Module 2.3: Skills System


Duration	2 hours
Day	3 of 5

Learning Objectives

By the end of this module, students will be able to:

Understand the skill architecture
Add and configure built-in skills
Compose multiple skills into agents
Check and manage loaded skills

Topics

1. What Are Skills? (20 min)

Definition

Skills are reusable capability modules that provide:

Functions (tools the AI can call)
Prompts (additional instructions)
Hints (speech recognition improvements)
Global data (shared information)

Skills vs Functions

Functions	Skills
Single capability	Bundle of capabilities
Defined in agent	Packaged separately
Agent-specific	Reusable across agents
No prompts	Can include prompts

Built-in Skills

Skill	Purpose
`datetime`	Current date/time info
`math`	Mathematical calculations
`web_search`	Internet searches
`datasphere`	SignalWire knowledge base
`native_vector_search`	Local document search

2. Adding Skills (25 min)

Basic Usage

from signalwire_agents import AgentBase

agent = AgentBase(name="helpful-agent")

# Add a skill
agent.add_skill("datetime")

That’s it! The agent now knows the current date and time.

Skills with Configuration

# Web search with API key
agent.add_skill("web_search", {
    "api_key": os.getenv("SEARCH_API_KEY")
})

# Vector search with index path
agent.add_skill("native_vector_search", {
    "index_path": "/data/knowledge.swsearch"
})

Multiple Skills

agent.add_skill("datetime")
agent.add_skill("math")
agent.add_skill("web_search", {"api_key": "..."})

Or chain them:

(agent
    .add_skill("datetime")
    .add_skill("math")
    .add_skill("web_search", {"api_key": "..."}))

3. Built-in Skills Deep Dive (30 min)

datetime Skill

Provides current date, time, timezone information.

agent.add_skill("datetime")

Functions Added:

get_current_time - Current time in timezone
get_current_date - Current date

Example Conversation:

Caller: “What time is it?” Agent: “It’s currently 2:30 PM Eastern Time.”

math Skill

Performs mathematical calculations.

agent.add_skill("math")

Functions Added:

calculate - Evaluate math expressions

Example Conversation:

Caller: “What’s 15% of 250?” Agent: “15% of 250 is 37.50.”

web_search Skill

Searches the internet for information.

agent.add_skill("web_search", {
    "api_key": os.getenv("TAVILY_API_KEY")
})

Functions Added:

search_web - Search and summarize results

Example Conversation:

Caller: “What’s the latest news about SignalWire?” Agent: “According to recent articles…”

native_vector_search Skill

Searches local document indexes.

agent.add_skill("native_vector_search", {
    "index_path": "/data/products.swsearch",
    "top_k": 5
})

Functions Added:

search_documents - Semantic search over documents

4. Skill Management (20 min)

Check Loaded Skills

# List all skills
skills = agent.get_skills()
print(skills)  # ['datetime', 'math']

# Check if specific skill is loaded
if agent.has_skill("datetime"):
    print("DateTime skill is active")

Remove Skills

agent.remove_skill("math")

Skill Priority

Skills added later can override earlier ones:

agent.add_skill("basic_search")    # Adds search function
agent.add_skill("advanced_search") # Overrides search function

5. Skill Composition Patterns (25 min)

Support Agent Pattern

class SupportAgent(AgentBase):
    def __init__(self):
        super().__init__(name="support")

        self.prompt_add_section(
            "Role",
            "You are a technical support agent."
        )

        # Knowledge skills
        self.add_skill("native_vector_search", {
            "index_path": "/data/support_docs.swsearch"
        })

        # Utility skills
        self.add_skill("datetime")

        # Custom functions for actions
        self._add_custom_functions()

Sales Agent Pattern

class SalesAgent(AgentBase):
    def __init__(self):
        super().__init__(name="sales")

        # Product knowledge
        self.add_skill("native_vector_search", {
            "index_path": "/data/products.swsearch"
        })

        # Pricing calculations
        self.add_skill("math")

        # Current promotions (time-sensitive)
        self.add_skill("datetime")

Research Agent Pattern

class ResearchAgent(AgentBase):
    def __init__(self):
        super().__init__(name="research")

        # Internal docs
        self.add_skill("native_vector_search", {
            "index_path": "/data/internal.swsearch"
        })

        # External research
        self.add_skill("web_search", {
            "api_key": os.getenv("SEARCH_API_KEY")
        })

        # Calculations
        self.add_skill("math")

Complete Example

Click to reveal complete solution

#!/usr/bin/env python3
"""Multi-skill support agent."""

import os
from signalwire_agents import AgentBase, SwaigFunctionResult


class SmartSupportAgent(AgentBase):
    def __init__(self):
        super().__init__(name="smart-support")

        self._configure_prompts()
        self._configure_voice()
        self._add_skills()
        self._add_custom_functions()

    def _configure_prompts(self):
        self.prompt_add_section(
            "Role",
            "You are an intelligent support agent for TechCorp. "
            "Use your skills to help customers effectively."
        )

        self.prompt_add_section(
            "Capabilities",
            bullets=[
                "Search our knowledge base for product information",
                "Calculate prices and discounts",
                "Tell customers the current date and time",
                "Create support tickets for complex issues"
            ]
        )

    def _configure_voice(self):
        self.add_language("English", "en-US", "rime.spore")

    def _add_skills(self):
        # Date/time awareness
        self.add_skill("datetime")

        # Price calculations
        self.add_skill("math")

        # Knowledge base search
        if os.path.exists("/data/support.swsearch"):
            self.add_skill("native_vector_search", {
                "index_path": "/data/support.swsearch",
                "top_k": 3
            })

    def _add_custom_functions(self):
        @self.tool(description="Create a support ticket")
        def create_ticket(
            issue: str,
            priority: str = "normal"
        ) -> SwaigFunctionResult:
            ticket_id = f"TKT-{hash(issue) % 10000:04d}"
            return SwaigFunctionResult(
                f"Created ticket {ticket_id} with {priority} priority."
            )


if __name__ == "__main__":
    agent = SmartSupportAgent()
    agent.run()

Key Takeaways

Skills bundle capabilities - Functions, prompts, hints together
Built-in skills are ready to use - datetime, math, search
Configuration via parameters - API keys, paths, options
Compose for power - Multiple skills work together
Manage what’s loaded - Check, add, remove as needed

Preparation for Lab 2.3

Review available built-in skills
Think about what capabilities your agent needs
Have API keys ready if using web_search

Lab Preview

In Lab 2.3, you will:

Add datetime and math skills
Configure web_search skill
Combine skills with custom functions
Test skill interactions