Lesson 4: Adding the Wikipedia Search Skill 🔗 ↑ TOC

Now we'll give Fred the ability to search Wikipedia! The SignalWire SDK's skills system makes this incredibly easy.

Table of Contents 🔗 ↑ TOC

  1. Understanding Skills
  2. Adding the Wikipedia Skill
  3. Configuring the Skill
  4. How Skills Work Internally
  5. Testing Wikipedia Search

Understanding Skills 🔗 ↑ TOC

Skills are modular capabilities that extend your agent's functionality. Think of them as plugins that add specific features.

What is a Skill? 🔗 ↑ TOC

A skill is:

Available Built-in Skills 🔗 ↑ TOC

The SDK includes several pre-built skills:

Adding the Wikipedia Skill 🔗 ↑ TOC

Let's add Wikipedia search to Fred with just one line of code!

Step 1: Basic Skill Addition 🔗 ↑ TOC

Add this to Fred's __init__ method, after the global data:

        # Add the Wikipedia search skill
        self.add_skill("wikipedia_search")

That's it! Fred can now search Wikipedia. But let's customize it for better results.

Step 2: Customized Configuration 🔗 ↑ TOC

Replace the basic addition with this enhanced version:

        # Add the Wikipedia search skill with custom configuration
        self.add_skill("wikipedia_search", {
            "num_results": 2,  # Get up to 2 articles for broader coverage
            "no_results_message": "Oh, I couldn't find anything about '{query}' on Wikipedia. Maybe try different keywords or let me know if you meant something else!",
            "swaig_fields": {
                "fillers": {
                    "en-US": [
                        "Let me look that up on Wikipedia for you...",
                        "Searching Wikipedia for that information...",
                        "One moment, checking Wikipedia...",
                        "Let me find that in the encyclopedia..."
                    ]
                }
            }
        })

Configuring the Skill 🔗 ↑ TOC

Let's understand each configuration option:

Configuration Parameters 🔗 ↑ TOC

num_results (default: 1) - Number of Wikipedia articles to return - More results = broader coverage but longer responses - Fred uses 2 for balance

no_results_message - Custom message when no articles are found - Use {query} as a placeholder for the search term - Should match your agent's personality

swaig_fields - Additional SWAIG function configuration - fillers: Phrases spoken while searching - Enhances conversation flow

Example Configurations 🔗 ↑ TOC

# Minimal configuration
self.add_skill("wikipedia_search")

# Academic assistant
self.add_skill("wikipedia_search", {
    "num_results": 3,
    "no_results_message": "No Wikipedia entries found for '{query}'. Please verify the spelling or try related terms."
})

# Casual helper
self.add_skill("wikipedia_search", {
    "num_results": 1,
    "no_results_message": "Hmm, Wikipedia doesn't have info on '{query}'. Got another topic?",
    "swaig_fields": {
        "fillers": {
            "en-US": ["Let me check...", "Looking that up..."]
        }
    }
})

How Skills Work Internally 🔗 ↑ TOC

Understanding the magic behind add_skill():

What Happens When You Add a Skill 🔗 ↑ TOC

  1. Skill Loading python # The SDK dynamically imports the skill module from signalwire_agents.skills.wikipedia_search.skill import WikipediaSearchSkill

  2. Skill Initialization python # Creates skill instance with your configuration skill = WikipediaSearchSkill(agent=self, params=config)

  3. Tool Registration python # The skill registers its functions with your agent self.define_tool( name="search_wiki", description="Search Wikipedia for information", parameters={"query": {"type": "string", "description": "Search term"}}, handler=skill._search_wiki_handler )

The search_wiki Function 🔗 ↑ TOC

After adding the skill, Fred gains the search_wiki function:

Function Signature:

search_wiki(query: string) -> string

What it does:

  1. Searches Wikipedia for articles matching the query
  2. Retrieves article summaries
  3. Returns formatted results

Example calls (by the AI):

Let's add some test code to verify the skill works:

Step 3: Update Main Function 🔗 ↑ TOC

def main():
    """Run Fred the Wiki Bot"""
    print("=" * 60)
    print("🤖 Fred - The Wikipedia Knowledge Bot")
    print("=" * 60)
    print()
    print("Fred is a friendly assistant who loves searching Wikipedia!")
    print("He can help you learn about almost any topic.")
    print()
    print("Example questions you can ask Fred:")
    print("  • 'Tell me about Albert Einstein'")
    print("  • 'What is quantum physics?'")
    print("  • 'Who was Marie Curie?'")
    print("  • 'Search for information about the solar system'")
    print()

    # Create Fred
    fred = FredTheWikiBot()

    # Show loaded skills
    loaded_skills = fred.list_skills()
    print(f"Fred's capabilities: {', '.join(loaded_skills)}")
    print()

    # The skill automatically adds tools to the agent
    # You can verify this by checking registered functions
    print("Wikipedia search is ready!")
    print("Fred can now search Wikipedia for any topic.")

if __name__ == "__main__":
    main()

Run it:

python fred.py

Expected output:

============================================================
🤖 Fred - The Wikipedia Knowledge Bot
============================================================

Fred is a friendly assistant who loves searching Wikipedia!
He can help you learn about almost any topic.

Example questions you can ask Fred:
  • 'Tell me about Albert Einstein'
  • 'What is quantum physics?'
  • 'Who was Marie Curie?'
  • 'Search for information about the solar system'

Fred's capabilities: wikipedia_search
Wikipedia search is ready!
Fred can now search Wikipedia for any topic.

When a user asks Fred about a topic:

  1. User says: "Tell me about Albert Einstein"

  2. Fred's AI recognizes the request needs Wikipedia

  3. Fred calls: search_wiki("Albert Einstein")

  4. Wikipedia returns article summary

  5. Fred responds with the information conversationally

Example Conversation Flow 🔗 ↑ TOC

User: "Hi there!"
Fred: "Hello! I'm Fred, your friendly Wikipedia assistant! I love helping people learn new things. What would you like to know about today?"

User: "Tell me about black holes"
Fred: [filler] "Let me look that up on Wikipedia for you..."
Fred: "Fascinating topic! According to Wikipedia, a black hole is a region of spacetime where gravity is so strong that nothing—no particles or even electromagnetic radiation such as light—can escape from it..."

User: "That's interesting! Who discovered them?"
Fred: [filler] "Searching Wikipedia for that information..."
Fred: "Great question! The concept of black holes has a rich history..."

Skill Benefits 🔗 ↑ TOC

Why Use Skills vs Custom Implementation? 🔗 ↑ TOC

With Skills:

# One line!
self.add_skill("wikipedia_search")

Without Skills:

# You'd need to:
# 1. Import Wikipedia API library
# 2. Handle API calls
# 3. Parse responses
# 4. Format results
# 5. Handle errors
# 6. Register SWAIG function
# 7. Implement handler
# ... 50+ lines of code

Skills Provide 🔗 ↑ TOC

  1. Tested functionality - Pre-built and debugged
  2. Consistent interface - Standard configuration pattern
  3. Error handling - Graceful failure modes
  4. Documentation - Clear usage instructions
  5. Maintenance - Updates with SDK

Next Steps 🔗 ↑ TOC

Fred can now search Wikipedia! But let's make him more fun by adding a custom function for sharing Wikipedia facts.

➡️ Continue to Lesson 5: Creating Custom Functions

Skill Checklist:

What Fred Can Do Now:

← Previous: Basic Agent | Back to Overview | Next: Custom Functions →