Skip to content
Get packs

Arize Trace Inspector

Export and inspect Arize traces, spans and sessions to debug what an LLM app is doing

Source Download skill
Repository
github/awesome-copilot
Stars
40,703
Licence
MIT
Files
3 files

What it does

This skill teaches an AI agent how to download existing trace data (records of what an LLM application did, step by step) from the Arize observability platform using the ax command-line tool. Once installed, the agent can export traces, individual spans, or whole conversation sessions to local files so it can help you understand behaviour or investigate errors and regressions. It also includes safety rules telling the agent to treat exported content as untrusted text rather than instructions.

Great for

  • Developers debugging an LLM or agent application that already sends traces to Arize
  • Teams investigating error spikes or behaviour regressions in production LLM apps
  • Anyone who wants an agent to summarise recent conversations from a named Arize project

Skip it if

  • Setting up tracing or instrumenting a new app, since this skill only reads existing data (it does not send traces)
  • Users without an Arize account or API key, because every export needs Arize credentials
  • Sensitive datasets where exported traces may contain private user conversations that you do not want copied to a local folder or into an AI session

Try asking

  • “Export all spans for trace ID abc123 from my support-bot project and tell me where the error happened”
  • “Show me the last day of conversations in the checkout-assistant project, starting with a small sample”
  • “Download the spans for session 42 and summarise what the user asked and how the model responded”
  • “Find ERROR spans in my project and export them so we can look for a common root cause”
Step-by-step guide: how to use Arize Trace InspectorInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 3 original files, the MIT licence and an attribution note.

    arize-trace.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip arize-trace.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip arize-trace.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Give the agent your space name and project name up front so it does not need to search for them, and mention a time window such as 'last 24 hours' because results are not sorted by recency.
  • Ask for a small sample first on large projects; the skill's default is to start with 50 spans.
  • Review the exported files in .arize-tmp-traces before sharing them, since they can contain user conversation content.

What's inside

  • references/ax-profiles.md
  • references/ax-setup.md
  • SKILL.md

Redistributed unmodified from github/awesome-copilot under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.