Explain the problem as you see it
The text parser for the AI command seems to be stripping out Tana Paste formatting conventions from the node title and node context. For example, if you try adding double brackets manually within a node title and use that node as a test it in the Prompt Workbench, you can see that they are removed in the preview.
Why is this a problem for you?
The option to send input formatted with Tana Paste would allow the AI to output more complete Tana Paste content, particularly when dealing with inline references. The AI could be instructed to maintain the format and bracket conventions of the input text, This would give it information about existing node titles, just by telling it to replicate any double brackets it sees in the input text. It could therefore maintain the links to those nodes in its output, so that we don't have to go and recreate those links manually.
My own personal use case: I'd like to use an AI command to extract the first instance of an inline reference in a node and send it back. This way, I can use it to quickly populate the value of another field. When creating a sub-task on the fly, I often will use an "@" reference while typing out the node title, and then manually need to go set the node's "Part of" field. I'd like to set up a command that instructs GPT to identify the first instance of bracketed text from the title and respond with it (also in brackets), so that I can use it to set the value of the "Part of" field automatically.
Suggest a solution
Add an option in the prompt workbench (or in the command configuration) to include Tana Paste formatting that is currently being stripped out before it is sent to OpenAI.