2_add_tools_tutorial.md

LangGraph Tutorial Summary: Add Tools to Chatbot

Purpose

• Extend the chatbot with external tools like web search.
• Enables the agent to answer questions beyond its training data.

Setup

• pip install -U langchain-tavily
• Set your API key: _set_env("TAVILY_API_KEY")

Tool Definition

• Define the tool:

  from langchain_tavily import TavilySearch
  tool = TavilySearch(max_results=2)
  tools = [tool]
  

• Optional test call: tool.invoke("What's a 'node' in LangGraph?")

Bind Tools to LLM

• Required so the LLM knows how to call tools:

  llm_with_tools = llm.bind_tools(tools)
  

Chatbot Node

• Updated to use the tool-aware LLM:

  def chatbot(state: State):
      return {"messages": [llm_with_tools.invoke(state["messages\])]}
  
  graph_builder.add_node("chatbot", chatbot)
  

Tool Execution Node

• Use prebuilt:

  from langgraph.prebuilt import ToolNode
  tool_node = ToolNode(tools=[tool])
  graph_builder.add_node("tools", tool_node)
  

Conditional Routing

• Add dynamic routing between chatbot tools:

  from langgraph.prebuilt import tools_condition
  graph_builder.add_conditional_edges("chatbot", tools_condition)
  graph_builder.add_edge("tools", "chatbot")
  graph_builder.add_edge(START, "chatbot")
  

Graph Behavior

• If LLM response includes tool_calls, graph routes to tools node.
• Tool executes and appends result as new message.
• Loop returns to chatbot to process result or continue conversation.

Compile

• Compile final graph:

  graph = graph_builder.compile()
  

Stream Execution

• Run as before using:

  graph.stream({"messages": [{"role": "user", "content": user_input}]})