03. Setting Up Triggers¶

This page is about deciding what wakes a workflow up. There are two ways.

Kind	When it runs	Example
Webhook (`WEBHOOK`)	Every time someone calls the address	When an order arrives, when a form is submitted
Schedule (`SCHEDULER`)	At the times you set	Every morning at 9, every Monday

Both are written on the ENTRYPOINT node. There is exactly one ENTRYPOINT per workflow.

Webhook triggers¶

The simplest form¶

- id: trigger
  name: Webhook trigger
  type: ENTRYPOINT

If you leave trigger out entirely, webhook is the default. In that case the request body you receive is passed on to the next node whole and unchanged.

If you want the incoming data validated¶

Attach an input-schema and eeumsae will validate the request for you.

- id: trigger
  name: Order webhook
  type: ENTRYPOINT
  trigger:
    kind: WEBHOOK
    input-schema:
      type: object
      required: [orderId, customerName]
      properties:
        orderId:      { type: string }
        customerName: { type: string }
        amount:       { type: number }

Using input-schema changes the behavior like this.

Situation	Result
A `required` field is missing from the request	The execution fails
The request contains a field not in `properties`	It is dropped silently (it does not reach the next node)
A field that is in `properties`	Passed through normally

In other words, input-schema is a whitelist. Every field you plan to use later has to be listed in properties. If you wrote a field and its value never arrives, check first whether you left it out of properties.

The call address¶

You can see and copy it on the workflow's Overview tab. The format is:

POST https://api.eeumsae.com/webhooks/{workspaceId}/{workflowSlug}

{workflowSlug} is the slug you chose when you created the workflow.

curl -X POST "https://api.eeumsae.com/webhooks/<workspaceId>/order-notify" \
  -H "Content-Type: application/json" \
  -d '{"orderId": "ORD-1234", "customerName": "Ada Lovelace", "amount": 25000}'

The call returns a response immediately, and the workflow runs asynchronously in the background. A 200 response does not mean the workflow succeeded — it means "received, thanks". Check the result in the execution history. → 10. Executions and Monitoring

Reading the received values¶

${nodes.trigger.response.body.orderId}

trigger here is the id of the ENTRYPOINT node. If you named the node my-trigger, it becomes ${nodes.my-trigger.response.body.orderId}.

Schedule triggers¶

Runs automatically at the times you set.

- id: trigger
  name: Every morning trigger
  type: ENTRYPOINT
  trigger:
    kind: SCHEDULER
    cron: "0 0 9 * * ?"
    timezone: "Asia/Seoul"
    lookback: PT24H

Field	Required	Description
`cron`	Required	A six-field cron expression (see below)
`timezone`	Optional	Defaults to `UTC`. If you mean local time, you must write it out — for example `Asia/Seoul`
`lookback`	Optional	Computes a query window for you and passes it along (see below)

cron expressions have six fields¶

This differs from the five-field cron you usually see. A seconds field is prepended.

┌───────────── second (0-59)
│ ┌─────────── minute (0-59)
│ │ ┌───────── hour (0-23)
│ │ │ ┌─────── day of month (1-31)
│ │ │ │ ┌───── month (1-12)
│ │ │ │ │ ┌─── day of week (0-7, 0 and 7 are Sunday)
│ │ │ │ │ │
0 0 9 * * ?

Here are the common ones.

What you want	cron
Every day at 9:00	`0 0 9 * * ?`
Every day at 9:30	`0 30 9 * * ?`
Every hour on the hour	`0 0 * * * ?`
Every 30 minutes	`0 0/30 * * * ?`
Weekdays (Mon-Fri) at 8:00	`0 0 8 ? * MON-FRI`
Every Monday at 10:00	`0 0 10 ? * MON`
Midnight on the 1st of every month	`0 0 0 1 * ?`

* and ?: the day-of-month field and the day-of-week field conflict with each other, so one of them takes a ?. If you are specifying by date, put ? in the day-of-week field; if by weekday, put ? in the day-of-month field.

Watch the time zone: if you omit timezone you get UTC. Writing only 0 0 9 * * ? runs at 6 PM Korea time. To run on Korea time, be sure to include timezone: "Asia/Seoul".

lookback — "just fetch the last 24 hours"¶

When you build a daily report, you always end up querying "the last day's worth". This feature has eeumsae compute the start and end of that window and pass them to you.

trigger:
  kind: SCHEDULER
  cron: "0 0 9 * * ?"
  timezone: "Asia/Seoul"
  lookback: PT24H        # 24 hours

With this, two values appear in the trigger output.

Value	Meaning
`${nodes.trigger.response.body.windowStart}`	Execution time − lookback
`${nodes.trigger.response.body.windowEnd}`	Execution time

Both are ISO-8601 strings (for example 2026-05-19T09:00:00.000+09:00).

- id: fetch-orders
  name: Fetch orders
  type: CALL
  integration: http_request
  input:
    uri: "https://api.example.com/orders"
    method: GET
    queryParams:
      from: "${nodes.trigger.response.body.windowStart}"
      to: "${nodes.trigger.response.body.windowEnd}"

The lookback value is written in ISO-8601 duration format.

Window you want	Notation
1 hour	`PT1H`
6 hours	`PT6H`
24 hours	`PT24H`
7 days	`P7D`
30 minutes	`PT30M`

If you omit lookback, windowStart / windowEnd are never created at all. Referencing them raises an error during execution.

Comparing the two¶

	Webhook	Schedule
Start condition	When called from outside	At the times you set
Input data	The request body (whatever you want)	None (window values if you use `lookback`)
Testing	Straight from `curl`	Wait for the time, or run it immediately over MCP
Good for	Reacting to events (orders, sign-ups, inquiries)	Recurring reports, checks and syncs

Next → 04. Integration Catalog