prompt

Collect interactive input from users during workflow execution. Supports text, password, select, confirm, and multiselect input types. Automatically skips prompts in non-interactive environments (CI/CD, LLM agents) using default values.

Parameters

Parameter	Type	Required	Description
message	string	Yes	The prompt message displayed to the user
type	string	No	Input type: text (default), password, select, confirm, multiselect
default	varies	No	Default value if user presses Enter or running non-interactively
options	array	For select/multiselect	List of choices

Return Values

Field	Type	Description
result.value	string	The user’s input (or default in non-interactive mode)
result.type	string	The prompt type used
result.interactive	string	”true” if user was prompted, “false” if default was used
result.output	string	Same as value (for consistency with shell function)

Non-Interactive Mode

Prompts are automatically skipped when:

1. ORCHSTEP_NON_INTERACTIVE=true environment variable is set
2. stdin is not a terminal (piped input, CI/CD runners)
3. A --var matching the step name provides the value

This means the same workflow works both interactively (human at terminal) and non-interactively (CI/CD pipeline, LLM agent).

# Interactive — prompts appear
orchstep run deploy

# Non-interactive — prompts skipped, defaults used
ORCHSTEP_NON_INTERACTIVE=true orchstep run deploy

# Override specific prompt — skip that prompt, use provided value
orchstep run deploy --var environment=prod

Examples

Text Input

- name: get_name
  func: prompt
  args:
    message: "Enter your name"
    type: text
    default: "World"
  outputs:
    name: "{{ result.value }}"

Password (Masked)

- name: get_password
  func: prompt
  args:
    message: "Database password"
    type: password

Select (Single Choice)

- name: choose_env
  func: prompt
  args:
    message: "Target environment"
    type: select
    options: [dev, staging, production]
    default: staging

Confirm (Yes/No)

- name: confirm_deploy
  func: prompt
  args:
    message: "Deploy to production?"
    type: confirm
    default: false

Multiselect (Multiple Choices)

- name: select_features
  func: prompt
  args:
    message: "Features to enable"
    type: multiselect
    options: [logging, monitoring, caching, cdn]
    default: [logging, monitoring]

Full Workflow: Interactive Deployment

name: interactive-deploy
desc: "Deployment with user confirmation"

defaults:
  app: "myapp"

tasks:
  deploy:
    steps:
      - name: environment
        func: prompt
        args:
          message: "Target environment"
          type: select
          options: [dev, staging, production]
          default: dev

      - name: version
        func: prompt
        args:
          message: "Version to deploy"
          type: text
          default: "latest"

      - name: confirm
        func: prompt
        args:
          message: "Deploy {{ defaults.app }} v{{ steps.version.value }} to {{ steps.environment.value }}?"
          type: confirm
          default: false

      - name: execute_deploy
        if: '{{ eq steps.confirm.value "true" }}'
        func: shell
        do: |
          echo "Deploying {{ defaults.app }} v{{ steps.version.value }} to {{ steps.environment.value }}..."
          echo "DEPLOY_STATUS=success"

Design Philosophy

The prompt function bridges the gap between interactive runbooks (human at terminal) and automated pipelines (CI/CD, LLM agents):

• For humans: Rich prompts with defaults, validation, and confirmation before destructive actions
• For automation: Same workflow, same YAML — prompts silently use defaults or --var overrides
• For LLM agents: ORCHSTEP_NON_INTERACTIVE=true or provide all values via --var

This means you write ONE workflow that serves all three audiences.