marked/ModelAsAService/code/API_definitions/qa-assistant-service/raw.md

qa-assistant-service

openapi: 3.0.3
info:
  title: Q&A Assistant - Service
  description: >-
    This API provides service interfaces for the Question and Answer (Q&A) Assistant, enabling users to submit query requests to a configured Q&A assistant and obtain answers generated based on the knowledge base and Large Language Model (LLM).
    It is suitable for scenarios such as building intelligent question-answering systems and customer service bots that require real-time responses to user inquiries.
    It must follow RFC 3339 and must have time zone. Recommended format is yyyy-MM-dd'T'HH:mm:ss.SSSZ.

    ## Security and Authentication

    This API uses OAuth 2.0 with OpenID Connect for authentication. The following scope is required:
    - qa-assistant-service:answer:read - Permission to query assistants

    # Authorization and authentication

    The "Camara Security and Interoperability Profile" provides details of how an API consumer requests an access token. Please refer to Identity and Consent Management (https://github.com/camaraproject/IdentityAndConsentManagement/) for the released version of the profile.

    The specific authorization flows to be used will be agreed upon during the onboarding process, happening between the API consumer and the API provider, taking into account the declared purpose for accessing the API, whilst also being subject to the prevailing legal framework dictated by local legislation.

    In cases where personal data is processed by the API and users can exercise their rights through mechanisms such as opt-in and/or opt-out, the use of three-legged access tokens is mandatory. This ensures that the API remains in compliance with privacy regulations, upholding the principles of transparency and user-centric privacy-by-design.


    # Additional CAMARA error responses

    The list of error codes in this API specification is not exhaustive. Therefore the API specification may not document some non-mandatory error statuses as indicated in `CAMARA API Design Guide`.

    Please refer to the `CAMARA_common.yaml` of the Commonalities Release associated to this API version for a complete list of error responses. The applicable Commonalities Release can be identified in the `API Readiness Checklist` document associated to this API version.

    As a specific rule, error `501 - NOT_IMPLEMENTED` can be only a possible error response if it is explicitly documented in the API.


    # Further info and support

    (FAQs will be added in a later version of the documentation)

  license:
    name: Apache 2.0
    url: https://www.apache.org/licenses/LICENSE-2.0.html
  version: 0.1.0
  x-camara-commonalities: 0.6

externalDocs:
  description: Product documentation at CAMARA
  url: https://github.com/camaraproject/ModelAsAService

servers:
  - url: '{apiRoot}/qa-assistant-service/v0.1'
    variables:
      apiRoot:
        default: http://localhost:9091
        description: API root, defined by the service provider, e.g. `api.example.com` or `api.example.com/somepath`

paths:
  /answer:
    post:
      operationId: queryAssistant
      summary: Get an answer from a QA assistant
      description: |
        Queries a configured QA assistant to generate answers using knowledge bases and LLMs.

        **Key Features:**
        - Utilizes pre-configured assistant settings (prompts, model parameters)
        - Leverages associated knowledge bases for contextual responses
        - Provides termination status for answer generation

        **Key Rules:**
        - Requires valid UUID `assistantId`
        - `prompt` must be 1-1000 characters
        - Needs `qa-assistant-service:answer:read` scope
        - Assistant must be properly configured and accessible

        **Response Characteristics:**
        - **200 OK**: Returns `QueryResponse` with answer text and termination status
          - `finished` indicates generation completion reason (Stop/Length/Error/Filter)
          - `reference` shows knowledge base source when available
        - **400 Bad Request**:
          - Missing/invalid prompt
          - Invalid assistantId format
          - Prompt length exceeds limits
        - **401 Unauthenticated**: Missing/invalid credentials
        - **403 Forbidden**:
          - Missing required scope
          - Token/assistantId mismatch
        - **404 Not Found**:
          - Assistant doesn't exist
          - Identifier not found
        - **500 Internal Server Error**:
          - LLM processing failure
          - Knowledge base access error
      tags:
        - Query
      parameters:
        - $ref: '#/components/parameters/x-correlator'  # Added correlator
      security:
        - openId:
            - qa-assistant-service:answer:read
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/QueryRequest'
      responses:
        '200':
          description: Successful answer retrieval
          headers:
            x-correlator:
              $ref: '#/components/parameters/x-correlator'
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/QueryResponse'
        '400':
          $ref: '#/components/responses/Generic400'
        '403':
          $ref: '#/components/responses/Generic403'
        '404':
          $ref: '#/components/responses/Generic404'
        '500':
          $ref: '#/components/responses/Generic500'

components:
  securitySchemes:
    openId:
      type: openIdConnect
      openIdConnectUrl: https://example.com/.well-known/openid-configuration
      description: OpenID Connect authentication

  parameters:
    x-correlator:
      name: x-correlator
      in: header
      description: Correlation ID for request tracing
      schema:
        $ref: "#/components/schemas/XCorrelator"

  headers:
    x-correlator:
      description: Correlation id for the different services
      schema:
        $ref: "#/components/schemas/XCorrelator"

  schemas:
    XCorrelator:
      type: string
      pattern: ^[a-zA-Z0-9-_:;.\/<>{}]{0,256}$
      example: "b4333c46-49c0-4f62-80d7-f0ef930f1c46"
    QueryRequest:
      type: object
      required:
        - assistantId
        - prompt
      properties:
        assistantId:
          type: string
          format: uuid
          description: Unique identifier of the QA assistant
          example: "a1b2c3d4-5678-90ab-cdef-1234567890ab"
        prompt:
          type: string
          minLength: 1
          maxLength: 1000
          description: The user's question or query
          example: "How do I reset my password?"

