Advanced Tips
Master SumoData Toolbox with these advanced techniques and workflows.
Power User Workflows
Chaining Tools
Combine multiple tools for complex tasks:
python
# Step 1: Generate boilerplate with ML Boilerplate
# Step 2: Add docstrings with Generate Docstring
# Step 3: Add type hints with Type Hinting
# Result: Production-ready, well-documented codeBatch Processing
Process multiple files efficiently:
- Open first file, select code, run tool
- While AI processes, open next file
- Review first result, apply changes
- Repeat for remaining files
Template Creation
Save frequently used results as templates:
- Generate code with SumoData
- Save result as snippet or template
- Customize for specific use cases
- Reuse across projects
Model Strategy
Cost Optimization
Minimize API costs with smart model selection:
Free/Cheap Tasks:
- Use ⚡ Lightning for SQL optimization
- Use 💨 Flash for cron expressions
- Use ✨ Spark for docstrings
Worth the Cost:
- Use 🔥 Titan Pro for ML boilerplate
- Use ⚡ Thunder for complex explanations
- Use 🌟 Nova for cutting-edge features
Quality vs Speed
Balance based on your needs:
| Scenario | Recommended Model | Why |
|---|---|---|
| Prototyping | Lightning/Flash | Speed matters |
| Production code | Titan Pro/Thunder | Quality matters |
| Learning | Spark/Nova | Balance both |
| Code review | Lightning | Quick feedback |
Integration with VS Code
Workspace Settings
Create project-specific settings:
json
// .vscode/settings.json
{
"sumodata.model": "claude-haiku-4-5",
"sumodata.maxCodeLength": 5000,
"sumodata.timeout": 60
}Multi-Root Workspaces
Different settings per project:
workspace/
├── project-a/
│ └── .vscode/settings.json (uses Lightning)
├── project-b/
│ └── .vscode/settings.json (uses Titan Pro)Tasks Integration
Automate with VS Code tasks:
json
// .vscode/tasks.json
{
"version": "2.0.0",
"tasks": [
{
"label": "Optimize All SQL",
"type": "shell",
"command": "echo 'Run SQL Optimizer on all .sql files'"
}
]
}Advanced Techniques
Context Enhancement
Improve results by adding context:
Before:
python
def process(data):
return data * 2Better (with context):
python
# This function processes sales data by doubling revenue
# Input: list of revenue numbers
# Output: doubled revenue for forecasting
def process(data):
return data * 2Incremental Refinement
Refine results iteratively:
- Generate initial code
- Review and identify issues
- Add comments with requirements
- Run tool again on updated code
- Repeat until satisfied
Combining with Other Extensions
Integrate with your existing tools:
- Prettier/Black: Format after inserting code
- ESLint/Pylint: Validate generated code
- GitLens: Track AI-generated changes
- Code Spell Checker: Verify documentation
Team Collaboration
Shared Configuration
Create team standards:
json
// .vscode/settings.json (committed to repo)
{
"sumodata.model": "claude-haiku-4-5",
"sumodata.enableContextMenu": true,
"sumodata.maxCodeLength": 8000
}Code Review Guidelines
Establish team practices:
- Mark AI-generated code with comments
- Review before committing to version control
- Test thoroughly especially for critical code
- Document changes in commit messages
Knowledge Sharing
Share successful patterns:
- Create internal wiki with examples
- Share useful prompts/contexts
- Document model selection strategies
- Build team-specific templates
Performance Optimization
Reduce Latency
Speed up tool execution:
- Use faster models for simple tasks
- Select smaller code blocks when possible
- Increase timeout for complex operations
- Check network connection quality
Manage API Quota
Avoid hitting rate limits:
- Monitor usage in sumopod.com dashboard
- Use appropriate models for each task
- Batch similar tasks together
- Cache common results as snippets
Debugging Generated Code
Validation Checklist
Before using generated code:
- [ ] Syntax is correct
- [ ] Logic matches requirements
- [ ] Edge cases are handled
- [ ] Performance is acceptable
- [ ] Security is not compromised
- [ ] Tests pass
Common Issues
Issue: Generated code doesn't run
- Check for missing imports
- Verify variable names match
- Ensure proper indentation
Issue: Logic is incorrect
- Provide more context in original code
- Try a more powerful model
- Add comments explaining requirements
Issue: Style doesn't match project
- Run formatter after insertion
- Adjust manually for consistency
- Create style guide in comments
Advanced Use Cases
Data Pipeline Automation
python
# 1. Generate SQL with JSON to DDL
# 2. Optimize with SQL Optimizer
# 3. Add Python wrapper with ML Boilerplate
# 4. Document with Generate Docstring
# Result: Complete ETL pipelineML Experiment Tracking
python
# 1. Generate training loop with ML Boilerplate
# 2. Add type hints for better IDE support
# 3. Document with Google-style docstrings
# 4. Integrate with MLflow/Weights & BiasesLegacy Code Modernization
python
# 1. Understand with SQL/Code Explainer
# 2. Add type hints to old code
# 3. Generate docstrings
# 4. Optimize with appropriate toolsCustomization
Environment Variables
Set defaults via environment:
bash
# .env or shell profile
export SUMODATA_MODEL="claude-haiku-4-5"
export SUMODATA_TIMEOUT="60"Extension Settings
Fine-tune behavior:
json
{
"sumodata.baseUrl": "https://ai.sumopod.com",
"sumodata.timeout": 30,
"sumodata.model": "claude-haiku-4-5",
"sumodata.enableContextMenu": true,
"sumodata.maxCodeLength": 10000
}Troubleshooting Advanced Issues
Performance Problems
Slow responses:
- Switch to faster model
- Reduce code selection size
- Check network latency
- Verify API endpoint status
High API costs:
- Use cheaper models for simple tasks
- Reduce unnecessary tool usage
- Cache common results
- Monitor quota usage
Quality Issues
Inconsistent results:
- Use same model for similar tasks
- Provide consistent context
- Add detailed comments
- Try more powerful model
Incorrect output:
- Verify input code is correct
- Add more context/comments
- Try different model
- Report issue on GitHub