How to beginner · 3 min read

How to use Langfuse prompt templates

Q: How to use Langfuse prompt templates

Use Langfuse prompt templates by defining reusable prompt strings with placeholders, then render them with variables before sending to your LLM client. This enables consistent prompt formatting and easy tracking with langfuse decorators or manual logging.

Quick answer

Use Langfuse prompt templates by defining reusable prompt strings with placeholders, then render them with variables before sending to your LLM client. This enables consistent prompt formatting and easy tracking with langfuse decorators or manual logging.

PREREQUISITES

Python 3.8+
pip install langfuse
OpenAI API key or compatible LLM API key
Basic familiarity with Python string formatting

Setup

Install the langfuse Python package and set your API keys as environment variables. Import the necessary classes to create prompt templates and initialize the Langfuse client.

bash

pip install langfuse

Step by step

Create a prompt template using Python's str.format style placeholders or f-string style. Use langfuse to observe the function that renders and sends the prompt to your LLM client. This example uses OpenAI's gpt-4o-mini model.

python

import os
from langfuse import Langfuse
from langfuse.decorators import observe
from openai import OpenAI

# Initialize Langfuse client
langfuse = Langfuse(
    public_key=os.environ["LANGFUSE_PUBLIC_KEY"],
    secret_key=os.environ["LANGFUSE_SECRET_KEY"],
    host="https://cloud.langfuse.com"
)

# Initialize OpenAI client
client = OpenAI(api_key=os.environ["OPENAI_API_KEY"])

# Define a prompt template with placeholders
PROMPT_TEMPLATE = "Write a short summary about {topic} in {language}."

@observe()
def generate_summary(topic: str, language: str) -> str:
    # Render the prompt template
    prompt = PROMPT_TEMPLATE.format(topic=topic, language=language)
    
    # Call the LLM
    response = client.chat.completions.create(
        model="gpt-4o-mini",
        messages=[{"role": "user", "content": prompt}]
    )
    
    # Return the generated text
    return response.choices[0].message.content

if __name__ == "__main__":
    summary = generate_summary("Langfuse", "English")
    print(summary)

output

Langfuse is a powerful tool for tracking and managing AI prompts, enabling developers to standardize prompt usage and improve observability.

Common variations

Use f-strings or other templating libraries like jinja2 for more complex prompt templates.
Integrate with other LLM providers by replacing the OpenAI client with any compatible client.
Use async functions with @observe() for asynchronous prompt calls.
Manually log prompts and responses with langfuse.log() if you prefer not to use decorators.

python

import asyncio
from langfuse.decorators import observe
from openai import OpenAI

client = OpenAI(api_key=os.environ["OPENAI_API_KEY"])

@observe()
async def generate_async_summary(topic: str, language: str) -> str:
    prompt = f"Write a short summary about {topic} in {language}."
    response = await client.chat.completions.create(
        model="gpt-4o-mini",
        messages=[{"role": "user", "content": prompt}]
    )
    return response.choices[0].message.content

async def main():
    summary = await generate_async_summary("Langfuse", "English")
    print(summary)

if __name__ == "__main__":
    asyncio.run(main())

output

Langfuse is a powerful tool for tracking and managing AI prompts, enabling developers to standardize prompt usage and improve observability.

Troubleshooting

If prompts are not tracked in the Langfuse dashboard, verify your LANGFUSE_PUBLIC_KEY and LANGFUSE_SECRET_KEY environment variables are set correctly.
Ensure the @observe() decorator wraps the function that generates and sends the prompt.
For template rendering errors, check that all placeholders in the prompt template have corresponding variables passed.
If you see API errors from your LLM client, confirm your API key and model name are correct and have sufficient quota.

✅

Key Takeaways

Define prompt templates as reusable strings with placeholders for consistent prompt formatting.
Use the @observe() decorator from langfuse to automatically track prompt usage and responses.
Render templates with variables before sending to your LLM client for clean, maintainable code.

Verified 2026-04 · gpt-4o-mini

Verify ↗