Review experiments and add summary comments
• Update README with findings
• Clean up old sandbox backups
• Check for utility improvements

Tips for Success

1. Commit often - Each successful experiment is worth saving
2. Document findings - Add comments explaining what you learned
3. Use branches - For larger experiments: git checkout -b feature/new-experiment
4. Stay organized - Move completed work out of sandbox regularly
5. Iterate quickly - Don’t overthink, test ideas fast

Getting Help

With This Workspace

• Check README.md for structure
• Look at example experiment: experiments/2025-11-17_string_concat_perf.py
• Use templates as starting points

With Claude Code

• Ask specific questions
• Reference files explicitly
• Request explanations for generated code
• Iterate on solutions

---

Ready to start? Try running the example experiment:

python experiments/2025-11-17_string_concat_perf.py

This site is open source. Improve this page.