Google Translate

Advanced Features

Overview

Node Sphere provides advanced features that enable sophisticated AI agent capabilities beyond basic social media automation. These features allow you to create more intelligent, context-aware agents that can access external data, utilize custom knowledge bases, and integrate with external tools.

Knowledge Management

What is Knowledge Management?

Knowledge Management allows you to provide your agents with custom information that they can reference when generating responses. This creates more accurate, contextually relevant outputs tailored to your specific domain or use case.

Adding Knowledge to Your Agent

  1. Navigate to Agent Configuration: Open your agent's configuration page
  2. Go to Knowledge Tab: Click on the "Knowledge" tab
  3. Add New Knowledge: Click "Add Knowledge" button

Knowledge Configuration

Title: A descriptive name for your knowledge item

Content: The actual information you want your agent to know

Source: Document upload or manual entry

Knowledge Types

Manual Knowledge

  • Use Case: Custom information, product details, company policies
  • Example: Product specifications, FAQ content, brand guidelines
  • Best Practice: Keep content focused and well-structured

Document Upload

  • Supported Formats: PDF, TXT, DOCX, MD
  • Use Case: Existing documentation, manuals, knowledge bases
  • Processing: Automatically chunked and embedded for semantic search

Knowledge Search in Tasks

Enable knowledge search in your agent tasks to allow the agent to reference relevant information:

  1. Task Configuration: Edit your agent task
  2. Enable Knowledge Search: Toggle "Use Knowledge Search" option

Best Practices for Knowledge Management

  1. Organize by Topic: Group related information together
  2. Keep Content Current: Regularly update time-sensitive information
  3. Use Clear Titles: Make knowledge items easy to identify
  4. Test Relevance: Verify that knowledge search returns relevant results

Data Sources

What are Data Sources?

Data Sources allow your agents to fetch real-time information from external APIs and services. This enables dynamic, up-to-date responses based on current data.

Configuring Data Sources

  1. Navigate to Agent Configuration: Open your agent's settings
  2. Go to Data Sources Tab: Click on "Data Sources"
  3. Add New Data Source: Click "Add Data Source" button

Data Source Configuration

Basic Settings

  • Name: Descriptive name for the data source
  • Endpoint: The API URL to fetch data from
  • Method: HTTP method (GET, POST, PUT, DELETE)
  • Cache Duration: How long to cache responses (in minutes)

Authentication

  • Headers: Add authentication headers (API keys, tokens)
  • Parameters: Query parameters or form data
  • Body: Request body for POST/PUT requests

Response Processing

  • Response Data Path: JSONPath expression to extract specific data
  • Example: $.weather.temperature to get temperature from weather API

Data Source Examples

Weather Information

{
  "name": "Current Weather",
  "endpoint": "https://api.openweathermap.org/data/2.5/weather",
  "method": "GET",
  "parameters": [
    {"name": "q", "value": "New York"},
    {"name": "appid", "value": "YOUR_API_KEY"}
  ],
  "responseDataPath": "$.main.temp",
  "cacheDurationMinutes": 30
}

Stock Prices

{
  "name": "Stock Price",
  "endpoint": "https://api.example.com/stock/AAPL",
  "method": "GET",
  "headers": [
    {"name": "Authorization", "value": "Bearer YOUR_TOKEN"}
  ],
  "responseDataPath": "$.price",
  "cacheDurationMinutes": 5
}

News Headlines

{
  "name": "Latest News",
  "endpoint": "https://newsapi.org/v2/top-headlines",
  "method": "GET",
  "parameters": [
    {"name": "country", "value": "us"},
    {"name": "apiKey", "value": "YOUR_API_KEY"}
  ],
  "responseDataPath": "$.articles[0].title",
  "cacheDurationMinutes": 60
}

Using Data Sources in Prompts

Node Sphere provides flexible ways to access data source content in your prompt templates:

Raw Content Access

Use DataSource.{Name}Raw to access the complete, unprocessed response from your data source:

Raw weather data: {{DataSource.WeatherRaw}}
Complete API response: {{DataSource.NewsAPIRaw}}

This is useful when you need:

  • Full XML or JSON responses
  • Complete text content without processing
  • Raw string data for custom parsing

Structured Data Access

Use DataSource.{Name}.{Property} to access specific fields from structured JSON responses:

Current temperature: {{DataSource.Weather.main.temp}}°F
Stock price: ${{DataSource.MarketData.Price}}
Latest headline: {{DataSource.News.articles[0].title}}

Top news headlines:
{{#each DataSource.News.articles}}
- {{this.title}} ({{this.source.name}})
{{/each}}

Stock prices:
{{#each DataSource.StockList}}
{{this.symbol}}: ${{this.price}} ({{this.change}})
{{/each}}

This allows you to:

  • Extract specific values from JSON responses
  • Access nested object properties
  • Get array elements by index
  • Loop through arrays of data with handlebars syntax
  • Create dynamic lists and formatted content

Dynamic Functions

Enable "AI Function with Parameters" to allow the AI model to dynamically call your data sources based on the context of the conversation. This enables intelligent, context-aware data retrieval where the AI determines when and how to fetch external information.

MCP (Model Context Protocol) Servers

What is MCP?

Model Context Protocol (MCP) is a standard for connecting AI models to external tools and data sources. MCP servers provide specific capabilities that your agents can use during conversations.

