n8n

Monitor your n8n AI workflows with Maev. Every workflow execution is tracked as a session — failures, LLM calls, tool usage, and costs all appear in your dashboard automatically.

How it works

Maev provides a dedicated endpoint for n8n:

POST https://<your-domain>/api/ingest/n8n

You configure one of two things in your n8n workflow:

A Webhook trigger that fires when the workflow is called
An HTTP Request node at the end of the workflow that POSTs execution data to Maev

Maev reads the execution data, maps each node run to an event, classifies failures, and closes the session.

Setup

1. Get your API key

Go to Settings → API Key in your Maev dashboard and copy your key (vl_xxx).

2. Add a Webhook node (recommended)

If your workflow starts with a Webhook trigger, you can use it to also send data back to Maev.

Add an HTTP Request node at the end of your workflow:

Field	Value
Method	POST
URL	`https://<your-domain>/api/ingest/n8n`
Authentication	Header Auth → `X-API-Key: vl_xxx`
Body Content Type	JSON
Body	See below

Body (enable "Send execution data" in your Webhook trigger settings, then reference fields):

{
  "execution": {
    "id": "{{ $execution.id }}",
    "workflowId": "{{ $workflow.id }}",
    "workflowName": "{{ $workflow.name }}",
    "status": "{{ $execution.status }}",
    "startedAt": "{{ $execution.startedAt }}",
    "stoppedAt": "{{ $now.toISO() }}",
    "mode": "{{ $execution.mode }}"
  }
}

3. Using the n8n "Execute Workflow" trigger with execution data

If you want full node-level telemetry (each node's timing, outputs, and errors), enable "Include Execution Data" in your webhook node settings. Maev will automatically parse each node's run and create a timeline event per node.

4. Test it

Run your workflow once and check the Sessions page in Maev — you should see a new session appear within seconds.

What Maev captures automatically

Source	What Maev reads
OpenAI / LLM nodes	Model, prompt, completion, token usage
Tool / Code nodes	Execution time, errors
HTTP Request nodes	Duration, errors
Workflow level	Total duration, status, failure reason

Error detection

Maev's classification engine runs on every session end. For n8n workflows it detects:

Tool failure — any node that throws an error
Latency spike — total workflow duration exceeds threshold
Cost anomaly — LLM token spend above baseline (when OpenAI node data is present)
Infinite loop — same node pattern repeated more than expected

Example: AI Agent workflow

[Webhook Trigger]
       ↓
[OpenAI Chat Model node]  ← LLM call captured
       ↓
[HTTP Request node]       ← tool call captured
       ↓
[HTTP Request → Maev]     ← sends execution data to Maev

Troubleshooting

Session not appearing in dashboard

Check that the HTTP Request node is the last step and executes even on error (use an Error Workflow or always-execute node)
Verify the X-API-Key header is set correctly
Check n8n execution logs for the HTTP Request node response

No node-level detail showing

Ensure "Include Execution Data" is enabled in your n8n Webhook trigger
Pass the full execution object including data.resultData.runData

Status always shows "completed" even on error

Make sure status is set to "error" in the body for failed runs — reference {{ $execution.status }} directly

GitHub Zapier