Fetch and analyze LLM application traces from the Phoenix observability platform.
npx skills add https://github.com/exiao/skills --skill phoenix-cli-exiao
Please help me install this Agent Skill. Skill: phoenix-cli Source: https://github.com/exiao/skills/tree/main/phoenix-cli Command: npx skills add https://github.com/exiao/skills --skill phoenix-cli-exiao
This Skill helps you debug and understand your LLM applications by providing tools to inspect traces, identify errors, and analyze experiments directly from the command line.
Use the phoenix-cli skill to fetch the 10 most recent traces for your application.
High-intent search queries and answers about installing and using this skill.
Trace analysis helps debug LLM application failures by fetching detailed execution traces to identify errors within specific spans. You can inspect these traces directly from the command line to pinpoint the root cause of inconsistent results.
You can fetch and inspect LLM app traces from the command line using Phoenix CLI. It enables fetching recent traces, filtering for failures, and supporting raw JSON output for programmatic access and integration with data manipulation tools.
Yes, you can use jq for advanced data manipulation when reviewing LLM experiments. The Phoenix CLI integrates with jq and supports raw JSON output, allowing you to programmatically filter and analyze experiment performance and error patterns.
The best way to diagnose inconsistent RAG system results is to fetch execution traces and identify failing retrieval or LLM spans. Using command-line tools to analyze these traces allows you to quickly pinpoint the root cause of the inconsistency.
Yes, Phoenix CLI supports raw JSON output for programmatic access to spans. This allows you to extract trace data and pipe it into other command-line tools for advanced data manipulation and automated debugging workflows.
You need an observability platform to review LLM experiments because it captures detailed execution traces and spans. Analyzing this data enables you to identify performance bottlenecks, debug failures, and understand error patterns within your application.
qcmuuCapture OpenTelemetry traces and evaluation results for LLM debugging.
kubesense-aiAnalyze Kubernetes distributed traces to diagnose latency and error patterns.
ChuaHanChongRun evaluation commands and profile GPU usage to establish ML model baseline metrics.
stephanjQuery Datadog logs, metrics, and dashboard configurations via CLI.
