CREATE AND RUN A WORKFLOW

Prerequisites: Swamp installed, a repository with at least one model. If you need a model first, see Create and Run a Model.

Create a workflow

swamp workflow create deploy-pipeline

Created: deploy-pipeline
Path: workflows/workflow-<uuid>.yaml

Jobs:
  main - Main job (edit or replace)
    Steps:
      example (model_method) - Example step (edit or replace)

This writes a scaffold YAML file under workflows/.

Edit the workflow definition

Open the generated file and replace the placeholder step with your model. A minimal workflow with one step:

id: <uuid>
name: deploy-pipeline
tags: {}
jobs:
  - name: main
    description: Run the greeter
    steps:
      - name: greet
        description: Say hello
        task:
          type: model_method
          modelIdOrName: my-greeter
          methodName: execute
          inputs:
            run: "echo Deploying..."
        dependsOn: []
        weight: 0
        allowFailure: false
    dependsOn: []
    weight: 0
version: 1

For the full YAML schema, see the workflows reference.

Validate the workflow

swamp workflow validate deploy-pipeline

Validating: deploy-pipeline
  ✓ Schema validation
  ✓ Unique job names
  ✓ Unique step names in job 'main'
  ✓ Valid job dependency references
  ✓ Valid step dependency references in job 'main'
  ✓ No cyclic job dependencies
  ✓ No cyclic step dependencies in job 'main'
  ✓ Step inputs for 'greet' in job 'main' (my-greeter.execute)
  ✓ GlobalArgument input references for 'greet' in job 'main' (my-greeter.execute)
Summary: 9 passed
Result: PASSED

Validation checks schema, naming, dependency cycles, and that referenced models and methods exist.

Run the workflow

swamp workflow run blocks until the run completes, fails, or suspends at an approval gate. The exit code reflects the outcome — 0 for success, non-zero otherwise. See the workflows reference for the full flag set.

For scripted or CI use, pass --json:

swamp workflow run deploy-pipeline --json

{
  "workflowName": "deploy-pipeline",
  "status": "succeeded",
  "jobs": [
    {
      "name": "main",
      "status": "succeeded",
      "steps": [
        {
          "name": "greet",
          "status": "succeeded",
          "duration": 23
        }
      ]
    }
  ],
  "duration": 28
}

Pass inputs at run time

If your workflow defines an inputs schema, supply values with --input:

swamp workflow run deploy-pipeline --input env=staging

You can also load inputs from a YAML file:

swamp workflow run deploy-pipeline --input-file params.yaml

View run history

To see the most recent run:

swamp workflow history deploy-pipeline

Workflow: "deploy-pipeline" (Run ID: "<uuid>")
  ✓ main (24ms)
    ✓ greet (23ms)
Result: SUCCEEDED (28ms)

To search past runs:

swamp workflow history search deploy-pipeline

For approval gates, see Gate a Workflow with Manual Approval.