Find failure patterns in your LLM app's traces.
What it does
Analyzes LLM application traces in Phoenix to identify failure patterns and prioritize fixes and evaluations.
When to use it
Use when investigating LLM or agent failures, making sense of traces, organizing failure observations, or deciding which evaluations to build.
How to use it
Give it a request to investigate LLM failures; it organizes observations into failure categories with counts and shares a Phoenix UI link.
Uses
What you need · 4
Requires access to the Phoenix CLI, either through npx with Node.js or through a global installation.
Requires a reachable Phoenix instance and a configured project to inspect.
Uses Python 3 to URL-encode the annotation filter when constructing the wrap-up Phoenix UI link.
Access · 1
This skill
Phoenix
Records notes and failure labels