Enable Langfuse for an Agent built with kAgent

[gururnag-te]

Describe your question

I have a chat UI agent built using https://github.com/kagent-dev/kagent.

I would like to get Human feedback and then evaluate the user sentiment with langfuse. Is this possible?

Langfuse Cloud or Self-Hosted?

Self-Hosted

If Self-Hosted

3.146.0 OSS

If Langfuse Cloud

No response

SDK and integration versions

No response

Pre-Submission Checklist

• I have checked for existing issues/discussions and consulted Langfuse AI.

[inkeep-ai-assistant[bot]]

Based on the available sources, I cannot find specific documentation about integrating Langfuse with kAgent (kagent-dev/kagent). The sources provided focus on LangGraph agents and general Langfuse evaluation features, but don't cover kAgent specifically(1)(2).

However, regarding your specific goals:

Human Feedback Collection:
Langfuse supports capturing user feedback through scores. You can record feedback using the Langfuse SDK by creating scores on traces(2). The typical approach involves:

• Capturing feedback as a numeric score (e.g., thumbs up/down as 1/-1)
• Attaching it to a specific trace using the trace ID
• Optionally including comments with the feedback(2)

Sentiment Evaluation:
For evaluating user sentiment, you can use Langfuse's LLM-as-a-Judge feature to automatically score traces based on criteria you define(2). This would allow you to set up automated evaluation of user interactions.

Important Note for Self-Hosted (v3.146.0 OSS):
Be aware that Annotation Queues, which facilitate human-in-the-loop evaluation workflows, require a commercial license when self-hosting Langfuse and are not included in the OSS version(3).

To integrate kAgent with Langfuse, you would likely need to manually instrument your kAgent application using Langfuse's Python SDK to create traces and capture the necessary data. Without kAgent-specific integration documentation, you may need to explore custom instrumentation approaches or reach out to Langfuse support for guidance on this specific framework.

Would you like guidance on general Langfuse instrumentation approaches that might apply to your kAgent setup?

📚 Sources:

1. Troubleshooting and FAQ
2. Evaluate LangGraph Agents
3. bug: Annotation Queues supported in Self-Hosted

---

Have another question? Just tag @inkeep.

[jannikmaierhoefer]

Hi @gururnag-te, inkeep is correct. You can ingest the user feedback via Lanfuse Scores: https://langfuse.com/docs/evaluation/evaluation-methods/scores-via-sdk