Criando ferramentas personalizadas do LangChain
Desenvolva e integre suas próprias ferramentas especializadas para ampliar a funcionalidade dos seus agentes além das opções prontas.
Criando ferramentas personalizadas do LangChain é uma aula grátis de AI Agents with LangChain & Autonomous Workflows no CoddyKit. Esta é a aula 1 de 4. Você pode ler a aula completa abaixo gratuitamente — depois pratica ao vivo no navegador com um editor de código integrado e um tutor de IA 24/7. Faz parte do caminho de aprendizado de AI Agents with LangChain & Autonomous Workflows, e seu progresso é sincronizado entre a web e o app CoddyKit. O curso de AI Agents with LangChain & Autonomous Workflows inclui 4 aulas no total.
Partes desta aula ainda não foram traduzidas e aparecem em inglês.
Extend Agent Capabilities
AI agents are powerful, but sometimes they need to perform very specific actions that aren't covered by standard tools. This is where custom tools come in!
Custom tools allow your agent to interact with unique APIs, proprietary databases, or perform highly specialized calculations tailored to your application.
Tools: Agent's Action Kit
In LangChain, a Tool is essentially a function that an agent can call to perform a specific action. Think of it as an item in the agent's utility belt.
- It takes a single string input (e.g., a query, a number).
- It returns a single string output (e.g., a result, an error message).
- The agent uses the tool's description to decide when and how to use it.
Anatomy of a Custom Tool
To build a custom tool, you primarily need two things:
- A standard Python function that contains the logic your tool will execute.
- A way to describe this function (its name, what it does, and what kind of input it expects) to the Large Language Model (LLM).
LangChain provides an easy way to define and expose these functions to your agents.
Crafting the Python Function
First, let's create a simple Python function. This function will be the core logic of our custom tool. Remember, it should ideally take a string and return a string.
Here's an example of a function that calculates the square of a number:
def calculate_square(number_str: str) -> str:
"""Calculates the square of a number."""
try:
number = int(number_str)
return str(number * number)
except ValueError:
return "Error: Input must be a valid integer."Making Tools Agent-Ready
LangChain provides the @tool decorator to easily turn a regular Python function into an agent-callable tool. This decorator automatically infers the tool's schema (inputs, description) for the LLM.
The docstring of your function becomes the tool's description, which is crucial for the LLM to understand its purpose.
from langchain_core.tools import tool
@tool
def calculate_square(number_str: str) -> str:
"""Calculates the square of a number.
Input should be a string representing an integer."""
try:
number = int(number_str)
return str(number * number)
except ValueError:
return "Error: Input must be a valid integer."
# You can test it directly:
# print(calculate_square("7"))
# print(calculate_square("hello"))Equipping Your Agent
Once your custom tool function is defined with the @tool decorator, you need to provide it to your agent. Agents are typically initialized with a list of tools they have access to.
The LLM will then "see" these tools and their descriptions, allowing it to decide when to call them based on the user's prompt.
Agent Setup for Tool Use
Before running our agent with the custom tool, we need a few more pieces:
- An LLM (like OpenAI's GPT models).
- A Prompt Template to guide the LLM's responses.
- The
create_tool_calling_agentfunction to combine the LLM, tools, and prompt. - An Agent Executor to run the agent.
Agent in Action: Custom Square Tool
Let's put it all together! This runnable example shows an agent using our calculate_square custom tool to answer a user's request. Observe the agent's thought process in the output.
from langchain_openai import ChatOpenAI
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.tools import tool
from langchain.agents import AgentExecutor, create_tool_calling_agent
# IMPORTANT: Set your OpenAI API key in your environment variables
# e.g., export OPENAI_API_KEY='your_key_here'
@tool
def calculate_square(number_str: str) -> str:
"""Calculates the square of a number.
Input should be a string representing an integer."""
try:
number = int(number_str)
return str(number * number)
except ValueError:
return "Error: Input must be a valid integer."
