arize-prompt-optimization - Optimize LLM prompts with trace feedback
Extract prompts and performance signals from Arize traces, datasets, and experiments, then use the ax CLI to run data-driven prompt optimization.
Tags
Updated: 2026-10-05Capabilities
Typical Inputs
Typical Outputs
What this skill does
- Extract prompts from spans
- Inspect prompt templates
- Gather evaluation feedback
- Analyze experiment results
- Identify optimization patterns
- Generate revised prompts
Inputs
- Arize space name or ID
- Arize project name
- Trace or span ID
- Dataset name
- Experiment name
- Exported trace data
- Evaluation and annotation data
- LLM provider credentials
Outputs
- Extracted prompt messages
- Prompt performance analysis
- Revised prompt text
- Exported trace files
- Exported dataset files
- Exported experiment files
Requirements
- ax CLI
- Arize API access
- Valid Arize profile
- LLM provider access
- Supported LLM model