Manages Langfuse datasets and traces to streamline LLM application evaluation and regression testing.
This skill enables seamless integration with Langfuse for managing datasets used in LLM evaluation, experimentation, and regression testing. It allows developers to programmatically create datasets, curate golden sets, and add traces—either individually or in batches—directly within their Claude Code environment. By centralizing dataset management, it facilitates the creation of high-quality benchmarks to monitor model performance and ensure consistent output quality over time.
Key Features
01Individual and batch trace addition to specific datasets
02Curation of golden sets and regression test cases
03Support for expected score assignment to dataset traces
04Dataset item inspection and listing via CLI
050 GitHub stars
06Programmatic creation and management of Langfuse datasets
Use Cases
01Building a regression suite from failing production traces to prevent future bugs
02Creating a 'golden set' of high-quality LLM outputs for benchmarking
03Automating the collection of specific traces for LLM fine-tuning or evaluation
