langfuse-advanced-filters

What problem does it solve?

Manually sifting through vast Langfuse traces and observations for specific issues is time-consuming and complex. This skill automates precise data retrieval, allowing you to quickly pinpoint performance bottlenecks, error patterns, or specific workflow behaviors without endless scrolling.

Core Features & Use Cases

• Advanced Filtering: Apply precise filters using operators like >, <, =, contains, in, and not_in across any trace or observation field.
• Metadata-based Queries: Filter by custom metadata such as case_id, profile_name, or workflow_version for highly targeted analysis.
• Aggregated Metrics: Query average latency, error counts, or P95 trends, grouped by dimensions, to gain high-level insights.
• Use Case: Quickly find all traces with latency greater than 5 seconds for a specific case_id to debug slow workflows, or identify all ERROR-level observations from a particular node within a custom time range.

Quick Start

Find all Langfuse traces for case_id=0001 that have a latency greater than 5000 milliseconds, from November 1st to November 4th, 2025. Save the results to /tmp/langfuse_queries/slow_traces.json.