Chapter 3: Agents and Tool Use

What are LLM Agents?

LLMs that can plan, use tools, and take actions autonomously.

ReAct Pattern

Reasoning + Acting:

Thought: I need to find the current weather
Action: search("weather in Paris")
Observation: It's 15°C and sunny
Thought: Now I can answer the user
Answer: The weather in Paris is 15°C and sunny

Tool Use Implementation

tools = [
    {
        "name": "web_search",
        "description": "Search the web",
        "parameters": {"query": "string"}
    },
    {
        "name": "calculator",
        "description": "Perform calculations",
        "parameters": {"expression": "string"}
    }
]

def agent_loop(query, tools, max_iterations=5):
    messages = [{"role": "user", "content": query}]
    
    for _ in range(max_iterations):
        response = llm.generate(messages, tools=tools)
        
        if response.tool_calls:
            for call in response.tool_calls:
                result = execute_tool(call)
                messages.append({"role": "tool", "content": result})
        else:
            return response.content

Agent Frameworks

Framework	Features
LangChain	Chains, agents, memory
AutoGPT	Autonomous goal pursuit
CrewAI	Multi-agent collaboration

Safety Considerations

• Limit tool permissions
• Human approval for actions
• Sandboxed execution
• Clear stopping conditions

---

← Chapter 2Quiz →