Skip to main content

Documentation Index

Fetch the complete documentation index at: https://arize-ax.mintlify.dev/docs/llms.txt

Use this file to discover all available pages before exploring further.

See a bad response? Open the LLM span directly in the Prompt Playground with its exact inputs pre-loaded — iterate using real production data, not synthetic examples.

​
How to do it

  1. Open a trace and click into an LLM span
  2. Click Open in Playground in the span toolbar
  3. The Playground opens with the prompt template, input variables, and model settings pre-filled
  4. Edit the prompt, adjust parameters, click Run to compare
  5. Save to Prompt Hub for version control and deployment
[screenshot: prompt playground with a span’s data pre-loaded]
Works with any LLM span that has prompt template data. Auto-instrumentation captures prompt templates automatically for most providers.
The Prompt Playground is designed for single LLM call iteration — editing prompt templates, adjusting parameters, and comparing outputs. It does not support replaying full agentic workflows (multi-step tool calls, agent chains). For agentic debugging, use the Agent Insights and Agent Trajectory views to understand the full execution path, then fix individual LLM prompts in the Playground.

​
What you can change

  • Prompt template — rewrite instructions, add constraints, change tone
  • Model — switch providers or versions to compare behavior
  • Parameters — temperature, max tokens, top-p
  • Input variables — swap in different values to test edge cases

​
Save and deploy

  1. Save to Prompt Hub to version the prompt
  2. Your app pulls the latest version from Prompt Hub automatically
  3. Run an experiment to validate the change before deploying