🎉 Syllabi - Open Source AI Chatbot Platform. Check it out →
docs
User Guide
Skills & Actions

Skills & Actions

Extend your chatbot's capabilities with custom skills and actions.

Overview

Skills allow your chatbot to:

  • 🔍 Search external databases
  • 📧 Send emails
  • 📅 Check calendars
  • 🌤️ Get real-time data (weather, stock prices)
  • 🔗 Call external APIs
  • ⚡ Execute custom logic

Built-in Skills

Document Retrieval (Default)

Automatically searches uploaded documents:

How it works:

  1. User asks question
  2. AI generates search query
  3. Retrieves relevant document chunks
  4. Generates answer with citations

Configuration:

Max chunks: 5-10
Similarity threshold: 0.7

Web Search

Search the internet for current information:

Setup:

  1. Skills tab → Add Skill → Web Search
  2. Configure:
    • Search engine: Google, Bing, DuckDuckGo
    • Max results: 3-5
    • Safe search: On/Off
  3. Save

Example use:

User: "What's the current weather in Tokyo?"
Bot: [Searches web] "The current temperature in Tokyo is 18°C with partly cloudy skies."

Calendar Integration

Check availability and schedule meetings:

Setup:

  1. Skills → Add Skill → Calendar
  2. Connect calendar:
    • Google Calendar
    • Microsoft Outlook
    • Apple Calendar (via CalDAV)
  3. Set permissions (read/write)
  4. Save

Example use:

User: "Am I free at 2pm tomorrow?"
Bot: [Checks calendar] "Yes, you're available at 2pm tomorrow."

User: "Schedule a meeting at 2pm tomorrow"
Bot: [Creates event] "Meeting scheduled for 2pm tomorrow."

Email

Send emails on behalf of users:

Setup:

  1. Skills → Email
  2. Configure SMTP:
  3. Save

Example use:

User: "Email support about my order #12345"
Bot: [Composes email]
"Subject: Order #12345 Inquiry
Body: [Generated based on conversation]

Send this email? (Yes/No)"

Database Query

Query your databases:

Setup:

  1. Skills → Database
  2. Connection:
    • Type: PostgreSQL, MySQL, MongoDB
    • Host: your-db.com
    • Credentials: username/password
  3. Allowed operations: SELECT only (recommended)
  4. Save

Example use:

User: "How many orders did we receive today?"
Bot: [Queries database] "You received 47 orders today."

Security: Use read-only credentials, limit to specific tables.

Creating Custom Skills

Simple Function Skill

Define custom JavaScript/Python functions:

Example: Calculate Shipping Cost

JavaScript:

// Skill Definition
{
  name: "calculateShipping",
  description: "Calculate shipping cost based on weight and destination",
  parameters: {
    weight_kg: "number",
    destination_country: "string",
    express: "boolean (optional)"
  },
  execute: async (params) => {
    const { weight_kg, destination_country, express = false } = params
 
    // Base rate
    let cost = weight_kg * 5
 
    // International surcharge
    if (destination_country !== "US") {
      cost += 15
    }
 
    // Express shipping
    if (express) {
      cost *= 1.5
    }
 
    return {
      cost: cost.toFixed(2),
      currency: "USD",
      estimated_days: express ? "2-3" : "5-7"
    }
  }
}

Usage:

User: "How much to ship 5kg to Canada?"
Bot: [Calls calculateShipping(5, "Canada", false)]
"Shipping 5kg to Canada will cost $40.00 USD and take 5-7 days."

Example: Get Stock Price

{
  name: "getStockPrice",
  description: "Get current stock price",
  parameters: {
    symbol: "string"
  },
  execute: async ({ symbol }) => {
    const response = await fetch(
      `https://api.example.com/stock/${symbol}`
    )
    const data = await response.json()
 
    return {
      symbol: data.symbol,
      price: data.price,
      change: data.change,
      percent_change: data.percent_change
    }
  }
}

API Integration Skill

Connect to external APIs:

Example: Weather API

{
  name: "getWeather",
  description: "Get current weather for a location",
  parameters: {
    location: "string",
    units: "metric | imperial (optional)"
  },
  execute: async ({ location, units = "metric" }) => {
    const apiKey = process.env.WEATHER_API_KEY
    const response = await fetch(
      `https://api.openweathermap.org/data/2.5/weather?q=${location}&units=${units}&appid=${apiKey}`
    )
    const data = await response.json()
 
    return {
      temperature: data.main.temp,
      condition: data.weather[0].description,
      humidity: data.main.humidity,
      wind_speed: data.wind.speed
    }
  }
}

