Dify.AI

​

General

​

Configuration Options

​

Prerequisites

1. Dify.AI Account: Sign up at Dify.AI or set up self-hosted instance
2. Published AI Application: Create and publish an AI app in Dify
3. API Key: Generate an App API key for your application
4. API Endpoint: Identify your Dify instance API endpoint

​

Authentication Setup

1. Navigate to your Dify.AI dashboard
2. Go to your published application
3. Navigate to API Access section
4. Copy the App API Key (not the App ID)
5. Note the API Base URL (default: https://api.dify.ai for cloud)
6. In QuickBot, add the Dify.AI block to your flow
7. Click Add Dify.AI credentials
8. Enter your API Endpoint URL
9. Enter your App API Key in the password field
10. Save the credentials

​

Configuration Parameters

• API Endpoint: Dify instance URL (default: https://api.dify.ai)
• App API Key: Your Dify application API key (required, stored encrypted)
• Query: User message to send to the AI assistant (supports variables)
• Conversation ID: Optional conversation identifier for context persistence
• User: Unique user identifier for conversation tracking
• Inputs: Custom inputs for your Dify application variables

​

Features

​

Core Capabilities

​

Create Chat Message

• Streaming Responses: Real-time message streaming for immediate user feedback
• Context Persistence: Maintain conversation context across multiple interactions
• Custom Inputs: Send custom variables and parameters to your Dify app
• Multi-format Support: Handle text, markdown, and rich content responses
• Token Tracking: Monitor and track AI model token usage

​

Conversation Management

• Automatic ID Generation: Automatically creates new conversation IDs when needed
• Context Continuity: Maintains conversation history for contextual responses
• User Identification: Track and associate conversations with specific users
• Session Handling: Manage long-running conversations and session state

​

Variable Integration

• Dynamic Queries: Support variable substitution in user queries
• Response Mapping: Map AI responses to multiple QuickBot variables
• Structured Outputs: Extract specific data points from AI responses
• Real-time Updates: Variables updated immediately during streaming responses

​

Data Handling

• Stream Processing: Real-time processing of streaming AI responses
• Link Detection: Automatic conversion of URLs to markdown links
• Text Formatting: Preserve formatting and structure in AI responses
• Error Recovery: Graceful handling of API failures and network issues

​

Advanced AI Features

• Knowledge Base Integration: Leverage Dify’s document knowledge capabilities
• Workflow Automation: Trigger complex AI workflows from simple queries
• Multi-model Support: Access various AI models through Dify’s platform
• Custom Instructions: Use pre-configured AI assistant personalities and behaviors

​

Advanced Features

​

Streaming Architecture

• Real-time Responses: Stream AI responses as they’re generated
• Progressive Display: Show partial responses to users immediately
• Connection Management: Robust handling of streaming connections
• Buffer Management: Efficient processing of streamed data chunks

​

Conversation Persistence

• Context Memory: AI assistants remember previous conversation context
• Cross-session Continuity: Maintain conversations across bot restarts
• User-specific Contexts: Separate conversation contexts per user
• Variable Persistence: Automatic conversation ID management

​

Error Handling & Reliability

• HTTP Error Management: Comprehensive handling of API errors
• Stream Error Recovery: Graceful handling of streaming interruptions
• Retry Logic: Automatic retry for transient failures
• Detailed Logging: Comprehensive error reporting and debugging information

​

Performance Optimization

• Efficient Streaming: Optimized stream processing for minimal latency
• Connection Pooling: Efficient API connection management
• Response Caching: Optional caching for frequently used responses
• Token Optimization: Monitor and optimize AI model token usage

​

Actions Reference

​

Create Chat Message

• Query: The user message or question to send to the AI assistant

• Conversation ID: Variable to store/retrieve conversation context
• User: Unique identifier for the user (for conversation tracking)
• Inputs: Custom key-value pairs for Dify app variables
• Response Mapping: Map different response elements to QuickBot variables

