Knowledge Base
The knowledge base is the “brain” of the Mendix Expert server. It contains curated, structured information about Mendix development.
Structure
knowledge/
├── best-practices.json # Coding patterns, guidelines
├── studio-pro.json # Studio Pro features, UI
├── model-sdk.json # Model SDK programming
├── platform-sdk.json # Platform SDK operations
├── troubleshooting.json # Common issues, solutions
├── advanced-patterns.json # Complex techniques
├── performance-guide.json # Optimization strategies
├── security-guide.json # Security best practices
└── harvested-*.json # Auto-crawled content
Entry Format
Every knowledge entry follows this structure:
{
"title": "Loop Pattern for Microflows",
"description": "How to iterate over lists in microflows",
"content": "Detailed explanation...",
"when_to_use": "When you need to process items one by one",
"best_practices": ["Use IterableList on loopSource", "Avoid deprecated loopVariableName"],
"examples": [
{
"description": "Basic loop",
"code": "loopAction.loopSource = new microflows.IterableList(...);"
}
],
"related_topics": ["microflows", "lists", "iteration"],
"mendix_version": "10.0+",
"_metadata": {
"id": "abc123",
"added": "2024-01-15T10:30:00Z",
"source": "docs.mendix.com",
"quality_score": 0.95,
"verified": true
}
}
Content Categories
Best Practices (best-practices.json)
Coding patterns and guidelines:
- Microflow design patterns
- Error handling strategies
- Naming conventions
- Performance optimization
- Code organization
Studio Pro (studio-pro.json)
Studio Pro features and usage:
- UI element configuration
- Modeling techniques
- Version-specific features
- Keyboard shortcuts
- Pro tips
Model SDK (model-sdk.json)
SDK programming patterns:
- Creating/modifying elements
- Navigation patterns
- Working with transactions
- Common pitfalls
- Real-world examples
Platform SDK (platform-sdk.json)
Platform operations:
- Creating apps
- Working copies
- Branch management
- Deployment operations
- Team Server integration
Troubleshooting (troubleshooting.json)
Common issues and fixes:
- Error messages explained
- Known bugs/workarounds
- Migration issues
- Performance problems
Harvested Content (harvested-*.json)
Auto-crawled from official sources:
- Release notes (10.x, 11.x)
- Documentation updates
- New feature announcements
- API changes
Quality Scoring
Every entry is automatically scored on:
| Factor | Weight | Description |
|---|---|---|
| Completeness | 30% | Has all relevant fields |
| Source quality | 25% | Official docs score higher |
| Verification | 20% | Manually verified entries |
| Freshness | 15% | Recent updates preferred |
| Usage | 10% | Frequently accessed entries |
Thresholds:
>= 0.8- High quality, trusted0.5 - 0.8- Good quality< 0.5- May need review
Coverage Statistics
Current knowledge base metrics:
| Category | Entries | Coverage |
|---|---|---|
| Best Practices | ~50 | Core patterns |
| Studio Pro | ~40 | Major features |
| Model SDK | ~60 | Common operations |
| Platform SDK | ~30 | Key operations |
| Troubleshooting | ~25 | Common issues |
| Harvested | ~100+ | Auto-updated |
Total: 300+ curated entries + 318 vector embeddings
Knowledge Gaps
The system tracks “missed queries” - searches that returned no results. These indicate knowledge gaps to fill.
View gaps:
@mendix-expert hello
# Shows recent missed queries in analytics
Common gaps are automatically added to the harvest priority list.
Validation
The knowledge base is automatically validated:
- Syntax errors - Invalid JSON
- Missing required fields - No title or description
- Staleness - Not updated in 6+ months
- Duplicates - Similar entries detected
- Broken references - Invalid cross-references
Run manual validation:
@mendix-expert Check the knowledge validation report
# Returns mendix://validation/report
Adding Knowledge
Manual Addition
@mendix-expert add_to_knowledge_base
knowledge_file="best-practices"
category="microflows"
content='{
"title": "My Pattern",
"description": "How to do X",
"when_to_use": "When you need Y"
}'
source="discovered through research"
Auto-Learning
When Beast Mode research discovers new information, the AI is instructed to save it:
I found this pattern in the SDK demos:
[Pattern details]
Let me save this to the knowledge base...
@mendix-expert add_to_knowledge_base ...
Harvesting
Weekly automatic harvests add fresh content from:
- docs.mendix.com
- Release notes
- API documentation
Keeping It Current
| Mechanism | Frequency | What It Does |
|---|---|---|
| Auto-Harvest | Weekly | Crawls official docs |
| Self-Learning | On-demand | Saves discoveries |
| Validation | Daily | Flags issues |
| Vector Re-index | After changes | Updates embeddings |