Zammad 7.0 - Gemini AI Integration Error: "Unknown name 'store': Cannot find field

Infos:

  • Used Zammad version: 7
  • Used Zammad installation type: any
  • Operating system: Ubuntu 22
  • Browser + version:

Expected behavior:

  • After successfully configuring the Custom (OpenAI Compatible) provider, AI features such as Ticket Summarization should work seamlessly.

  • The AI provider should generate a summary and display it in the ticket interface

Actual behavior:

• While the initial connection test/setup succeeds, the actual usage of AI features (e.g., clicking “Summarize”) fails.
• The Zammad production log shows a 400 Bad Request from the API provider.
• The error returned is: Invalid JSON payload received. Unknown name “store”: Cannot find field.
• It appears that Zammad successfully validates the endpoint during setup (likely via a simple handshake), but the specific payload sent during summarization tasks includes the “store” parameter, which is rejected by the Google Gemini OpenAI-compatible API.

Steps to reproduce the behavior:

  1. Successfully configure a Custom (OpenAI Compatible) provider using the Google Gemini API endpoint.
  2. Open any ticket with existing articles.
  3. Click on the AI Summarize button .

The process fails, and the logs confirm that the “store” field in the JSON payload is the cause of the INVALID_ARGUMENT error

According to my research the Gemini-API is not OpenAI compatible.

Gemini REST-API vs OpenAI REST-API.

Or do you have a specific OpenAI compatible endpoint from Google?

Hi,

Actually, Google does provide a specific OpenAI-compatible endpoint for Gemini. You can find the official documentation here: OpenAI compatibility  |  Gemini API  |  Google AI for Developers

The base URL I am using is: https://generativelanguage.googleapis.com/v1beta/openai/

The integration actually works for the initial handshake/validation. However, the issue arises during the actual request (e.g., when clicking “Summarize”).

Oh, in that case, can you please create a bug issue?

1 Like
2 Likes

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.