Flowise [BUG] Advanced Structured Output Parser Failing

Describe the bug The parser is failing for me quite often. Just connected to LangSmith and it seems the response usually includes System: before the JSON.

Failed to parse. Text: " System: {"fieldsOfStudy": ["Philosophy", "Psychology", "Sociology"], "researchSpecializations": ["Moral Psychology", "Existentialism", "Social Psychology"]}".

@KennyVaneetvelde Thank you for the feature! Do you see anything I could be doing wrong? I have 3 other Chatflows with similar results.

To Reproduce Create LLM Chain with Advanced Structured Output Parser with the following JSON

z.object({
  fieldsOfStudy: z.array(z.string()),
  researchSpecializations: z.array(z.string())
});

OpenAI: gpt-3.5 turbo-instruct Chat Prompt template:

Role and Goal:
This GPT identifies and suggests academic topics based on a given initial topic. 
It provides two lists in response: 
- Fields of Study: 5-10 high-level field of study or areas
- Research Specializations: 5-10 more specific topics, suitable for in-depth academic research or publication.

Constraints: The GPT responds only with two lists in json format and does not provide additional commentary or explanations.

Guidelines: The GPT focuses on academic relevance and specificity, ensuring that the topics are suitable for scholarly discussion and exploration. It avoids suggesting topics outside academic research or studies.

Clarification: The GPT should never ask for clarification.

Expected behavior Output to be in JSON

Screenshots