[BUG] Post-processing text does not appear correctly

[nikitas-novatix]

Describe the bug

Markdown formatting is not rendered correctly in the chat output when a Post-Processing is activated in the Flowise flow, even if the function is just a passthrough. Specifically, newline characters seem to be replaced by the literal string FLOWISE_NEWLINE, and Markdown elements like titles (using #) are not interpreted. This results in a broken and unformatted chat message display.

To Reproduce

Steps to reproduce the behavior:

• Create a new Chat Flow in Flowise.
• Add an LLM Chain (e.g., using OpenAI or Azure OpenAI) configured to generate Markdown output.
• Add a Post Processing function (in Settings->Post Processing)
• In the Post Processing function, simply return the raw output from the previous node: return $flow.rawOutput;
• In the chat interface, ask a question that prompts the LLM to generate a response containing Markdown formatting
• Observe the chat output. Markdown formatting will be broken; newlines will be replaced with FLOWISE_NEWLINE, and titles will not be rendered as headings.

Expected behavior

Markdown formatting generated by the LLM Chain should be rendered correctly in the Chat Output, even when passing through a Post-processing function. The chat message should display formatted text with newlines, headings, and other Markdown elements.

Screenshots

Setup

Installation: Azure App Service

Flowise Version: 2.2.7-patch.1

OS: macOS

Browser: Chrome

Additional context

The issue occurs even when the Post Processing function is a simple passthrough function, just returning the $flow.rawOutput directly. This suggests that Flowise may be processing outputs from the JS function differently from direct LLM node outputs, possibly treating them as plain text and applying unwanted sanitization or string replacements that break Markdown rendering. This significantly impacts the ability to use Post Processing in flows where maintaining Markdown formatting is important for the user interface.