Node reference: Synthetic and observability

Add these nodes from the designer palette (Synthetic Tests and Observability groups). Click a node on the canvas to open its config panel — everything below is filled in there.

Web Test

Drives a real browser through a multi-step journey on a synthetic agent. Use it as the "do what a user does" step of a Workbook: log in, search, add to cart, check out.

What you fill in

Field	What to enter
Assigned Agents	Which synthetic agents may run the test. Leave empty to use any available agent; pick your private agent for apps behind your firewall.
Starting URL	The page the journey starts on. Supports `{{variable}}` values.
Wait Strategy	How the browser decides a page has settled before the next step (e.g. wait for network idle).
Width / Height	The browser window size the test runs at.
Test Steps	The journey itself, built step by step in the step builder: Navigate, Click, Type Text, Read Text, Wait, Assert (Validate) and more — the same builder Synthetics web tests use (see Web (browser) tests).
Timeout (ms) / Retries	How long the test may take overall, and how many times to retry it on a transient error.

What it outputs

Whether the journey passed, how long it took, how many steps ran and passed, the network requests the browser made, and the text of any Read Text step. That text becomes a variable for later nodes — for example, read the order id off the confirmation page and check it in the database next.

Example — login journey: Navigate to the login page → Type Text into the email and password fields (use {{user}} / {{password}} from secrets) → Click the submit button → Read Text from the welcome banner into greeting → Assert the banner contains "Welcome".

API Test

Sends an HTTP request (or a multi-step API flow) from a synthetic agent and checks the response.

What you fill in

Field	What to enter
URL / Method	The endpoint and HTTP method. The URL supports `{{variable}}` values.
Expected Status	The HTTP status that counts as a pass (e.g. 200, 201).
Headers / Body	Optional request headers (added row by row) and a request body.
Auth Type	None, Bearer Token, username + password, or an API key in a custom header. Store the credential as a secret and reference it as a `{{variable}}`.
Base URL	For multi-step flows: the prefix shared by every step.
Assigned Agents	As for Web Test — pick your private agent for internal APIs.
Timeout (ms) / Retries	Response deadline and retry count for transient errors.

What it outputs

Pass/fail, the status code, response time, and the captured response. Use the node's Variable Extraction panel to pull a value out of the response body — e.g. bind orderId to body.id — for downstream nodes.

APM Observe

This is what makes a Workbook more than a ping — after your synthetic step runs, APM Observe pulls the APM traces those requests generated inside your service, so you can assert on what actually happened in the backend.

What you fill in

Field	What to enter
Service Name	The service to inspect, exactly as it reports to APM. The panel checks live whether recent traces exist under that name.
Operation Name	Optional — narrow the check to one operation.
Time Window (seconds)	How far back to look for traces.
Collection Delay (seconds)	Wait this long before collecting, so traces have time to arrive.
Max Retries / Retry Interval (seconds)	Keep re-checking until traces appear.
Correlate web-test requests	Match the exact requests your synthetic step made to the traces they produced.

What it outputs

The number of spans found, whether any had errors, the error count, and the service's latency in the window (e.g. p95). Assert on these with an Assertion node — for example: no errors and p95 under 800 ms. You can also capture a span tag into a variable for later steps.

Place APM Observe after the synthetic node so the traces exist; the Collection Delay field covers ingest lag.