Trace Explorer
Browse stored AI runs from Agent Lab, coach sessions, chat, and API ingestion.
Trace Explorer is where you search and open traces saved in your workspace.
Traces come from:
- Run agent and Agent chat (local Ollama runs)
- Coach sessions (local draft and coach review)
- SDK / API ingestion (production apps sending
POST /api/traces)
For each trace you can inspect prompt, response, model, tokens, latency, cost, and metadata (including trainingPreferred for coach picks).
What you can do
- Search by trace ID, action name, or model (
/dashboard/traces?q=...) - Open a trace detail page for full prompt/response text
- Review linked execution records (tokens, latency,
estimatedCostUsd) - Optionally review governance or verification state when those features are in use
Trace investigation workflow
Execution Timeline
Review the full lifecycle of an AI request from initiation to completion.
This helps teams understand:
- trace ID
- request timestamps
- workflow IDs
- model providers
- execution duration
- completion status
Prompt and Response Inspection
Inspect what entered the model and what was returned.
This helps teams investigate:
- hallucinations
- malformed prompts
- unexpected outputs
- workflow regressions
Cost Intelligence
Review operational economics tied to each trace.
Track:
- input tokens
- output tokens
- total spend
- model allocation
- latency costs
- cost anomalies
Governance Events
Review policy enforcement activity tied to each execution.
Examples include:
- PII detection
- credential exposure
- policy violations
- risk escalations
Verification Records
Move directly from trace investigation into verification workflows.
Validate:
- SHA-256 integrity hashes
- execution records
- timestamps
- verification artifacts
Operational Benefits
Without trace visibility, teams often struggle to answer basic operational questions:
- Why did this model behave unexpectedly?
- Why did this request cost more than expected?
- Which workflow triggered this issue?
- Was this record modified after execution?
Trace Explorer allows teams to investigate these issues without jumping between fragmented logs, provider dashboards, and disconnected monitoring systems.
It becomes the operational entry point for governance, optimization, verification, and audit workflows.