# Define the LLM (replace with your setup if not using OpenAI)
llm = ChatOpenAI(model="gpt-3.5-turbo", temperature=0)
# Define the tools the agent can use
tools = [calculate_square]
# Create the prompt template
prompt = ChatPromptTemplate.from_messages([
("system", "You are a helpful assistant. Use the tools provided to answer questions."),
("human", "{input}"),
("placeholder", "{agent_scratchpad}")
])
# Create the agent
agent = create_tool_calling_agent(llm, tools, prompt)
# Create an agent executor with verbosity to see the steps
agent_executor = AgentExecutor(agent=agent, tools=tools, verbose=True)
# Invoke the agent with a question it needs the tool for
result = agent_executor.invoke({"input": "What is the square of 15?"})
print(f"\nAgent's final answer: {result['output']}")Tips for Effective Tools
When designing your custom tools, consider these best practices:
- Clear Descriptions: The tool's docstring is vital. Make it precise so the LLM knows when to use it.
- Robust Error Handling: Build try-except blocks into your tool functions to handle unexpected inputs or external service failures.
- Single Responsibility: Each tool should ideally do one thing well. Avoid making overly complex tools.
- String I/O: Remember, tools typically expect string inputs and return string outputs. Convert types as needed.
Custom Tool Check
Which of the following is the primary purpose of the @tool decorator in LangChain when creating a custom tool?
Custom Tool Power-Up
You've learned how to create and integrate your own custom tools into LangChain agents!
- We saw why custom tools are important for extending agent capabilities.
- We understood the core components: a Python function and its description.
- We used the
@tooldecorator to make functions agent-ready. - Finally, we built a full agent that successfully used our custom tool.
This skill is fundamental for building truly versatile and domain-specific AI agents.
Aprenda AI Agents with LangChain & Autonomous Workflows com um tutor de IA — grátis
Escreva e execute código real no seu navegador, obtenha ajuda instantânea de um tutor de IA 24/7 e continue de onde parou na web ou no app.
- Cursos
- 12
- Aulas
- 50
Perguntas Frequentes
A aula “Criando ferramentas personalizadas do LangChain” é grátis?
Sim — o texto completo de “Criando ferramentas personalizadas do LangChain” é grátis para ler aqui na web. Para praticá-la interativamente (um editor de código integrado e um tutor de IA 24/7) e desbloquear o restante do curso de AI Agents with LangChain & Autonomous Workflows, atualize para CoddyKit PRO. O curso de AI Agents with LangChain & Autonomous Workflows inclui 4 aulas no total.
O que vou aprender em “Criando ferramentas personalizadas do LangChain”?
Desenvolva e integre suas próprias ferramentas especializadas para ampliar a funcionalidade dos seus agentes além das opções prontas. Você pratica AI Agents with LangChain & Autonomous Workflows com código prático que executa diretamente no navegador, e um tutor de IA 24/7 responde suas dúvidas enquanto trabalha na aula.
Preciso ter experiência prévia para começar AI Agents with LangChain & Autonomous Workflows?
Nenhuma experiência prévia é necessária. AI Agents with LangChain & Autonomous Workflows no CoddyKit é estruturado para alunos iniciantes até avançados, então você pode começar aqui ou desde o início e aprender no seu ritmo. Esta é a aula 1 de 4.
Quanto tempo leva a aula “Criando ferramentas personalizadas do LangChain”?
A maioria das aulas CoddyKit leva cerca de 5–10 minutos. Cada uma é compacta e interativa, então você faz progresso constante e retoma exatamente de onde parou entre web e app.
Posso escrever e executar código nesta aula de AI Agents with LangChain & Autonomous Workflows?
Sim. Cada aula de AI Agents with LangChain & Autonomous Workflows inclui um editor de código integrado, então você escreve e executa código real direto no navegador e recebe feedback de IA instantaneamente — nenhuma configuração local necessária.
Todas as aulas deste curso
- Criando ferramentas personalizadas do LangChain
- Integrando APIs externas
- Extração de dados da web e enriquecimento de dados
- Kits de ferramentas e entradas estruturadas