Extracting a Figma Make Transcript?

replying to keep this bumped.  I want to have more user tools to enhance the AI agent experience with me as a non tech builder.  

It really is quite unfortunate that one cannot simply just export the Figma Make AI transcript to a text file or something. It’s weird because Figma Make can certainly create all kinds of wonderful markdown files in the code to document minute things that come up during the development process.

Copying and pasting each page scroll in the chat sucks and it isn’t accurate as some stuff just doesn’t get selected. I ended up taking hundreds of screenshots and creating a cropping macro (I asked CoPilot to write it) to then use in the Scripter plugin for Figma.

And, I’ve still to actually copy the text from the screenshots to paste into a text document (good thing Live Text is a thing on Mac). I might just call it good at this point because boy, it’s a PITA.

Why would I even need this anyways? I guess I just wanted to be able to feed the transcript into another AI tool for context to be able to ask newb developers type questions. You know… to learn stuff. But, alas, it is not to be. 

If you communicate clearly with the figma agent and say this is the beginning of the session as you sit down at the beginning of the session and you indicate that at the end of the session today you want all the text or a very detailed summary to be put into a markdown file it's not a bad workaround

I keep trying to make little workarounds like this I developed one that takes a screenshot. It's a button on my application in figma and it does the screenshot of just the figma preview. It then downloads automatically to the browser and you see it in the top right corner so you can grab it with your mouse and drop it into the input dialog box hit the return button and it's a fantastic way to do quick screenshots to give the agent a graphical view of what is going on as well 

Absolutely feel the need. I sit down with the Agent to detail out a prototype. I want all my original explanations so that I can generate other complementary documentation resources from it. 

UPDATE: I tried to OCR the screenshots using AI (Copilot) and that was a waste of time. Even though the AI could read the screenshots, it just gave me the run around when it came to transcribing it into a document. Ultimately, I was able to cobble together a workflow using CLI tools namely ocrmypdf (and sips to remove alpha channel from png screenshots)

find ~/Documents/OCR/screenshots -type f -name "*.png" -print0 \

| LC_ALL=C sort -z \

| while IFS= read -r -d '' f; do \

    tmp_img="$(mktemp).jpg"; \

    tmp_txt="$(mktemp)"; \

    sips -s format jpeg "$f" --out "$tmp_img" >/dev/null; \

    ocrmypdf -j 1 -f --image-dpi 144 --output-type none --sidecar "$tmp_txt" "$tmp_img" /dev/null; \

    printf "\n\n===== %s =====\n\n" "$(basename "$f")" >> ~/Documents/OCR/full_transcript_raw.txt; \

    cat "$tmp_txt" >> ~/Documents/OCR/full_transcript_raw.txt; \

    rm "$tmp_img" "$tmp_txt"; \

  done

So, I just put all my screenshots into a folder and pointed this to it. Next step would be to take the raw transcript and format it and/or feed it back into an AI as a context document.