⛓️ Langflow Integration
Langflow (GitHub) is a UI for LangChain, designed with react-flow to provide an effortless way to experiment and prototype flows.
With the native integration (since langflow v1.0.17), you can use Langflow to quickly create complex LLM applications in no-code and then use Langfuse to monitor and improve them.
Integration
Obtain Langfuse API keys
- Create account and project on cloud.langfuse.com
- Copy API keys for your project
Setup Langflow
# API keys from project settings in Langfuse
export LANGFLOW_LANGFUSE_SECRET_KEY=secret_key
export LANGFLOW_LANGFUSE_PUBLIC_KEY=public_key
# Optionally, set the host, defaults to Langfuse Cloud
# LANGFLOW_LANGFUSE_HOST=http://localhost:3000
# Install Langflow
pip install langflow
# Run Langflow
langflow runAlternatively, you can run the Langflow CLI command with the environment variables set:
LANGFLOW_LANGFUSE_SECRET_KEY=secret_key LANGFLOW_LANGFUSE_PUBLIC_KEY=public_key langflowMonitoring in Langfuse
Now, when you use Langflow’s chat or API, you can view the trace of your conversations in Langfuse.
Running Langfuse and Langflow with Docker Compose
If you prefer to self-host Langfuse, you can run both services using Docker Compose. By combining the two docker-compose files, you can streamline the networking between them.
version: "3.5"
services:
# Adapted from https://github.com/logspace-ai/langflow/blob/dev/docker_example/docker-compose.yml
langflow:
build:
context: .
dockerfile: Dockerfile
ports:
- "7860:7860"
environment:
+ # Tokens are to be created in Langfuse, then copy-pasted here. Then restart docker-compose.
+ - LANGFLOW_LANGFUSE_SECRET_KEY=sk-lf-...
+ - LANGFLOW_LANGFUSE_PUBLIC_KEY=pk-lf-...
+ - LANGFLOW_LANGFUSE_HOST=http://langfuse-server:3000
command: langflow run --host 0.0.0.0
# https://github.com/langfuse/langfuse/blob/main/docker-compose.yml
langfuse-server:
image: ghcr.io/langfuse/langfuse:latest
depends_on:
- db
ports:
- "3000:3000"
environment:
- NODE_ENV=production
- DATABASE_URL=postgresql://postgres:postgres@db:5432/postgres
- NEXTAUTH_SECRET=mysecret
- SALT=mysalt
- NEXTAUTH_URL=http:localhost:3000
db:
image: postgres
restart: always
environment:
- POSTGRES_USER=postgres
- POSTGRES_PASSWORD=postgres
- POSTGRES_DB=postgres
ports:
- 5432:5432
volumes:
- database_data:/var/lib/postgresql/data
volumes:
database_data:
driver: localTo test the connectivity between Langflow and Langfuse, run the following command:
docker compose exec langflow python -c "import requests, os; addr = os.environ.get('LANGFLOW_LANGFUSE_HOST'); print(addr); res = requests.get(addr, timeout=5); print(res.status_code)"
# which should output the following:
# http://langfuse-server:3000
# 200