Verify it worked

A step-by-step guide to confirming your events landed in the Obsivara dashboard.

Step 1 — Send a test event

If you haven't already, send a quick test event to confirm your API key and connection are working:

bash

curl -X POST https://api.obsivara.com/ingest/webhook/generic \
  -H "Authorization: Bearer obs_live_YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "event": "run_finished",
    "run_id": "verify-run-1",
    "asset_id": "my-agent",
    "outcome": "success",
    "duration_ms": 500
  }'

You should receive HTTP 202 with:

json

{ "source": "generic", "accepted": 1, "rejected": 0, "errors": [] }

If you get a 401 or 403, see Errors & troubleshooting to fix the key or scope.

A 202 response with accepted: 1 and rejected: 0 means your event was accepted by the ingest API. It will appear in the dashboard within about one minute.

Step 2 — Open the dashboard

Go to app.obsivara.com and log in to your workspace.

Obsivara dashboard — inventory overview showing assets and recent runs

Step 3 — Find your asset

On the main dashboard (inventory/health page), look for the asset_id you sent in your event — for example my-agent. Click it to open the asset detail view.

If the asset does not appear immediately, wait one minute and refresh. New assets can take up to one minute to be indexed.

Timing: Run events and errors appear within ~1 minute. Health scores and recommendations are computed by background workers on a schedule and may lag by a few minutes after first ingest.

Step 4 — Inspect the run

Click into a run to see the full trace: spans, LLM calls with token and cost breakdowns, tool calls, and any errors recorded during the run.

Run detail view in Obsivara — showing spans, LLM token usage, and cost breakdown

What to check

Run status: The run should show as success (or whatever status you sent). If it shows error, check your event's outcome field.
Token + cost data: If you sent llm_call events (or are using the batch endpoint), the run detail should show token counts and cost per call. This data is not available from n8n sync — only from direct instrumentation.
Spans: If you used span events or OTLP, the trace view will show a waterfall of spans.
Errors: Any error events will appear in the Errors tab of the asset view.

Troubleshooting

If the run is not visible after two minutes:

Re-check the 202 response — confirm accepted is greater than zero and rejected is zero.
Verify the asset_id in your event matches what you are searching for in the dashboard (case-sensitive).
Check that your API key is associated with the correct workspace — a key from workspace A will not write to workspace B.
See Errors & troubleshooting for a full error reference.

Next steps

Once your first event is confirmed, you can:

Send the full run lifecycle (run_started → llm_call → run_finished) for richer trace data — see Connect your AI agent.
Use the /ingest/batch endpoint to send multiple events per request for efficiency — see Event schema.
Forward runs from n8n, LangSmith, or Langfuse — see the Connect section.