How to Use ChatPromptTemplate in Langchain: Syntax and Example

Use ChatPromptTemplate in Langchain to create structured chat prompts by defining a list of message templates with variables. Instantiate it with from_messages passing message prompt templates, then call format_prompt with variables to generate the final prompt.

Syntax

The ChatPromptTemplate is created using the from_messages method, which takes a list of message prompt templates like SystemMessagePromptTemplate and HumanMessagePromptTemplate. Each message template can have variables that you fill later. Use format_prompt with a dictionary of variables to get the full prompt ready for the chat model.

python

from langchain.prompts.chat import (
    ChatPromptTemplate,
    SystemMessagePromptTemplate,
    HumanMessagePromptTemplate
)

# Define system and human message templates with variables
system_template = SystemMessagePromptTemplate.from_template("You are a helpful assistant.")
human_template = HumanMessagePromptTemplate.from_template("Translate this text to {language}: {text}")

# Create ChatPromptTemplate from messages
chat_prompt = ChatPromptTemplate.from_messages([system_template, human_template])

# Format prompt with variables
final_prompt = chat_prompt.format_prompt(language="French", text="Hello, how are you?")

Example

This example shows how to create a chat prompt that instructs the assistant to translate text into a specified language. It uses ChatPromptTemplate with system and human messages, then formats the prompt with actual values.

python

from langchain.prompts.chat import (
    ChatPromptTemplate,
    SystemMessagePromptTemplate,
    HumanMessagePromptTemplate
)

# Define the system message template
system_msg = SystemMessagePromptTemplate.from_template("You are a helpful assistant.")

# Define the human message template with variables
human_msg = HumanMessagePromptTemplate.from_template("Translate this text to {language}: {text}")

# Create the chat prompt template
chat_prompt = ChatPromptTemplate.from_messages([system_msg, human_msg])

# Format the prompt with actual values
formatted_prompt = chat_prompt.format_prompt(language="Spanish", text="Good morning")

# Print the formatted messages
for message in formatted_prompt.messages:
    print(f"{message.type}: {message.content}")

Output

system: You are a helpful assistant.
human: Translate this text to Spanish: Good morning

Common Pitfalls

• Not using from_messages to create ChatPromptTemplate causes errors because direct instantiation is not supported.
• Forgetting to include all required variables in format_prompt leads to missing value errors.
• Mixing message types incorrectly or not using message prompt templates like SystemMessagePromptTemplate and HumanMessagePromptTemplate can cause unexpected prompt structure.

python

from langchain.prompts.chat import ChatPromptTemplate

# Wrong: Direct instantiation (will error)
try:
    chat_prompt = ChatPromptTemplate(messages=["Hello {name}"])
except Exception as e:
    print(f"Error: {e}")

# Right: Use from_messages with proper message templates
from langchain.prompts.chat import HumanMessagePromptTemplate
human_msg = HumanMessagePromptTemplate.from_template("Hello {name}")
chat_prompt = ChatPromptTemplate.from_messages([human_msg])
formatted = chat_prompt.format_prompt(name="Alice")
print(formatted.messages[0].content)

Output

Error: ChatPromptTemplate() takes no arguments
Hello Alice

Quick Reference

ChatPromptTemplate Quick Tips:

• Use from_messages to create the template from message prompt templates.
• Message templates include SystemMessagePromptTemplate, HumanMessagePromptTemplate, and others.
• Call format_prompt with variables to get the final prompt object.
• The final prompt's messages list contains typed message objects ready for chat models.

Key Takeaways

Create ChatPromptTemplate using from_messages with message prompt templates.

Use format_prompt with all required variables to generate the final prompt.

Always use SystemMessagePromptTemplate and HumanMessagePromptTemplate for proper message types.

Avoid direct instantiation of ChatPromptTemplate; it requires from_messages method.

The formatted prompt contains a list of messages ready to send to chat models.