Overview

A workflow is a YAML file in .rigg/. Its filename doesn’t matter — the id field is the identifier.

Top-level fields

id: my-workflow             # Required: unique identifier
inputs:                     # Optional: typed parameters
  name:
    type: string
    default: world
env:                        # Optional: environment variables (supports templates)
  API_KEY: ${{ inputs.key }}
steps:                      # Required: non-empty step array
  - id: greet
    type: shell
    with:
      command: echo "hello ${{ inputs.name }}"

Inputs

JSON Schema objects. Supported types: string, number, integer, boolean, object, array.

inputs:
  name:
    type: string
    default: world
    enum: ["Alice", "Bob"]
    minLength: 1
  count:
    type: integer
    minimum: 0
  config:
    type: object          # requires 'properties'
    required: [name]
    properties:
      name:
        type: string
  tags:
    type: array           # requires 'items'
    items:
      type: string
    minItems: 1

Only top-level inputs support default. Nested properties cannot.

Common step fields

Every step type supports these:

- id: my_step               # Optional: unique across the workflow
  type: ...                  # Required
  if: ${{ condition }}       # Optional: skip when false
  env:                       # Optional: step-scoped env vars
    KEY: value

ID rules: starts with ASCII letter or _, then alphanumeric, _, or -.

Scoping rules

• Step results are visible only to subsequent steps in the same scope
• Exports promote results through container boundaries (group, loop, branch, parallel)
• Access exports via steps.<container_id>.result.<export_name>
• Step IDs must be unique across the entire workflow, not just within a scope