Partial prompt templates in LangChain

Introduction

Partial prompt templates let you reuse parts of prompts easily. They help you build bigger prompts step-by-step without repeating yourself.

When you want to create a prompt with some fixed parts and some parts that change later.

When you build complex prompts from smaller pieces to keep things organized.

When you want to reuse a common prompt section in different prompts.

When you want to update only part of a prompt without rewriting the whole thing.

Syntax

LangChain

from langchain.prompts import PromptTemplate

partial = PromptTemplate(input_variables=["name"], template="Hello, {name}!")

The input_variables list defines which parts you will fill later.

The template string contains placeholders wrapped in curly braces like {name}.

Examples

This partial prompt waits for the city name to be filled later.

LangChain

from langchain.prompts import PromptTemplate

partial = PromptTemplate(input_variables=["city"], template="Welcome to {city}.")

This partial prompt can be reused with different foods.

LangChain

from langchain.prompts import PromptTemplate

partial = PromptTemplate(input_variables=["food"], template="I love eating {food}.")

Sample Program

This program creates a partial prompt that says hello to a name. Then it fills the name with "Alice" and prints the full prompt.

LangChain

from langchain.prompts import PromptTemplate

# Create a partial prompt with one variable
partial = PromptTemplate(input_variables=["name"], template="Hello, {name}!")

# Fill the variable to get the full prompt
full_prompt = partial.format(name="Alice")

print(full_prompt)

Important Notes

You can combine multiple partial prompts to build bigger prompts.

Partial prompts help keep your code clean and avoid repeating the same text.

Always list all variables you want to fill later in input_variables.

Summary

Partial prompt templates let you create reusable prompt pieces.

They keep your prompts organized and easy to update.

Fill variables later to get the full prompt text.