Key Concepts to Understand:

Prompt Templates:

Used to format user input into a structured format that an LLM can work with.

ChatPromptTemplate:

A class that helps create structured prompts, which can include both system and user messages.

MessagesPlaceholder:

Allows us to dynamically insert messages into the prompt template.

RunnableWithMessageHistory:

A utility that retains and manages chat history for maintaining context between sessions.

Implementing the Prompt Templates

1. Setting Up a Basic ChatPromptTemplate:

```
ChatPromptTemplate
```
allows the separation of different input components (e.g., system, user messages).
With placeholders like
```
MessagesPlaceholder
```
, you can dynamically change what gets passed to the LLM.

from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder
from langchain.schema import HumanMessage

# Create a simple ChatPromptTemplate
prompt = ChatPromptTemplate.from_messages(
    [
        ("system", "You are a helpful assistant. Answer all questions to the best of your ability."),
        MessagesPlaceholder(variable_name="messages"),
    ]
)

# Define the chain by connecting the prompt with a language model (model should be instantiated beforehand)
chain = prompt | model

# Invoke the chain with some input
response = chain.invoke({"messages": [HumanMessage(content="Hi! I'm Bob")]})

print(response.content)

2. Adding Dynamic Inputs to the Prompt Template:

You can add variables like
```
{language}
```
to customize responses based on user preference.

# Modify the prompt to include a dynamic `language` variable
prompt = ChatPromptTemplate.from_messages(
    [
        ("system", "You are a helpful assistant. Answer all questions to the best of your ability in {language}."),
        MessagesPlaceholder(variable_name="messages"),
    ]
)

# Redefine the chain with the updated prompt
chain = prompt | model

# Invoke the chain, specifying the language
response = chain.invoke(
    {"messages": [HumanMessage(content="Hi! I'm Bob")], "language": "Spanish"}
)

print(response.content)  # Output: ¡Hola, Bob! ¿En qué puedo ayudarte hoy?

3. Managing Conversation History:

```
RunnableWithMessageHistory
```
ensures the model retains context across multiple interactions, making it ideal for building conversational agents.

from langchain_core.runnables import RunnableWithMessageHistory

# Initialize RunnableWithMessageHistory to track conversation history
with_message_history = RunnableWithMessageHistory(
    chain,
    get_session_history,  # A function that retrieves session history
    input_messages_key="messages",
)

# Define some configurations (e.g., session ID for the chat)
config = {"configurable": {"session_id": "abc11"}}

# Pass in inputs along with config to maintain session
response = with_message_history.invoke(
    {"messages": [HumanMessage(content="Hi! I'm Todd")], "language": "Spanish"},
    config=config,
)

print(response.content)  # Output: ¡Hola Todd! ¿En qué puedo ayudarte hoy?

# Model retains context and responds accurately based on session history
response = with_message_history.invoke(
    {"messages": [HumanMessage(content="What's my name?")], "language": "Spanish"},
    config=config,
)

print(response.content)  # Output: Tu nombre es Todd.

Key Concepts to Understand:

Used to format user input into a structured format that an LLM can work with.

A class that helps create structured prompts, which can include both system and user messages.

Allows us to dynamically insert messages into the prompt template.

A utility that retains and manages chat history for maintaining context between sessions.

1. Setting Up a Basic ChatPromptTemplate:

ChatPromptTemplate

allows the separation of different input components (e.g., system, user messages).

With placeholders like

MessagesPlaceholder

, you can dynamically change what gets passed to the LLM.

from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder
from langchain.schema import HumanMessage

# Create a simple ChatPromptTemplate
prompt = ChatPromptTemplate.from_messages(
    [
        ("system", "You are a helpful assistant. Answer all questions to the best of your ability."),
        MessagesPlaceholder(variable_name="messages"),
    ]
)

# Define the chain by connecting the prompt with a language model (model should be instantiated beforehand)
chain = prompt | model

# Invoke the chain with some input
response = chain.invoke({"messages": [HumanMessage(content="Hi! I'm Bob")]})

print(response.content)

2. Adding Dynamic Inputs to the Prompt Template:

You can add variables like

{language}

to customize responses based on user preference.

# Modify the prompt to include a dynamic `language` variable
prompt = ChatPromptTemplate.from_messages(
    [
        ("system", "You are a helpful assistant. Answer all questions to the best of your ability in {language}."),
        MessagesPlaceholder(variable_name="messages"),
    ]
)

# Redefine the chain with the updated prompt
chain = prompt | model

# Invoke the chain, specifying the language
response = chain.invoke(
    {"messages": [HumanMessage(content="Hi! I'm Bob")], "language": "Spanish"}
)

print(response.content)  # Output: ¡Hola, Bob! ¿En qué puedo ayudarte hoy?

3. Managing Conversation History:

RunnableWithMessageHistory

ensures the model retains context across multiple interactions, making it ideal for building conversational agents.

from langchain_core.runnables import RunnableWithMessageHistory

# Initialize RunnableWithMessageHistory to track conversation history
with_message_history = RunnableWithMessageHistory(
    chain,
    get_session_history,  # A function that retrieves session history
    input_messages_key="messages",
)

# Define some configurations (e.g., session ID for the chat)
config = {"configurable": {"session_id": "abc11"}}

# Pass in inputs along with config to maintain session
response = with_message_history.invoke(
    {"messages": [HumanMessage(content="Hi! I'm Todd")], "language": "Spanish"},
    config=config,
)

print(response.content)  # Output: ¡Hola Todd! ¿En qué puedo ayudarte hoy?

# Model retains context and responds accurately based on session history
response = with_message_history.invoke(
    {"messages": [HumanMessage(content="What's my name?")], "language": "Spanish"},
    config=config,
)

print(response.content)  # Output: Tu nombre es Todd.

Key Concepts to Understand:

Key Concepts to Understand:

Prompt Templates:

ChatPromptTemplate:

MessagesPlaceholder:

RunnableWithMessageHistory:

Implementing the Prompt Templates

1. Setting Up a Basic ChatPromptTemplate:

2. Adding Dynamic Inputs to the Prompt Template:

3. Managing Conversation History:

Related Skills

English Translator and Improver

Bakery Merge Bounty Game Overview

VEO3 Video Generation Assistant System Prompt

Key Concepts to Understand:

Prompt Templates:

ChatPromptTemplate:

MessagesPlaceholder:

RunnableWithMessageHistory:

Implementing the Prompt Templates

1. Setting Up a Basic ChatPromptTemplate:

2. Adding Dynamic Inputs to the Prompt Template:

3. Managing Conversation History: