[Bug]: Tool call arguments are being appended to the end of the answer.

[ricac1]

OpenRAG Version

0.3.2

Deployment Method

Docker

Operating System

macOS Tahoe 26.2

Python Version

3.13.12

Affected Area

Chat (chat interface, conversations, AI responses)

Bug Description

The model (meta-llama/llama-3-405b-instruct model) is returning correct natural language answers, however, the problem is that the tool call arguments (for example, {“search_query": "..."}) are being appended to the end of the response text.

Examples:

Steps to Reproduce

I uploaded a document to ingest, waited for ingestion to complete, and started chatting with the document by filtering for that specific one.

Expected Behavior

The response text should not have the JSON query appended at the end.

Actual Behavior

When I ask a question in Chat, the JSON query appears for each response text.

Relevant Logs

Screenshots

No response

Additional Context

No response

Checklist

• I have searched existing issues to ensure this bug hasn't been reported before.
• I have provided all the requested information.