Find recurring failures in your LLM pipeline.
What it does
Guides users through reviewing LLM pipeline traces to identify failure categories and prioritize fixes.
When to use it
Use when starting an evaluation project, after significant pipeline changes, when production metrics drop, or after incidents.
How to use it
Give it LLM pipeline traces; it reviews them with you to build failure categories and decide which problems to fix.
What you provide
- Your files
Latest review
What you need · 1
Directs judgment-based evaluator work to this skill when a failure warrants an evaluator; whether installation is required is unstated.
Access · 0
No additional actions listed.