ontology-mcp-self-healing
Self-healing MCP server that monitors database schema changes, detects differences, and automatically updates ontology mappings using AI to ensure agents continue working without manual intervention.
README
Self-Healing Ontology MCP Agent System
A production-ready self-healing multi-agent system that uses ontologies and MCP (Model Context Protocol) to automatically adapt when database schemas change.
Overview
This system solves a critical problem in modern AI agent deployments: when database schemas change, agent queries break. Instead of manually updating every agent and query, this system:
- Monitors database schemas continuously
- Detects schema changes automatically
- Analyzes changes using Claude AI
- Heals ontology mappings automatically
- Reloads MCP tools without downtime
The result: agents continue working seamlessly even when databases evolve.
Architecture
┌─────────────┐ ┌──────────────┐ ┌─────────────┐
│ Agents │─────▶│ MCP Server │─────▶│ Database │
│ (Analytics, │ │ (Ontology) │ │ (SQLite, │
│ Support) │ │ │ │ PostgreSQL)│
└─────────────┘ └──────┬───────┘ └─────────────┘
│
▼
┌─────────────────┐
│ Schema Monitor │
│ (SHA-256 Hash) │
└────────┬────────┘
│
▼
┌─────────────────┐
│ Diff Engine │
│ (Change Detect) │
└────────┬────────┘
│
▼
┌─────────────────┐
│ Ontology Remap │
│ (Claude AI) │
└─────────────────┘
Features
- ✅ Automatic Schema Change Detection - SHA-256 hash-based monitoring
- ✅ Intelligent Diff Analysis - Detects renames, additions, deletions
- ✅ AI-Powered Healing - Uses Claude to update ontology mappings
- ✅ MCP Protocol Support - Native Model Context Protocol integration
- ✅ Multi-Agent Support - Shared semantic understanding across agents
- ✅ Hot Reload - MCP server reloads without downtime
- ✅ Audit Logging - Complete JSON audit trail
- ✅ Alert Integration - Slack/Teams webhook support
- ✅ Production Ready - Docker, tests, error handling
Quickstart (< 5 minutes)
1. Install Dependencies
pip install -r requirements.txt
2. Set Up Environment
# Create .env file
echo "ANTHROPIC_API_KEY=your_api_key_here" > .env
3. Initialize Database and Ontology
# Create sample database
python scripts/init_db.py
# Generate initial ontology
python scripts/setup_ontology.py
4. Run Quickstart Example
python examples/quickstart.py
You should see:
- ✓ MCP Server initialized
- ✓ Tools generated from ontology
- ✓ Schema monitoring active
- ✓ Agent system ready
5. Test Schema Change Healing
# In another terminal, modify the database schema
sqlite3 test_database.db "ALTER TABLE customers ADD COLUMN phone TEXT;"
# Watch the system automatically detect and heal
python examples/full_system.py
Installation
From Source
git clone https://github.com/yourusername/ontology-mcp-self-healing.git
cd ontology-mcp-self-healing
pip install -r requirements.txt
pip install -e .
Using Docker
# Build and run
docker-compose up -d
# View logs
docker-compose logs -f
Configuration
Configuration is managed via config/config.yaml:
# Database Configuration
database:
type: sqlite # sqlite, postgresql, mysql
connection_string: sqlite:///./test_database.db
# Ontology Configuration
ontology:
main_file: ontologies/business_domain.owl
auto_reload: true
# Schema Monitoring
monitoring:
enabled: true
check_interval: 60 # seconds
detect_renames: true
# Auto-Healing
healing:
enabled: true
auto_approve: false # Set to true for automatic healing
claude_model: claude-3-5-sonnet-20241022
validation_enabled: true
# Alerts
alerts:
enabled: true
webhook_url: ${ALERT_WEBHOOK_URL} # Optional
Usage Examples
Basic MCP Server
from src.mcp_server.server import OntologyMCPServer
# Initialize server
server = OntologyMCPServer()
# Get available tools
tools = server.get_tools()
# Execute query
result = await server.execute_tool(
"query_order",
{"query": "all orders", "limit": 10}
)
Create an Agent
from src.mcp_server.server import OntologyMCPServer
from src.agents.examples.analytics_agent import AnalyticsAgent
# Initialize MCP server
mcp_server = OntologyMCPServer()
# Create agent
agent = AnalyticsAgent(mcp_server, claude_api_key="your_key")
# Query using natural language
response = await agent.query("What are the total sales for last month?")