Example: CRM Integration

{
  name: "searchCustomer",
  description: "Search for customer in CRM",
  parameters: {
    email: "string (optional)",
    phone: "string (optional)",
    customer_id: "string (optional)"
  },
  execute: async (params) => {
    const response = await fetch(
      `https://your-crm.com/api/customers/search`,
      {
        method: 'POST',
        headers: {
          'Authorization': `Bearer ${process.env.CRM_API_KEY}`,
          'Content-Type': 'application/json'
        },
        body: JSON.stringify(params)
      }
    )
    const customer = await response.json()
 
    return {
      name: customer.name,
      email: customer.email,
      status: customer.status,
      total_orders: customer.orders.length,
      lifetime_value: customer.lifetime_value
    }
  }
}

Skill Configuration

Access Control

Control who can use skills:

Public skills:

  • Available to all users
  • No authentication required
  • Rate limited

Authenticated skills:

  • Requires user login
  • Access based on user role
  • User-specific data

Admin skills:

  • Admin users only
  • Sensitive operations
  • Audit logged

Rate Limiting

Prevent abuse:

{
  name: "sendEmail",
  rate_limit: {
    calls_per_minute: 5,
    calls_per_hour: 20,
    calls_per_day: 50
  }
}

Caching

Cache results for performance:

{
  name: "getProductInfo",
  cache: {
    enabled: true,
    ttl_seconds: 3600, // Cache for 1 hour
    key_params: ["product_id"] // Cache per product
  }
}

Error Handling

Graceful error handling:

execute: async (params) => {
  try {
    // Skill logic
    return result
  } catch (error) {
    return {
      error: true,
      message: "Failed to fetch data. Please try again.",
      user_friendly: true
    }
  }
}

Testing Skills

Test Console

  1. Skills tab → Select skill → Test
  2. Input parameters: 
    {
  "weight_kg": 5,
  "destination_country": "Canada"
}
  3. Click Run Test
  4. View result: 
    {
  "cost": "40.00",
  "currency": "USD",
  "estimated_days": "5-7"
}

Integration Testing

Test in chat interface:

1. Create test chatbot
2. Enable skill
3. Try sample queries
4. Verify responses
5. Check error cases

Skill Marketplace (Coming Soon)

Browse and install community skills:

Categories:

  • 📧 Communication (Email, SMS)
  • 📊 Data & Analytics
  • 💰 E-commerce
  • 🎓 Education
  • 🏥 Healthcare
  • 🏢 Business

Example skills:

  • Stripe payment processing
  • Shopify order lookup
  • Zendesk ticket creation
  • Google Sheets integration
  • HubSpot CRM sync

Advanced: Multi-Step Skills

Execute multiple steps:

{
  name: "bookAppointment",
  description: "Book an appointment with availability check",
  steps: [
    {
      name: "checkAvailability",
      execute: async ({ date, time }) => {
        // Check calendar
        const available = await checkCalendar(date, time)
        return { available }
      }
    },
    {
      name: "createAppointment",
      condition: (prev) => prev.available,
      execute: async ({ date, time, user_email }) => {
        // Create appointment
        const appointment = await createEvent(date, time, user_email)
        return { appointment_id: appointment.id }
      }
    },
    {
      name: "sendConfirmation",
      execute: async ({ appointment_id, user_email }) => {
        // Send confirmation email
        await sendEmail(user_email, "Appointment Confirmed", ...)
        return { confirmed: true }
      }
    }
  ]
}

Best Practices

Design

✅ Do:

  • Clear, descriptive names
  • Detailed descriptions
  • Validate input parameters
  • Return structured data
  • Handle errors gracefully

❌ Don't:

  • Vague skill names
  • Missing parameter descriptions
  • Assume valid input
  • Return raw API responses
  • Expose error stack traces

Security

✅ Do:

  • Use environment variables for API keys
  • Validate all inputs
  • Rate limit sensitive operations
  • Log skill usage
  • Sanitize outputs

❌ Don't:

  • Hardcode credentials
  • Trust user input
  • Allow unlimited API calls
  • Expose sensitive data
  • Skip authentication

Performance

✅ Do:

  • Cache when possible
  • Set reasonable timeouts
  • Batch operations
  • Use async/await
  • Monitor execution time

❌ Don't:

  • Make unnecessary API calls
  • Block on slow operations
  • Retry failed requests indefinitely
  • Load large datasets

Troubleshooting

Skill Not Being Called

Check:

  1. Skill is enabled
  2. Description is clear
  3. Parameters are well-defined
  4. AI model supports function calling
  5. No syntax errors in skill code

Skill Returns Error

Debug:

  1. Check skill logs
  2. Verify API credentials
  3. Test with skill console
  4. Check parameter types
  5. Validate API endpoint

Slow Response Times

Optimize:

  1. Enable caching
  2. Reduce API calls
  3. Batch requests
  4. Use faster endpoints
  5. Implement timeouts

Next Steps

IntegrationsAnalytics