arazzo: 1.0.1
info:
  title: ChatGPT Multi-Turn Conversation
  summary: Start a conversation and continue it by chaining previous_response_id.
  description: >-
    Demonstrates the canonical Responses API multi-turn pattern. A first
    response is created from an opening prompt, and a follow-up prompt is then
    sent with previous_response_id set to the first response's id so the model
    carries the prior turn as context. The conversation's input items are
    listed afterward to confirm the chained history. Every step spells out its
    request inline so the flow can be read and executed without opening the
    underlying OpenAPI description.
  version: 1.0.0
sourceDescriptions:
- name: responsesApi
  url: ../openapi/chatgpt-responses-api-openapi.yml
  type: openapi
workflows:
- workflowId: multi-turn-conversation
  summary: Chain two Responses API turns and list the resulting input items.
  description: >-
    Creates an opening response, then creates a follow-up response that
    references the first via previous_response_id, then lists the input items
    of the follow-up to confirm the conversation history was threaded.
  inputs:
    type: object
    required:
    - apiKey
    - model
    - firstInput
    - followUpInput
    properties:
      apiKey:
        type: string
        description: OpenAI API key used as the Bearer credential.
      model:
        type: string
        description: Model ID used to generate the responses (e.g. gpt-4o).
      firstInput:
        type: string
        description: The opening prompt for the conversation.
      followUpInput:
        type: string
        description: The follow-up prompt that should reuse the prior context.
  steps:
  - stepId: openTurn
    description: >-
      Create the first stored response that opens the conversation.
    operationId: createResponse
    parameters:
    - name: Authorization
      in: header
      value: "Bearer $inputs.apiKey"
    requestBody:
      contentType: application/json
      payload:
        model: $inputs.model
        input: $inputs.firstInput
        store: true
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      firstResponseId: $response.body#/id
      firstStatus: $response.body#/status
      firstText: $response.body#/output/0/content/0/text
  - stepId: followUpTurn
    description: >-
      Create the follow-up response, passing previous_response_id so the model
      treats the opening turn as conversation context.
    operationId: createResponse
    parameters:
    - name: Authorization
      in: header
      value: "Bearer $inputs.apiKey"
    requestBody:
      contentType: application/json
      payload:
        model: $inputs.model
        previous_response_id: $steps.openTurn.outputs.firstResponseId
        input:
        - role: user
          content: $inputs.followUpInput
        store: true
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      followUpResponseId: $response.body#/id
      followUpText: $response.body#/output/0/content/0/text
  - stepId: listHistory
    description: >-
      List the input items of the follow-up response to confirm the chained
      conversation history is present.
    operationId: listResponseInputItems
    parameters:
    - name: Authorization
      in: header
      value: "Bearer $inputs.apiKey"
    - name: response_id
      in: path
      value: $steps.followUpTurn.outputs.followUpResponseId
    - name: order
      in: query
      value: asc
    - name: limit
      in: query
      value: 100
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      itemCount: $response.body#/data
      hasMore: $response.body#/has_more
  outputs:
    firstResponseId: $steps.openTurn.outputs.firstResponseId
    firstText: $steps.openTurn.outputs.firstText
    followUpResponseId: $steps.followUpTurn.outputs.followUpResponseId
    followUpText: $steps.followUpTurn.outputs.followUpText
    historyHasMore: $steps.listHistory.outputs.hasMore