Ragflow Observability & Tracing

Ragflow (GitHub) is a no-code / low-code framework to design and run Retrieval-Augmented-Generation (RAG) pipelines.

With the native Langfuse integration you can build complex RAG pipelines in Ragflow and then use Langfuse to monitor, debug and improve them.

The integration works for all execution modes of Ragflow: interactive runs in the UI, scheduled jobs and API calls.

Integration

Get Langfuse API keys

Langfuse cloud

local or self-hosted

Create an account and project on cloud.langfuse.com
Copy the public and secret API keys for your project

Follow the self-hosting guide to run Langfuse locally or on-prem
Copy the public and secret API keys for your project

Configure Ragflow

You can configure Langfuse in the Ragflow API settings (scroll to bottom). Please note that you first have to configure an LLM in the Ragflow settings.

Run your pipeline and inspect traces

Execute a pipeline from the Ragflow UI or trigger it via the API. Ragflow will automatically send structured trace data to Langfuse.

Example Ragflow trace in
Langfuse

(Example trace in Langfuse)

Version requirements

Library	Minimum version
Ragflow	`>= 0.3.0`

If you are on an older Ragflow version please upgrade to benefit from the Langfuse integration.