# Smart Defaults for Elicitation Forms
The Smart Defaults system learns from user patterns and organizational conventions to pre-populate elicitation forms, significantly improving user experience by reducing repetitive input.
## Overview
Smart Defaults uses three layers of intelligence to suggest values:
1. **Template-based defaults** - Pre-configured best practices for common operations
2. **Pattern-based defaults** - Learned from analyzing existing schemas and naming conventions
3. **History-based defaults** - Learned from user choices and feedback
## Features
### Pattern Recognition
- Detects naming conventions (camelCase, snake_case, hyphenated, etc.)
- Identifies common prefixes and suffixes
- Analyzes field patterns across schemas
- Learns organizational conventions
### Context-Aware Suggestions
- Different defaults for dev/staging/production environments
- Team and project-specific preferences
- Compatibility mode suggestions based on context
- Smart field type detection
### Learning Engine
- Tracks user choices and acceptance rates
- Adjusts confidence scores based on feedback
- Persists learning data locally
- Privacy-aware with configurable exclusions
### Seamless Integration
- Works transparently with existing elicitation forms
- Shows confidence scores and reasoning
- Allows users to accept or override suggestions
- No changes required to existing workflows
## How It Works
When an elicitation form is created, the Smart Defaults system:
1. **Analyzes the context** - Determines the operation type and environment
2. **Gathers suggestions** - Combines template, pattern, and history-based defaults
3. **Enhances the form** - Adds suggested values with confidence scores
4. **Presents to user** - Shows suggestions in the elicitation form
5. **Learns from feedback** - Records whether suggestions were accepted
## Configuration
Smart Defaults can be configured via environment variables or a JSON configuration file.
### Environment Variables
```bash
# Enable/disable smart defaults
export SMART_DEFAULTS_ENABLED=true
# Enable specific features
export SMART_DEFAULTS_PATTERN_RECOGNITION=true
export SMART_DEFAULTS_LEARNING=true
# Set confidence thresholds
export SMART_DEFAULTS_MIN_CONFIDENCE=0.3
export SMART_DEFAULTS_AUTO_FILL_CONFIDENCE=0.7
# Privacy settings
export SMART_DEFAULTS_ANONYMIZE=false
export SMART_DEFAULTS_EXCLUDED_FIELDS=password,secret,token
export SMART_DEFAULTS_EXCLUDED_CONTEXTS=sensitive-data
# UI settings
export SMART_DEFAULTS_SHOW_CONFIDENCE=true
```
### Configuration File
Create `~/.kafka-schema-mcp/smart_defaults_config.json`:
```json
{
"enabled": true,
"enable_pattern_recognition": true,
"enable_learning": true,
"min_confidence_for_suggestion": 0.3,
"min_confidence_for_auto_fill": 0.7,
"show_confidence_scores": true,
"excluded_fields": ["password", "secret", "token"],
"environment_defaults": {
"production": {
"compatibility": "FULL",
"dry_run": true
},
"development": {
"compatibility": "NONE",
"dry_run": false
}
}
}
```
## Usage Examples
### Schema Registration with Smart Defaults
When registering a schema, the system will suggest:
- Subject naming based on detected patterns
- Compatibility mode based on context
- Common fields based on schema type
```python
# The elicitation form will show:
# Subject: [order-events-created] (suggested - 85% confidence)
# Compatibility: [BACKWARD] (suggested - 70% confidence)
# Schema Type: [AVRO] (default - 80% confidence)
```
### Migration with Learned Preferences
For migrations, the system remembers your preferences:
- Preserve IDs setting from previous migrations
- Dry run preferences
- Version migration patterns
### Context Creation with Metadata
When creating contexts, get suggestions for:
- Environment detection from context name
- Owner based on similar contexts
- Tags from organizational patterns
## Privacy and Security
Smart Defaults is designed with privacy in mind:
- **Local storage only** - All learning data is stored locally
- **Configurable exclusions** - Exclude sensitive fields from learning
- **No external data sharing** - Patterns stay within your organization
- **Anonymization option** - Can anonymize stored values
- **Retention controls** - Automatic cleanup of old data
### Excluded Fields
By default, the following field patterns are excluded from learning:
- password
- secret
- key
- token
- credential
### Data Retention
Learning data is retained for 90 days by default. This can be configured via:
```bash
export SMART_DEFAULTS_RETENTION_DAYS=30
```
## Implementation Details
### Pattern Detection
The system uses regular expressions and statistical analysis to detect:
- Naming conventions (75%+ occurrence rate)
- Common prefixes/suffixes (40%+ occurrence rate)
- Field type patterns (appears in 2+ schemas)
### Confidence Scoring
Confidence scores are calculated based on:
- **Pattern strength** - How consistently a pattern appears
- **Historical acceptance** - How often users accept suggestions
- **Recency** - Recent choices are weighted higher
- **Context match** - Higher confidence for exact context matches
### Learning Algorithm
The learning engine uses:
- **Incremental updates** - Learns from each user interaction
- **Decay factor** - Old patterns gradually lose weight
- **Feedback loops** - Acceptance/rejection affects future confidence
## Troubleshooting
### Suggestions Not Appearing
1. Check if smart defaults are enabled:
```bash
export SMART_DEFAULTS_ENABLED=true
```
2. Verify minimum occurrences are met:
- Patterns need 2+ occurrences by default
- Adjust via `minimum_occurrences_for_pattern`
3. Check confidence thresholds:
- Suggestions below `min_confidence_for_suggestion` are hidden
### Incorrect Suggestions
1. Clear learning history:
```bash
rm -rf ~/.kafka-schema-mcp/smart_defaults/learning_history.json
```
2. Exclude problematic fields:
```bash
export SMART_DEFAULTS_EXCLUDED_FIELDS=field1,field2
```
3. Adjust confidence thresholds for better accuracy
### Performance Issues
1. Disable caching if needed:
```json
{"enable_caching": false}
```
2. Reduce pattern tracking:
```json
{"max_patterns_to_track": 20}
```
3. Use async learning:
```json
{"async_learning": true}
```
## Future Enhancements
- Machine learning models for better predictions
- Cross-registry pattern sharing (opt-in)
- A/B testing for suggestion effectiveness
- Integration with schema documentation
- Custom pattern definitions
- Team-specific learning profiles
## Contributing
To contribute to Smart Defaults development:
1. Check the implementation in `smart_defaults.py`
2. Review integration in `smart_defaults_integration.py`
3. Add tests to `tests/test_smart_defaults.py`
4. Update configuration in `smart_defaults_config.py`
## License
This feature is part of the Kafka Schema Registry MCP server and follows the same license terms.
