workflows-automations

name	workflows-automations
description	Workflow engine — triggers, node types, conditions, variables, approval flows. This skill should be used when the user asks to create workflows, set up triggers, build approval flows, or automate business processes.

Workflows & Automations

Expert guidance on NocoBase workflow engine — triggers, node types, execution modes, variables, and common automation patterns.

Workflow Types

Type	Trigger	Use Case
Collection event	Record created/updated/deleted	Notifications, validations, cascading updates
Schedule	Cron expression	Reports, cleanup, data sync
Action	User clicks button	Manual approval, batch processing

Trigger Configuration

Collection Event Trigger

Trigger: Collection event
Collection: orders
Event: afterCreate
Condition: {{ $context.data.total > 1000 }}
Mode: async (recommended for non-blocking)

Events: afterCreate, afterUpdate, afterDelete, beforeCreate, beforeUpdate, beforeDelete

Schedule Trigger

Trigger: Schedule
Expression: "0 9 * * 1-5"  (weekdays at 9:00 AM)
Mode: async

Common cron patterns:

• Every hour: 0 * * * *
• Daily at 9 AM: 0 9 * * *
• Weekdays at 9 AM: 0 9 * * 1-5
• Every Monday: 0 0 * * 1
• First of month: 0 0 1 * *

Action Trigger

Trigger: Action (button click)
Collection: purchase_requests
Button label: "Submit for Approval"

Node Types

Condition

Branching logic based on expressions.

Node: Condition
Expression: {{ $context.data.amount > 10000 }}
True branch → Manager Approval
False branch → Auto Approve

Calculation

Compute values and store in variables.

Node: Calculation
Expression: {{ $context.data.price * $context.data.quantity * (1 - $context.data.discount / 100) }}
Variable: lineTotal

Query

Fetch records from a collection.

Node: Query
Collection: contacts
Filter: { email: {{ $context.data.contactEmail }} }
Result variable: contactRecord

Create Record

Node: Create
Collection: notifications
Data: {
  title: "New order #{{ $context.data.id }}",
  message: "Order created by {{ $context.data.createdBy.name }}",
  type: "order"
}

Update Record

Node: Update
Collection: orders
Filter: { id: {{ $context.data.id }} }
Data: {
  status: "approved",
  approvedAt: {{ NOW() }}
}

Delete Record

Node: Delete
Collection: temp_records
Filter: { createdAt: { $lt: {{ DATEADD(NOW(), -30, 'days') }} } }

HTTP Request

Node: Request
Method: POST
URL: https://hooks.slack.com/services/xxx
Body: {
  "text": "New order #{{ $context.data.id }} - ${{ $context.data.total }}"
}

Manual (Approval)

Node: Manual
Assignees: {{ $context.data.department.manager }}
Actions: Approve, Reject
Timeout: 48 hours

Loop

Node: Loop
Target: {{ $jobsData.queryResult }}
Loop body nodes: Process each record

Parallel Branch

Node: Parallel
Branches:
  - Branch A: Send email notification
  - Branch B: Create audit log
  - Branch C: Update dashboard
(All execute simultaneously)

Delay

Node: Delay
Duration: 24 hours
Then: Send follow-up email

Variable System

Context Variables

Variable	Description
$context.data	Trigger record data
$context.user	Current user
$context.params	Action parameters

Job Variables

Variable	Description
$jobsData.nodeKey	Output from a specific node
$jobsData.queryResult	Result of Query node
$jobsData.calculationResult	Result of Calculation node

System Functions

Function	Description
NOW()	Current datetime
DATEADD(date, n, unit)	Add time to date
DATEDIFF(date1, date2, unit)	Difference between dates

Common Workflow Patterns

Record Create → Notify

Trigger: orders.afterCreate
1. Query: Get order creator details
2. Request: POST to Slack webhook
   Body: "New order #{{$context.data.id}} by {{$jobsData.query.name}}"

Record Update → Cascade

Trigger: deals.afterUpdate
Condition: status changed to "Won"
1. Update: Set all related tasks to "Completed"
2. Create: Add entry to revenue_log
3. Request: Notify sales team

Scheduled Cleanup

Trigger: Schedule (daily at 2 AM)
1. Query: Find records older than 90 days with status "Draft"
2. Loop: For each record
   2a. Delete: Remove record
3. Create: Add cleanup log entry

Approval Flow

Trigger: Action (Submit button)
1. Update: Set status to "Pending Approval"
2. Condition: amount > 10000?
   Yes → Manual: Assign to Director
   No → Manual: Assign to Manager
3. Condition: Approved?
   Yes → Update: status = "Approved"
   No → Update: status = "Rejected"
4. Request: Send notification

Execution Modes

Mode	Description	Use When
Sync	Blocks until complete	Before-event validation
Async	Non-blocking	After-event notifications, heavy processing

Error Handling in Workflows

HTTP Request Failures

Trigger: orders.afterCreate
1. Request: POST to external API
2. Condition: Check request result
   - Success → Continue normal flow
   - Failure → Create: Log error to error_logs collection
              → Request: Send alert to admin webhook
              → Delay: 5 minutes
              → Request: Retry the original request

Missing Data Handling

Trigger: contacts.afterCreate
1. Condition: {{ $context.data.email != null }}
   - True → Request: Send welcome email
   - False → Create: Log "missing email" to audit_logs

Cascading Error Prevention

Trigger: orders.afterUpdate
1. Query: Get related order_items
2. Condition: {{ $jobsData.queryResult.length > 0 }}
   - True → Loop: Update each item
   - False → (skip, no items to update)

Variable Passing Between Nodes

Using $context.data

Access trigger record fields:

{{ $context.data.id }}          // Record ID
{{ $context.data.status }}      // Field value
{{ $context.data.createdBy }}   // Creator reference

Using $jobsData

Access output from previous nodes:

{{ $jobsData.queryNodeKey }}              // Full query result
{{ $jobsData.queryNodeKey[0].name }}      // First result's name field
{{ $jobsData.calculationNodeKey }}        // Calculation result

Chaining Multiple Conditions

Trigger: deals.afterUpdate
1. Condition: status changed?
   - Yes:
     2. Condition: new status == "Won"?
        - Yes:
          3. Calculate: commission = amount * 0.05
          4. Create: commission_record
          5. Request: Notify sales team
        - No:
          2b. Condition: new status == "Lost"?
              - Yes:
                3b. Create: lost_reason_record
                4b. Request: Notify manager
              - No: (skip)
   - No: (skip, no status change)

Best Practices

1. Use async mode — unless you need to block the user operation
2. Add conditions early — filter out irrelevant triggers to save processing
3. Test incrementally — build one node at a time and test
4. Use variables — pass data between nodes via the variable system
5. Handle errors — add condition nodes to check for null/missing data before using them
6. Log important events — create audit trail records in workflows
7. Keep workflows focused — one workflow per business process
8. Use parallel branches — for independent actions (notify + log simultaneously)
9. Check for null — always verify query results are not empty before accessing fields
10. Retry on failure — for HTTP requests, add retry logic with delay nodes