Workflow Examples

This page shows complete workflow definitions you can adapt. Every example uses step types and condition syntax that exist in @loop/workflow-engine today.

Example 1: Thyroid review

This is the smallest useful workflow: check a biomarker, then add a recommendation when the condition is met.


id: thyroid-review-v1
name: Thyroid Review
version: 1.0.0
triggers:
  - type: biomarker_result
    params:
      biomarker: TSH
steps:
  - id: check-tsh
    type: check_biomarker
    action:
      type: check_value
      params:
        biomarker: TSH
        threshold: 2.5
        operator: ">"
  - id: recommend-selenium
    type: recommend_supplement
    condition: "check-tsh.result === true"
    action:
      type: recommend
      params:
        supplement: Selenium
        dosage: 200mcg

Run it with patient data:


import { WorkflowEngine } from '@loop/workflow-engine';
 
const workflow = WorkflowEngine.load(thyroidYaml);
 
const result = await workflow.execute({
  patientId: 'patient-42',
  data: {
    biomarker: {
      TSH: 4.2,
    },
  },
});
 
console.log(result.recommendations);

Example 2: Contraindication-aware recommendation

This flow checks a biomarker first, then inspects current medications before allowing a recommendation.


id: vitamin-k-review-v1
name: Vitamin K Review
version: 1.0.0
steps:
  - id: check-inflammation
    type: check_biomarker
    action:
      type: check_value
      params:
        biomarker: hsCRP
        threshold: 1.0
        operator: ">"
  - id: check-warfarin
    type: check_contraindication
    condition: "check-inflammation.result === true"
    action:
      type: check
      params:
        medication: warfarin
        supplement: Vitamin K
  - id: recommend-vitamin-k
    type: recommend_supplement
    condition: "check-warfarin.result.hasContraindication === false"
    action:
      type: recommend
      params:
        supplement: Vitamin K
        dosage: 100mcg

Example context:


const result = await workflow.execute({
  patientId: 'patient-99',
  data: {
    biomarker: { hsCRP: 2.1 },
    medications: ['metformin'],
  },
});

Example 3: Branching with `next`

By default, the engine walks steps sequentially. Use next when you want a non-linear flow.


id: follow-up-routing-v1
name: Follow-Up Routing
version: 1.0.0
steps:
  - id: check-glucose
    type: check_biomarker
    next: notify-patient
    action:
      type: check_value
      params:
        biomarker: fastingGlucose
        threshold: 100
        operator: ">"
  - id: order-a1c
    type: order_lab
    action:
      type: order
      params:
        test: a1c
        priority: normal
  - id: notify-patient
    type: send_notification
    action:
      type: notify
      params:
        channel: email
        message: "Your glucose review is ready."

In this definition, check-glucose routes directly to notify-patient, skipping order-a1c.

Example 4: Tool-eligible workflow

run_tool is useful when a workflow needs a named capability, but the current executor returns a stubbed result. This is best used to prove the contract and reserve a stable tool ID.


id: risk-evaluation-v1
name: Risk Evaluation
version: 1.0.0
steps:
  - id: invoke-risk-tool
    type: run_tool
    action:
      type: tool
      params:
        toolId: risk-calculator
        labPanel: advanced-metabolic

Execution context:


const result = await workflow.execute({
  patientId: 'patient-101',
  tools: ['risk-calculator'],
});

Current output shape:


{
  "executed": true,
  "toolId": "risk-calculator",
  "params": {
    "toolId": "risk-calculator",
    "labPanel": "advanced-metabolic"
  }
}

Example 5: Scheduled workflow

Triggers are declarative metadata and can be used by an external scheduler or orchestration layer.


id: quarterly-labs-v1
name: Quarterly Labs Reminder
version: 1.0.0
triggers:
  - type: schedule
    params:
      cron: "0 9 1 */3 *"
steps:
  - id: order-panel
    type: order_lab
    action:
      type: order
      params:
        test: comprehensive_metabolic_panel
        priority: normal
  - id: notify-patient
    type: send_notification
    action:
      type: notify
      params:
        channel: email
        message: "Your quarterly lab panel is ready to order."

Picking the right pattern

Need	Pattern
Threshold check	`check_biomarker`
Recommendation output	`recommend_supplement`
Operational next step	`order_lab` or `send_notification`
Medication gate	`check_contraindication`
Future external capability	`run_tool`

Next steps

Continue to Creating Workflows
Validate behavior in Testing Workflows
Keep the field-level contract handy with DSL Reference