Full Self-Healing System
from src.system.self_healing import SelfHealingAgentSystem
# Initialize system
system = SelfHealingAgentSystem()
# Start monitoring and healing
system.start()
# System runs forever, auto-healing on schema changes
Architecture Details
MCP Server
The MCP server loads OWL ontologies and generates tools dynamically:
- Extracts class → table mappings
- Extracts property → column mappings
- Translates semantic queries to SQL
- Caches queries for performance
Schema Monitor
Continuous monitoring using:
- SQLAlchemy inspector for schema capture
- SHA-256 hashing for change detection
- Configurable check intervals
- Event callbacks on changes
Diff Engine
Intelligent diff computation:
- Detects table/column additions/removals
- Heuristic-based rename detection
- Type change detection
- Detailed diff reporting
Auto Remapper
AI-powered ontology healing:
- Extracts current ontology mappings
- Generates LLM prompts with schema changes
- Validates proposed RDF triples
- Applies updates to ontology files
- Supports manual approval mode
Production Deployment
Docker Deployment
# Build image
docker build -t ontology-mcp-self-healing .
# Run container
docker run -d \
-e ANTHROPIC_API_KEY=your_key \
-v $(pwd)/ontologies:/app/ontologies \
-v $(pwd)/logs:/app/logs \
ontology-mcp-self-healing
Docker Compose
# Start all services
docker-compose up -d
# View logs
docker-compose logs -f self-healing
# Stop services
docker-compose down
See docs/deployment.md for Kubernetes, Helm, and production best practices.
Testing
# Run all tests
pytest tests/ -v
# Run with coverage
pytest tests/ --cov=src --cov-report=html
# Run specific test
pytest tests/test_mcp_server.py -v
Documentation
- Architecture Guide - Detailed system architecture
- Deployment Guide - Production deployment instructions
- Troubleshooting - Common issues and solutions
Project Structure
ontology-mcp-self-healing/
├── README.md
├── requirements.txt
├── setup.py
├── docker-compose.yml
├── Dockerfile
├── config/
│ └── config.yaml
├── ontologies/
│ └── business_domain.owl
├── src/
│ ├── mcp_server/ # MCP server implementation
│ ├── monitoring/ # Schema monitoring
│ ├── healing/ # Auto-healing
│ ├── system/ # Orchestration
│ └── agents/ # Agent implementations
├── tests/ # Test suite
├── examples/ # Example scripts
├── scripts/ # Setup scripts
└── docs/ # Documentation
Getting Started
New to this project? Follow our comprehensive SETUP_GUIDE.md for step-by-step instructions on:
- Cloning the repository
- Setting up your environment
- Running examples
- Running tests
- Troubleshooting common issues
Contributing
Contributions are welcome! Please see CONTRIBUTING.md for guidelines.
License
MIT License - see LICENSE file for details.
Publishing to GitHub
Want to publish this project? See GITHUB_SETUP.md for step-by-step instructions.
Acknowledgments
- Built with owlready2 for ontology management
- Uses Anthropic Claude for AI-powered healing
- Implements MCP Protocol for agent communication
- Powered by LangChain for agent orchestration
Related Articles
- Medium Article: Self-Healing AI Agents - Deep dive into the system design
- Blog Post: MCP for Production - Using MCP in production systems
Made with ❤️ for the AI agent community
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.