• Answer: The main AI response content
• Conversation ID: Unique identifier for the conversation session
• Total Tokens: Token count used by the AI model

Query: "{{userQuestion}}"
Conversation ID: conversationId (variable)
User: "{{userEmail}}"
Inputs:
  - Key: "context", Value: "{{previousContext}}"
  - Key: "language", Value: "{{userLanguage}}"
Response Mapping:
  - Answer → aiResponse
  - Conversation ID → conversationId
  - Total Tokens → tokenUsage

​

Streaming vs Non-Streaming

• Real-time response display
• Progressive user feedback
• Better user experience for long responses
• Immediate variable updates

• Complete response before display
• Simpler integration for basic use cases
• Better for short responses
• Batch variable updates

​

Best Practices

​

Implementation Recommendations

1. Conversation Design: Design conversation flows that leverage AI context effectively
2. User Identification: Use consistent user identifiers for conversation continuity
3. Input Optimization: Structure custom inputs to maximize AI assistant performance
4. Response Handling: Plan for variable-length AI responses in your flow design

​

Security Best Practices

1. API Key Protection: Never expose App API keys in client-side code
2. User Privacy: Be mindful of user data sent to AI services
3. Data Retention: Understand Dify’s data retention policies
4. Access Control: Use appropriate API key permissions and restrictions

​

Performance Guidelines

1. Query Optimization: Write clear, specific queries for better AI responses
2. Context Management: Balance conversation context with performance
3. Token Monitoring: Monitor token usage to manage costs effectively
4. Streaming Benefits: Use streaming for better user experience with longer responses

​

AI Assistant Guidelines

1. Clear Instructions: Provide clear instructions in your Dify app configuration
2. Knowledge Base: Optimize your knowledge base for relevant, accurate responses
3. Testing: Thoroughly test AI responses across different scenarios
4. Fallback Handling: Implement fallbacks for when AI responses are unclear

​

Troubleshooting

​

Common Issues

​

Authentication Problems

• Invalid API Key: Verify the App API key is correctly copied from Dify
• Wrong Endpoint: Ensure the API endpoint matches your Dify instance
• App Not Published: Verify your Dify application is published and accessible
• Permission Denied: Check API key permissions and app access settings

​

Connection and Streaming Issues

• Stream Interruption: Check network stability and connection reliability
• Timeout Errors: Verify network connectivity to Dify API endpoints
• Incomplete Responses: Check for streaming connection issues
• Connection Refused: Verify Dify service availability and endpoint correctness

​

AI Response Issues

• Poor Response Quality: Review AI assistant instructions and knowledge base
• Context Loss: Check conversation ID persistence and variable handling
• Irrelevant Responses: Optimize knowledge base and AI assistant training
• Response Length Issues: Adjust AI assistant settings for response length

​

Variable and Integration Issues

• Variables Not Set: Verify response mapping configuration
• Conversation ID Problems: Check conversation variable persistence
• Input Parameter Issues: Verify custom input format and content
• Token Tracking Issues: Check token usage tracking and limits

​

Debugging Steps

1. Test API Access: Verify API key and endpoint access directly
2. Check Dify App: Confirm your Dify application is working properly
3. Review Logs: Check QuickBot logs for detailed error messages
4. Validate Inputs: Ensure all input parameters are properly formatted
5. Test Streaming: Verify streaming functionality with simple queries

​

Error Messages

• “Failed to read response stream”: Check network connectivity and API status
• HTTP error responses: Check Dify API documentation for specific error codes
• “API key is required”: Ensure App API key is configured in credentials
• Stream processing errors: Check for network interruptions during streaming
• Conversation context errors: Verify conversation ID variable configuration

​

Performance Issues

• Slow AI Responses: Check Dify service status and AI model performance
• High Token Usage: Optimize queries and conversation context management
• Memory Issues: Monitor conversation context size and cleanup old conversations
• Connection Timeouts: Implement appropriate timeout settings for AI responses