Control flow

workflow

Calls another workflow in the same .rigg project.

- id: call_plan
  type: workflow
  with:
    workflow: plan
    inputs:
      requirements: ${{ inputs.requirements }}
      retries: 3
      dry_run: true

Rules:

• with.workflow must be a static string, not a template expression
• String with.inputs values are rendered as templates in the caller scope
• Boolean and number YAML literals pass through unchanged
• steps.call_plan.result is null in v1, so workflows stay independent and share data through files or git state
• workflow steps may set retry

group

Encapsulates steps into a single unit. Internal results are hidden unless exported.

- id: gather
  type: group
  steps:
    - id: log
      type: shell
      with:
        command: git log --oneline -20
        stdout:
          mode: text
    - id: diff
      type: shell
      with:
        command: git diff --stat
        stdout:
          mode: text
  exports:
    log: ${{ steps.log.result }}
    diff: ${{ steps.diff.result }}

Downstream steps access exports via steps.gather.result.log.

loop

Repeats until a condition is met or max iterations is reached.

- id: fix_loop
  type: loop
  max: 5 # required
  until: ${{ len(steps.review.result.findings) == 0 }} # optional
  steps:
    - id: review
      type: codex
      with:
        kind: review
        target:
          type: uncommitted
    - id: fix
      if: ${{ len(steps.review.result.findings) > 0 }}
      retry:
        max: 3
        delay: 1s
      type: codex
      with:
        kind: turn
        prompt: |
          Address these review findings:
          ${{ toJSON(steps.review.result) }}
  exports:
    clean: ${{ len(steps.review.result.findings) == 0 }}

Loop results always include reason.

• until_satisfied when until becomes truthy
• max_reached when the loop uses all iterations

Loop context variables

Inside loop bodies, run.* variables are available:

Variable	Type	Description
run.iteration	integer	Current iteration (1-based)
run.max_iterations	integer	Max iterations value
run.node_path	string	Node path

branch

Conditional execution. First matching if wins. else is the fallback.

- id: decide
  type: branch
  cases:
    - if: ${{ steps.check.result == 'needs_fix' }}
      steps:
        - id: fix
          type: codex
          with:
            kind: turn
            prompt: Fix the issues.
      exports:
        outcome: ${{ steps.fix.result }}
    - else:
      steps: []
      exports:
        outcome: "no issues"

Rules:

• All cases must export the same shape, or none at all
• If any case has exports, an else case is required
• else: takes no value (just else: on its own)

parallel

Runs branches concurrently.

- id: checks
  type: parallel
  branches:
    - id: unit # id required on each branch
      steps:
        - id: run_unit
          type: shell
          with:
            command: npm test
    - id: lint
      steps:
        - id: run_lint
          type: shell
          with:
            command: npm run lint
  exports:
    unit: ${{ steps.run_unit.result }}
    lint: ${{ steps.run_lint.result }}

Exports can reference steps.* from any branch in the parallel block.

Exports

All control flow types (group, loop, branch, parallel) support exports to promote internal results to the outer scope:

exports:
  name: ${{ steps.inner_step.result }}
  field: ${{ steps.inner_step.result.nested_field }}

Without exports, inner step results are invisible to downstream steps.