Arize Link

Generate deep links to the Arize UI for traces, spans, and sessions.

When to Use

User wants a link to a specific trace, span, or session
You have trace/span/session IDs from exported data or logs and need to link back to the UI
User asks to "open" or "view" a trace/span/session in Arize

Required Inputs

Collect these from the user or from context (e.g., exported trace data, parsed URLs):

org_id -- Base64-encoded organization ID (from URL path or user)
space_id -- Base64-encoded space ID (from URL path or user)
project_id -- Base64-encoded project/model ID (from URL path or user)
One of:
- trace_id (and optionally span_id) for trace/span links
- session_id for session links

URL Construction

Base URL: https://app.arize.com (override for on-prem if the user specifies a custom base URL)

Trace Link

Opens the trace slideover showing all spans in the trace.

{base_url}/organizations/{org_id}/spaces/{space_id}/projects/{project_id}?selectedTraceId={trace_id}&queryFilterA=&selectedTab=llmTracing&timeZoneA=America%2FLos_Angeles&startA={start_epoch_ms}&endA={end_epoch_ms}&envA=tracing&modelType=generative_llm

If a span_id is also available, add &selectedSpanId={span_id} to highlight that span within the trace.

Span Link

Opens a specific span within a trace. Both trace_id and span_id are required.

{base_url}/organizations/{org_id}/spaces/{space_id}/projects/{project_id}?selectedTraceId={trace_id}&selectedSpanId={span_id}&queryFilterA=&selectedTab=llmTracing&timeZoneA=America%2FLos_Angeles&startA={start_epoch_ms}&endA={end_epoch_ms}&envA=tracing&modelType=generative_llm

Session Link

Opens the session view for a conversation/interaction flow.

{base_url}/organizations/{org_id}/spaces/{space_id}/projects/{project_id}?selectedSessionId={session_id}&queryFilterA=&selectedTab=llmTracing&timeZoneA=America%2FLos_Angeles&startA={start_epoch_ms}&endA={end_epoch_ms}&envA=tracing&modelType=generative_llm

Time Range

CRITICAL: startA and endA are required query parameters. Without them, the Arize UI defaults to the last 7 days and will show a "Your model doesn't have any recent data" error if the trace/span falls outside that window.

startA: Start of the time window as epoch milliseconds
endA: End of the time window as epoch milliseconds

How to Determine the Time Range

Use these sources in priority order:

User-provided URL: If the user shared an Arize URL, extract startA and endA from it and reuse them. This is the most reliable approach since it preserves the user's original time window.

Exported span data: If you have span data (e.g., from ax spans export), use the span's start_time field to calculate a range that covers the data:

# Convert span start_time to epoch ms, then pad ±1 day
python -c "
from datetime import datetime, timedelta
t = datetime.fromisoformat('2026-03-07T05:39:15.822147Z'.replace('Z','+00:00'))
start = int((t - timedelta(days=1)).timestamp() * 1000)
end = int((t + timedelta(days=1)).timestamp() * 1000)
print(f'startA={start}&endA={end}')
"

Default fallback: Use the last 90 days. Calculate:
- startA: (now - 90 days) as epoch milliseconds
- endA: current time as epoch milliseconds

Instructions

Gather the required IDs from the user or from available context (URLs, exported trace data, conversation history).
Determine startA and endA epoch milliseconds using the priority order above.
Substitute values into the appropriate URL template above.
Present the URL as a clickable markdown link.

Example Output

Given: org_id=QWNjb3VudE9yZ2FuaXphdGlvbjoxOmFiQzE=, space_id=U3BhY2U6MTp4eVo5, project_id=TW9kZWw6MTpkZUZn, trace_id=0123456789abcdef0123456789abcdef

Trace link:

https://app.arize.com/organizations/QWNjb3VudE9yZ2FuaXphdGlvbjoxOmFiQzE=/spaces/U3BhY2U6MTp4eVo5/projects/TW9kZWw6MTpkZUZn?selectedTraceId=0123456789abcdef0123456789abcdef&queryFilterA=&selectedTab=llmTracing&timeZoneA=America%2FLos_Angeles&startA=1700000000000&endA=1700086400000&envA=tracing&modelType=generative_llm

arize-link