What metrics are included in the analysis?

The skill aggregates data across traces to show turn counts, total duration, token usage, total cost, and average quality scores.

How do I find sessions with errors?

You can use the find-issues command with the --has-errors flag to quickly identify any sessions that encountered problems over a specified timeframe.

Can I filter sessions by specific users?

Yes, the skill allows you to list sessions filtered by a specific user-id, making it easy to troubleshoot issues reported by specific clients.

What does the Langfuse Session Analysis skill do?

It provides tools to analyze multi-trace user sessions in Langfuse, helping you debug conversation flows, track performance metrics, and understand user journeys.

Langfuse Session Analysis

Name: Langfuse Session Analysis
Author: mberto10

bymberto10

0•

Analytics & Monitoring

Analyzes multi-turn conversation flows and session-level metrics within Langfuse to debug user journeys and track LLM performance.

Langfuse Session Analysis is a specialized observability tool designed for developers to manage and optimize AI interactions tracked via Langfuse. It provides a structured way to dive into multi-trace sessions, allowing users to aggregate metrics such as turn counts, total costs, and durations while generating detailed chronological timelines of user journeys. By enabling the automated discovery of sessions with errors or low quality scores, this skill helps developers pinpoint friction points in complex conversation flows and improve the overall reliability of LLM-powered applications.

Key Features

01Automated discovery of problematic sessions using error and score filters

02Comprehensive session listing with aggregate usage statistics

03Chronological timeline generation for multi-turn conversations

04Deep-link analysis into specific user sessions and trace histories

050 GitHub stars

06In-depth analysis of turn counts, duration, and token costs

Use Cases

01Debugging user complaints by reconstructing and analyzing specific conversation histories

02Monitoring LLM application costs and token usage at the aggregate session level

03Identifying and fixing performance bottlenecks in multi-turn AI interactions

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add mberto10/mberto-compound session-analysis

For use in Claude.ai and ChatGPT

Download Skill