Quickstart: n8n

In about five minutes, you add the verified Mibo Testing node to an n8n AI workflow, send one trace, and create a test from it. Mibo evaluates the traces your workflow sends. It never calls your workflow. It runs next to the observability tool you already use.

Before you start

• A Mibo account with a verified email. Mibo rejects traces and AI features until you verify. See Email verification.
• An n8n workflow that starts with a When chat message received trigger and runs an AI Agent node. A Webhook trigger works too. A Manual Trigger does not carry a user message.
• The owner or admin role on your n8n instance. Only they can install community nodes.
• Access to the n8n API, which this quickstart uses. The n8n API is not available during the n8n Cloud free trial. On the trial, use the HTTP Request node instead.

1. Set up your Agent in Mibo

1. Create an account at app.mibo-ai.com, or sign in, and click Set up your Agent.
2. Enter an Agent Name, such as “Support Bot”, and choose the n8n Agent type. Click Continue to API key.
3. Click Create Agent API key and copy the key. Mibo shows it only once.
4. Click I’ve saved it — continue to traces.

The key is scoped to this Agent, so the node does not need an Agent ID. If your project already has an Agent, see Add another Agent and Creating an API key.

2. Create an n8n API key

The node reads your workflow’s structure through the n8n API, so each span carries the node name you see in n8n.

1. In n8n, open Settings → n8n API.
2. Click Create an API key, enter a label, and choose an expiration.
3. Copy the key. On n8n Enterprise, you can limit it to the workflow:read scope.

3. Install the Mibo Testing node

1. Open your workflow and open the nodes panel with + or the N key.
2. Search for Mibo Testing. It appears under More from the community.
3. Select it and click Install.

The steps are the same on n8n Cloud and self-hosted. On self-hosted n8n, you can also install the npm package @mibo-ai/n8n-nodes-mibo-testing from Settings → Community Nodes. See n8n’s guide to installing verified community nodes.

4. Add the node to your workflow

1. Add Mibo Testing as the last node of the workflow, after the AI Agent node. The node captures the nodes that ran before it.

2. Under Credential to connect with, create a Mibo Testing API credential:

   Field	Value
   API Key	Your Mibo API key from step 1.
   n8n API Key (Optional)	The n8n API key from step 2. This quickstart uses it to read the workflow. The node also accepts an upstream Get Workflow node instead, but that node calls the n8n API too.
   n8n API URL	On n8n Cloud: https://<your-name>.app.n8n.cloud/api/v1. On self-hosted n8n: the default http://localhost:5678/api/v1 works when n8n listens on port 5678. Otherwise, use https://<your-n8n-host>/api/v1.

3. Leave Agent ID empty.

4. Under Options, add Include Input Data in Output and turn it on. The chat trigger replies with the last node’s output, and this option passes the AI Agent’s output fields through.

5. Open the node’s Settings tab and set On Error to Continue. If Mibo is unreachable, the workflow still finishes.

6. Save the workflow.

5. Send your first trace

1. Open the workflow’s chat and send a message, such as “Where is my order A-1042?”. The node sends a trace on every execution, including runs from the editor.
2. In Mibo, open the Agent’s Traces page. The new trace has one span per captured node.

If the trace does not appear, check the Mibo Testing node’s output in the execution log. See Verifying it works for common causes.

6. Create your first test

1. In the Agent navigation, open Create tests with AI.
2. Click Use real trace. In Ground generation on a real trace, select your trace and click Use this trace.
3. Review the generated test cases and click Save. Keep the test enabled, with Passive behavior set to Always run or Run when relevant.
4. Send another chat message. Mibo evaluates each new trace against your saved tests.
5. Open the test and check Last Result.

For status meanings and next actions, see Read the result status.

What Mibo needs from the trace

Data	Where the node finds it	Without it
User input	The trigger node’s output, such as the chat message	Run when relevant tests are skipped, and the trace shows No usable input. Always run tests still run.
Final response	The AI Agent node’s output	Checks on the reply have no text to read.
Steps and tool calls	Every node that ran before the Mibo Testing node, and the AI Agent’s tool calls	Step and tool checks cannot match.

The node does not capture a Manual Trigger. See If no input was available for scenario routing.

What the node sends

The node posts plain JSON to POST https://api.mibo-ai.com/public/traces with your key in the x-api-key header. It does not gzip requests, although Mibo accepts gzip. Mibo accepts up to 10 MB per request, and the node warns when its trace is above about 8 MB. Mibo routes the trace by the Agent-scoped key, not by an OTLP service name. Before sending, the node replaces values under common secret field names. Mibo redacts secrets again when the trace arrives. See Data handling and the node’s protection settings.

Note

Using n8n’s native OTLP export instead? Self-hosted n8n 2.19.0 and later can export OTLP/HTTP protobuf on its own. Set N8N_OTEL_EXPORTER_OTLP_TRACING_PATH=/traces, because n8n otherwise appends /v1/traces to the endpoint. Set N8N_OTEL_EXPORTER_SERVICE_NAME to the Agent’s OTLP Service Name. That export carries no node input or output, so most traces have no user input and no reply for Mibo to check. See Native OTLP export.

What’s next

• Creating test cases to refine the generated tests.
• Passive testing to understand which tests run on each trace.
• Send traces from n8n for every node setting.