Provides deep insights into multi-turn LLM conversations by analyzing and debugging Langfuse trace sessions.
Langfuse Session Analysis is a specialized Claude Code skill designed for developers monitoring LLM applications. It allows for seamless inspection of multi-trace sessions grouped by session ID, enabling users to debug complex, multi-turn conversations, track session quality via scores, and isolate problematic interactions. By integrating directly into the CLI, it streamlines the observability workflow, making it easy to identify sessions with errors or low-performance metrics across specific timeframes without leaving the terminal.
Key Features
01Visual timeline view for tracking the sequence of session events
02Inspect detailed session metadata and multi-trace hierarchies
03List and filter LLM sessions with customizable limits
04Advanced issue detection for identifying errors and low-scoring interactions
05Automated quality analysis based on session scores and feedback
060 GitHub stars
Use Cases
01Debugging long-running, multi-turn chatbot conversations that resulted in user dissatisfaction
02Proactively monitoring production logs to find and fix sessions with high error rates
03Analyzing performance trends by filtering sessions by specific quality metrics over time
