Hyperbolic Multimodal Story

arazzo: 1.0.1
info:
  title: Hyperbolic Multimodal Story
  summary: List models, write a short story, illustrate it, and narrate it across four endpoints.
  description: >-
    A complete multimodal storytelling pipeline that chains all four Hyperbolic
    inference modalities. The workflow confirms the catalog, writes a short
    story with a chat model, renders a cover illustration from the story with a
    diffusion model, and narrates the story with the text-to-speech endpoint.
    Every step inlines its request and inline Authorization Bearer credential so
    the flow reads and runs without opening the OpenAPI sources.
  version: 1.0.0
sourceDescriptions:
- name: modelsApi
  url: ../openapi/hyperbolic-models-api-openapi.yml
  type: openapi
- name: chatCompletionsApi
  url: ../openapi/hyperbolic-chat-completions-api-openapi.yml
  type: openapi
- name: imageGenerationApi
  url: ../openapi/hyperbolic-image-generation-api-openapi.yml
  type: openapi
- name: audioGenerationApi
  url: ../openapi/hyperbolic-audio-generation-api-openapi.yml
  type: openapi
workflows:
- workflowId: multimodal-story
  summary: Generate a story, illustrate it, and narrate it in one pass.
  description: >-
    Confirms the catalog, writes a short story from a theme, renders a cover
    image from the story, and produces narrated speech audio of the story.
  inputs:
    type: object
    required:
    - apiKey
    - chatModel
    - imageModel
    - theme
    properties:
      apiKey:
        type: string
        description: Hyperbolic API key passed as a Bearer token.
      chatModel:
        type: string
        description: Chat model id used to write the story.
      imageModel:
        type: string
        description: Image model name used to render the cover (e.g. SDXL1.0-base).
      theme:
        type: string
        description: The theme or premise for the short story.
      language:
        type: string
        description: TTS language code for narration.
      speed:
        type: number
        description: Speech speed between 0.5 and 2.0.
  steps:
  - stepId: listCatalog
    description: >-
      List the live inference catalog so the chat and image models can be
      confirmed before generation begins.
    operationId: listModels
    parameters:
    - name: Authorization
      in: header
      value: "Bearer $inputs.apiKey"
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      models: $response.body#/data
  - stepId: writeStory
    description: >-
      Ask the chat model to write a short, self-contained story based on the
      theme.
    operationId: createChatCompletion
    parameters:
    - name: Authorization
      in: header
      value: "Bearer $inputs.apiKey"
    requestBody:
      contentType: application/json
      payload:
        model: $inputs.chatModel
        messages:
        - role: system
          content: >-
            You write a short, self-contained story of about six sentences in
            plain prose with no markdown.
        - role: user
          content: $inputs.theme
        max_tokens: 600
        temperature: 0.9
        stream: false
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      story: $response.body#/choices/0/message/content
  - stepId: illustrate
    description: >-
      Render a cover illustration using the story text as the diffusion prompt.
    operationId: generateImage
    parameters:
    - name: Authorization
      in: header
      value: "Bearer $inputs.apiKey"
    requestBody:
      contentType: application/json
      payload:
        model_name: $inputs.imageModel
        prompt: $steps.writeStory.outputs.story
        n: 1
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      coverImage: $response.body#/images/0/image
  - stepId: narrate
    description: >-
      Convert the story to speech audio and capture the base64 audio and its
      duration.
    operationId: generateAudio
    parameters:
    - name: Authorization
      in: header
      value: "Bearer $inputs.apiKey"
    requestBody:
      contentType: application/json
      payload:
        text: $steps.writeStory.outputs.story
        language: $inputs.language
        speed: $inputs.speed
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      audio: $response.body#/audio
      duration: $response.body#/duration
  outputs:
    story: $steps.writeStory.outputs.story
    coverImage: $steps.illustrate.outputs.coverImage
    audio: $steps.narrate.outputs.audio