Workflow Triggers

Besides starting flows explicitly (API, SDK, admin console Run dialog), flows can start automatically when nodes change. Triggers are raisin:Trigger nodes in the functions workspace that reference a flow through function_flow.

Node-Event Triggers

A node-event trigger starts the referenced flow whenever a matching node event occurs.

node_type: raisin:Trigger
properties:
  name: on-article-published
  title: On Article Published
  trigger_type: node_event
  enabled: true
  config:
    event_kinds: [Updated]            # which events fire the trigger
  filters:
    workspaces: [content]             # glob patterns
    node_types: ["my:Article"]
    # paths: ["/articles/**"]         # optional path globs
    # property_filters:               # optional conditions on the node's properties
    #   status: published
    #   "seo.score": { $gte: 80 }
  function_flow:
    raisin:ref: /flows/publish-pipeline
    raisin:workspace: functions

Property	Description
trigger_type	node_event
config.event_kinds	Any of Created, Updated, Deleted, Published, Unpublished, Moved, Renamed
filters.workspaces	Workspace glob patterns
filters.paths	Path glob patterns. * matches within one segment, ** matches across segments
filters.node_types	Node type list, for example ["raisin:Page", "raisin:Asset"]
filters.property_filters	Object of property path (dot notation for nested values) to either a literal value or an operator object: $exists, $eq, $ne, $gt, $gte, $lt, $lte, $in
function_flow	Reference to the raisin:Flow node to start. raisin:ref may be the node's path or id; the server resolves a path to the id on write
enabled	Default true
priority	Execution priority (lower runs first)
max_retries	Retry budget for the trigger's execution job

Trigger nodes live under a folder in the functions workspace, since the workspace root only allows folders. New repositories come with a /triggers folder.

The Flow Input

When a trigger starts a flow, the flow input describes the event and carries the node:

{
  "event": {
    "type": "Created",
    "node_id": "JyR6mKiKBUV5Om7lhGZDC",
    "node_type": "raisin:Page",
    "node_path": "/hello"
  },
  "node": {
    "id": "JyR6mKiKBUV5Om7lhGZDC",
    "path": "/hello",
    "name": "hello",
    "node_type": "raisin:Page",
    "workspace": "blog",
    "properties": { "title": "Hello", "featured": true, "tags": ["a", "b"] },
    "created_by": "system",
    "created_at": "2026-09-06T18:40:18.849950+00:00"
  },
  "workspace": "blog"
}

So a step reads the changed node's properties as input.node.properties.*. Trigger metadata is also available under trigger.* (event_type in lower case, node_id, node_type, node_path, workspace, tenant_id, repo_id, branch):

- id: notify
  node_type: raisin:FlowStep
  properties:
    action: Notify the author
    function_ref: /lib/notify-author
    arguments:
      path: "{{ trigger.node_path }}"
      event: "{{ trigger.event_type }}"
      title: "{{ input.node.properties.title }}"

See Data and Templates.

Triggers That Call a Function

A trigger can reference a single function through function_path instead of a flow. The function receives the event directly. When a trigger has both, function_flow wins. Use function_flow when you want routing, retries, compensation, or human tasks, even for single-step automations.

Scheduled Triggers

Schedule-based triggers run on a cron expression. Every minute the server evaluates the enabled schedule triggers of every repository and queues an execution for each match:

node_type: raisin:Trigger
properties:
  name: daily-cleanup
  title: Daily Cleanup
  trigger_type: schedule
  enabled: true
  config:
    cron_expression: "0 2 * * *"      # 02:00 UTC every day
  function_path: /lib/cleanup-old-data

Supported cron syntax (5 fields: minute, hour, day, month, day-of-week, evaluated in UTC):

Pattern	Meaning
*	Any value
*/15	Every 15 units (step values)
1-5	Range
1,3,5	List
@hourly, @daily / @midnight, @weekly (Monday), @monthly, @yearly / @annually	Presets

Scheduled triggers invoke a function (function_path). To run a workflow on a schedule, point the trigger at a small function that starts the flow through the flow API, or have the function do the work itself.

Starting Flows Explicitly

Channel	How
HTTP	POST /api/flows/{repo}/run with { "flow_path": "/flows/...", "input": {...} }
HTTP (test run)	POST /api/flows/{repo}/test with an additional test_config (mocks, isolated branch)
SDK	flows.run(path, input), flows.runAndWait(...), flows.runAndCollect(...)
Admin console	Repository sidebar, Flows, Run dialog (JSON input and live event view)

For an API start, trigger.event_type is manual and the flow input is exactly the input you sent.

Next Steps

• Function Triggers: triggers that invoke functions directly, including HTTP triggers
• Examples: complete runnable workflows