arazzo: 1.0.1
info:
  title: Hyperbolic Tool Calling Roundtrip
  summary: Run an OpenAI-compatible tool call and feed the tool result back for a final answer.
  description: >-
    Exercises Hyperbolic's OpenAI-compatible function/tool calling over two
    chat completion turns within the same endpoint. The first turn offers a tool
    catalog and lets the model request a function call; the second turn submits
    the caller-supplied tool result back to the model to produce the final
    natural-language answer. Both steps reuse the createChatCompletion operation,
    so they are referenced by operationPath. 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: chatCompletionsApi
  url: ../openapi/hyperbolic-chat-completions-api-openapi.yml
  type: openapi
workflows:
- workflowId: tool-calling-roundtrip
  summary: Request a tool call, then return the tool result for a final answer.
  description: >-
    First chat turn presents a single function tool and lets the model decide to
    call it; the second turn supplies the tool output back as a tool message so
    the model can finish the answer.
  inputs:
    type: object
    required:
    - apiKey
    - model
    - question
    - toolName
    - toolDescription
    - toolResult
    properties:
      apiKey:
        type: string
        description: Hyperbolic API key passed as a Bearer token.
      model:
        type: string
        description: Chat model id that supports tool calling.
      question:
        type: string
        description: The user question that may require the tool.
      toolName:
        type: string
        description: Name of the single function tool offered to the model.
      toolDescription:
        type: string
        description: Human description of what the tool does.
      toolResult:
        type: string
        description: The caller-computed result string returned to the model.
  steps:
  - stepId: requestToolCall
    description: >-
      Send the question with a single function tool available and let the model
      decide whether to call it, capturing any returned tool call.
    operationPath: '{$sourceDescriptions.chatCompletionsApi.url}#/paths/~1chat~1completions/post'
    parameters:
    - name: Authorization
      in: header
      value: "Bearer $inputs.apiKey"
    requestBody:
      contentType: application/json
      payload:
        model: $inputs.model
        messages:
        - role: user
          content: $inputs.question
        tools:
        - type: function
          function:
            name: $inputs.toolName
            description: $inputs.toolDescription
        tool_choice: auto
        max_tokens: 256
        stream: false
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      finishReason: $response.body#/choices/0/finish_reason
      toolCallId: $response.body#/choices/0/message/tool_calls/0/id
      toolArguments: $response.body#/choices/0/message/tool_calls/0/function/arguments
  - stepId: submitToolResult
    description: >-
      Return the tool result back to the model as a tool message so it can
      produce the final natural-language answer.
    operationPath: '{$sourceDescriptions.chatCompletionsApi.url}#/paths/~1chat~1completions/post'
    parameters:
    - name: Authorization
      in: header
      value: "Bearer $inputs.apiKey"
    requestBody:
      contentType: application/json
      payload:
        model: $inputs.model
        messages:
        - role: user
          content: $inputs.question
        - role: tool
          tool_call_id: $steps.requestToolCall.outputs.toolCallId
          name: $inputs.toolName
          content: $inputs.toolResult
        max_tokens: 400
        stream: false
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      answer: $response.body#/choices/0/message/content
  outputs:
    toolArguments: $steps.requestToolCall.outputs.toolArguments
    answer: $steps.submitToolResult.outputs.answer