Creating Custom MCP Servers

You can build your own MCP servers using the official SDKs provided by the Model Context Protocol. The protocol supports multiple programming languages:

Available SDKs

  • Python SDK: For Python-based integrations
  • TypeScript SDK: For Node.js and web applications
  • C# SDK: For .NET applications
  • Java SDK: For Java applications
  • Kotlin SDK: For Kotlin/Android applications
  • Swift SDK: For iOS/macOS applications
  • Ruby SDK: For Ruby applications

Getting Started with Custom Servers

  1. Choose Your SDK: Visit modelcontextprotocol.io to select the appropriate SDK
  2. Follow Quickstart Guide: Use the official quickstart documentation
  3. Implement Your Logic: Create tools, resources, and prompts for your specific use case
  4. Test with MCP Inspector: Use the debugging tools to test your server
  5. Deploy and Connect: Host your server and connect it to Node Sphere

Configuring MCP Servers

  1. Navigate to Agent Configuration: Open your agent's settings
  2. Go to MCP Servers Tab: Click on "MCP Servers"
  3. Add New MCP Server: Click "Add MCP Server" button

MCP Server Configuration

Basic Settings

  • Name: Descriptive name for the MCP server
  • Endpoint: Connection endpoint for the MCP server
  • Transport Type: SSE (Server-Sent Events)
  • Active: Enable/disable the MCP server

Authentication

  • Headers: Custom headers for authentication
  • Credentials: API keys or tokens if required

Common MCP Servers

Pipedream MCP Servers

  • Purpose: Connect to thousands of APIs and services
  • Use Case: Integration with popular services like GitHub, Slack, Google Sheets, and more
  • Platform: Pipedream MCP provides pre-built MCP servers
  • Benefits: No-code integration with 2000+ apps and APIs

Zapier MCP Servers

  • Purpose: Connect AI assistants to over 7,000+ apps and 30,000+ actions
  • Use Case: Enable AI to perform real tasks like sending messages, managing data, scheduling events, and updating records
  • Platform: Zapier MCP provides direct access to Zapier's vast integration network
  • Benefits:
    • No complex API integrations required
    • Built-in authentication and security
    • Free tier with 300 tool calls per month
    • Transform AI from conversational tool to functional extension
  • Supported Apps: Gmail, Slack, Google Workspace, HubSpot, Salesforce, Notion, Trello, and thousands more

MCP Server Examples

Web Search MCP

{
  "name": "Web Search",
  "endpoint": "http://yourdomain/mcp/search",
  "transportType": "SSE",
  "headers": [
    {"name": "Authorization", "value": "Bearer YOUR_TOKEN"}
  ],
  "isActive": true
}

Pipedream Integration MCP

{
  "name": "Pipedream GitHub",
  "endpoint": "https://mcp.pipedream.com/github",
  "transportType": "SSE",
  "isActive": true
}

Zapier MCP Integration

{
  "name": "Zapier Actions",
  "endpoint": "https://your-zapier-mcp-endpoint.zapier.com",
  "transportType": "SSE",
  "headers": [
    {"name": "Authorization", "value": "Bearer YOUR_ZAPIER_MCP_TOKEN"}
  ],
  "isActive": true
}

Multi-Modal Content Generation

Image Generation

Configure image generation for your agents to create visual content automatically.

Supported Providers

  • HuggingFace: Free and paid models
  • OpenAI: DALL-E models
  • xAI: Image generation models
  • Pi API: Midjourney, Flux AI, and other models

Configuration

  1. Go to Agent Configuration: Open your agent settings
  2. Image Generation Tab: Configure image model settings
  3. Model Selection: Choose your preferred image model
  4. API Configuration: Add your API credentials

Video Generation

Create short video content for platforms like TikTok and YouTube Shorts.

Supported Providers

  • Pi API: Kling, WanX, Hunyuan, Luma and other video generation models

Video Task Configuration

  • Video Prompt Template: Description for video generation
  • Use AI Audio: Generate audio and sound effects for the video
  • Video Styles: Predefined style options

Advanced Task Configuration

Command-Based Interactions

Set up commands that users can trigger to interact with your agents.

Command Configuration

  1. Task Type: Select appropriate task type (Discord Bot, Telegram Bot, etc.)
  2. Commands Tab: Add custom commands
  3. Command Settings:
    • Command: The trigger word/phrase
    • Match Type: Exact, Contains, Regex
    • Response: How the agent should respond

Match Types Explained

Exact Match:

  • Triggers only when the message exactly matches the command
  • Case-sensitive by default
  • Example: /weather only triggers on exactly /weather

Contains Match:

  • Triggers when the message contains the command anywhere within it
  • Useful for natural language interactions
  • Example: weather triggers on "What's the weather like?" or "Check weather please"

Regex Match:

  • Uses regular expressions for advanced pattern matching
  • Allows complex matching rules and parameter extraction
  • Perfect for understanding natural language patterns and user intent
  • Example: (?i).*(?:what|how).*(?:price|cost).* matches various ways users ask about pricing

Example Commands

Exact Command:

{
  "command": "/weather",
  "matchType": "Exact",
  "description": "Get current weather",
  "response": "{{DataSource.Weather}}"
}

Natural Language Command (Contains):

{
  "command": "weather",
  "matchType": "Contains", 
  "description": "Weather information when mentioned",
  "response": "Current weather: {{DataSource.Weather}}"
}

Advanced Regex Command:

{
  "command": "(?i).*(?:what|how).*(?:price|cost).*(?:of|for)\\s+(.+?)(?:\\?|$)",
  "matchType": "Regex",
  "description": "Detect price inquiries in natural language",
  "response": "Let me check the price for {{$1}}. {{DataSource.PriceCheck}}"
}

Help Command (Included):

{
  "command": "/help",
  "matchType": "Exact",
  "description": "Show help information",
  "response": "Available commands: /weather, /news, /joke"
}
Previous
Getting Started
Next
Best Practices
In this document