    QueryResponse:
      type: object
      required:
        - success
        - finished
        - answerText
      properties:
        success:
          type: boolean
          description: Indicates whether the query was processed successfully
        finished:
          type: string
          description: Termination status of the response generation
          enum:
            - Stop
            - Length
            - Error
            - Filter
          example: "Stop"
        answerText:
          type: string
          description: The assistant's generated answer
          example: "To reset your password, go to settings > security..."
        reference:
          type: string
          description: Optional reference to the knowledge base source
          example: "kb123_section5"

    ErrorInfo:
      type: object
      required:
        - status
        - code
        - message
      properties:
        status:
          type: integer
          description: HTTP status code
        code:
          type: string
          description: Error code identifier
        message:
          type: string
          description: Human-readable error message

  responses:
    Generic400:
      description: Bad request due to invalid input
      headers:
        x-correlator:
          $ref: '#/components/parameters/x-correlator'
      content:
        application/json:
          schema:
            allOf:
              - $ref: '#/components/schemas/ErrorInfo'
              - type: object
                properties:
                  status:
                    enum:
                      - 400
                  code:
                    enum:
                      - MISSING_PROMPT
                      - INVALID_ASSISTANT_ID
                      - INVALID_ARGUMENT
                      - OUT_OF_RANGE
          examples:
            GENERIC_400_MISSING_PROMPT:
              description: Missing required prompt parameter
              value:
                status: 400
                code: MISSING_PROMPT
                message: "Required parameter 'prompt' is missing"
            GENERIC_400_INVALID_ASSISTANT_ID:
              description: Invalid assistant ID format
              value:
                status: 400
                code: INVALID_ASSISTANT_ID
                message: "Assistant ID must be a valid UUID"
            GENERIC_400_INVALID_ARGUMENT:
              description: Invalid argument
              value:
                status: 400
                code: INVALID_ARGUMENT
                message: "Client specified an invalid argument"
            GENERIC_400_OUT_OF_RANGE:
              description: Parameter value out of range
              value:
                status: 400
                code: OUT_OF_RANGE
                message: "Prompt length exceeds maximum allowed characters"

    Generic403:
      description: Forbidden
      headers:
        x-correlator:
          $ref: '#/components/parameters/x-correlator'
      content:
        application/json:
          schema:
            allOf:
              - $ref: '#/components/schemas/ErrorInfo'
              - type: object
                properties:
                  status:
                    enum:
                      - 403
                  code:
                    enum:
                      - PERMISSION_DENIED
                      - INVALID_TOKEN_CONTEXT
          examples:
            GENERIC_403_PERMISSION_DENIED:
              description: User lacks required permissions
              value:
                status: 403
                code: PERMISSION_DENIED
                message: User lacks required permissions to query this assistant
            GENERIC_403_INVALID_TOKEN_CONTEXT:
              description: Token context inconsistency
              value:
                status: 403
                code: INVALID_TOKEN_CONTEXT
                message: "Assistant ID is not consistent with access token"

    Generic404:
      description: Resource not found
      headers:
        x-correlator:
          $ref: '#/components/parameters/x-correlator'
      content:
        application/json:
          schema:
            allOf:
              - $ref: '#/components/schemas/ErrorInfo'
              - type: object
                properties:
                  status:
                    enum:
                      - 404
                  code:
                    enum:
                      - NOT_FOUND
                      - ASSISTANT_NOT_FOUND
                      - IDENTIFIER_NOT_FOUND
          examples:
            GENERIC_404_NOT_FOUND:
              description: Resource is not found
              value:
                status: 404
                code: NOT_FOUND
                message: The specified resource is not found.
            GENERIC_404_ASSISTANT_NOT_FOUND:
              description: Assistant not found
              value:
                status: 404
                code: ASSISTANT_NOT_FOUND
                message: "Specified assistant ID does not exist"
            GENERIC_404_IDENTIFIER_NOT_FOUND:
              description: Identifier not found
              value:
                status: 404
                code: IDENTIFIER_NOT_FOUND
                message: Assistant identifier not found.

    Generic500:
      description: Internal server error
      headers:
        x-correlator:
          $ref: '#/components/parameters/x-correlator'
      content:
        application/json:
          schema:
            allOf:
              - $ref: '#/components/schemas/ErrorInfo'
              - type: object
                properties:
                  status:
                    enum:
                      - 500
                  code:
                    enum:
                      - INTERNAL_SERVER_ERROR
                      - INTERNAL
          examples:
            GENERIC_500_INTERNAL_SERVER_ERROR:
              description: Unexpected server error
              value:
                status: 500
                code: INTERNAL_SERVER_ERROR
                message: "Unexpected server error occurred while processing query"
            GENERIC_500_INTERNAL:
              description: Internal server error
              value:
                status: 500
                code: INTERNAL
                message: Unknown server error