Zammad-AI

Appearance

Zammad Integration ​

The Zammad integration layer provides the clients used by both services to talk to Zammad.

Architecture ​

The integration is built around BaseZammadClient, which defines the shared operations for the workflow and index job.

BaseZammadClient ​

The abstract interface located in zammad-ai-workflow/app/zammad/base.py. It defines methods for:

  • get_ticket(id): Retrieving a ticket with its articles.
  • post_answer(ticket_id, text, internal): Posting a reply or internal note.
  • post_shared_draft(ticket_id, text): Saving a draft.

Implementations ​

ZammadAPIClient (REST API) ​

Located in zammad-ai-workflow/app/zammad/api.py and zammad-ai-index/job/zammad/api.py.

  • Uses Bearer Token authentication when the API token secret is configured.
  • Handles ticket retrieval, knowledge base lookup, and answer synchronization.
  • Maps Zammad JSON responses to Pydantic models.

ZammadEAIClient (Internal EAI) ​

Located in zammad-ai-workflow/app/zammad/eai.py and zammad-ai-index/job/zammad/eai.py.

  • Used when the deployment talks to Zammad through EAI instead of the REST API.
  • The client is selected through the zammad.type setting.

Models ​

Incoming Zammad data and outgoing requests are validated using Pydantic models defined in app/models/zammad.py and job/models/zammad.py.

  • zammad.type
  • zammad.base_url -> Base URL of the Zammad instance. Use it for API calls and to generate links, such as knowledge base articles.
  • zammad.knowledge_base_id
  • zammad.ai_ticket_group_id
  • zammad.ai_ticket_group_name
  • zammad.ai_ticket_author
  • zammad.auth_token or EAI OAuth fields
  • zammad.document_parsing
  • zammad.eai_url (if using EAI)