​

Integrations Overview

​

How Integrations Work

• Authenticate with external services
• Discover available data resources
• Access data through a unified interface
• Transform data for AI analysis

​

Built-in Integrations

​

Databases

• Full SQL query support
• Schema discovery
• Relationship mapping

• Complete database access
• Table and view discovery
• Performance optimization

• Document collection access
• Aggregation pipeline support
• Index optimization

• Local database support
• File-based storage
• Lightweight queries

​

APIs & Services

• CRM data access
• Object and field discovery
• Real-time sync capabilities

• Marketing and sales data
• Contact and deal management
• Pipeline analytics

• Payment and subscription data
• Transaction history
• Customer analytics

• Website traffic data
• User behavior analysis
• Conversion tracking

​

File Systems

• CSV, JSON, Excel support
• Directory scanning
• File metadata extraction

• Cloud storage access
• Bucket and object discovery
• Secure file handling

• Document access
• Folder structure mapping
• Collaborative file handling

​

Using Integrations

​

Setting Up an Integration

1. Choose Integration Type Select from available integrations in your dashboard
2. Configure Authentication Provide credentials (API keys, OAuth, database connections)
3. Test Connection Verify connectivity and permissions
4. Discover Resources Let Erdo scan and catalog available data

​

In Agent Workflows

from erdo import Agent, state
from erdo.actions import llm, memory

# Agent with database integration
data_analyzer = Agent(
    name="sales_analyzer",
    description="Analyzes sales data from CRM"
)

# Query data from integrated source
query_step = data_analyzer.step(
    llm.message(
        model="claude-sonnet-4",
        query=f"Analyze sales trends from: {state.database_query_result}",
        system_prompt="You are a sales analyst. Provide insights on sales performance."
    )
)

​

Data Access Patterns

# SQL databases
query_result = integration.query("SELECT * FROM sales WHERE date > '2024-01-01'")

# NoSQL databases
documents = integration.find({"status": "active"})

# APIs
data = integration.get("/api/customers")

# List available tables/collections
resources = integration.discover_resources()

# Get schema information
schema = integration.get_schema("customers")

​

Authentication Methods

​

OAuth 2.0

oauth_config = {
    "client_id": "your_client_id",
    "client_secret": "your_client_secret",
    "scopes": ["read:data", "read:metadata"]
}

​

API Keys

api_config = {
    "api_key": "your_api_key",
    "api_secret": "your_api_secret"  # if required
}

​

Database Connections

db_config = {
    "host": "localhost",
    "port": 5432,
    "database": "mydb",
    "username": "user",
    "password": "password"
}

​

Data Security

​

Credential Management

• Credentials are encrypted at rest
• Secure transmission using TLS
• Role-based access controls
• Audit logging for all access

​

Data Privacy

• Data remains in your environment
• No unnecessary data copying
• Configurable data retention
• GDPR/CCPA compliance support

​

Network Security

• VPC/private network support
• IP whitelisting capabilities
• Secure tunneling options
• Certificate-based authentication

​

Best Practices

​

Performance Optimization

• Use indexed columns in WHERE clauses
• Limit result sets appropriately
• Cache frequently accessed data
• Use connection pooling

• Close connections properly
• Monitor connection limits
• Implement retry logic
• Handle timeouts gracefully

​

Error Handling

try:
    data = integration.query(sql_query)
except ConnectionError as e:
    # Handle connection failures
    logger.error(f"Database connection failed: {e}")
    # Implement fallback or retry logic

try:
    api_data = integration.get("/api/data")
except AuthenticationError as e:
    # Handle auth failures
    logger.error(f"Authentication failed: {e}")
    # Refresh tokens or re-authenticate

​

Monitoring & Maintenance

• Regular connection testing
• Performance monitoring
• Error rate tracking
• Usage analytics

• Credential rotation
• Schema change detection
• Performance optimization
• Security updates

​

Creating Custom Integrations

1. Implement Provider Interface Create a provider that implements authentication and data access
2. Define Resource Discovery Specify how to discover and catalog available data
3. Test Integration Comprehensive testing with real data sources
4. Deploy and Register Make available for use in agents

​

Troubleshooting

​

Common Issues

• Increase timeout values
• Check network connectivity
• Verify firewall settings

• Verify credentials are correct
• Check token expiration
• Confirm required permissions

• Validate query syntax
• Check data permissions
• Verify resource exists

• Optimize queries
• Add appropriate indexes
• Implement caching
• Use connection pooling

​

Getting Help

• Check integration logs
• Review error messages
• Consult API documentation
• Contact support team