Nodepad_嵌入式文本

I'm going to save you an hour: Nodepad_嵌入式文本 ("embedded text") does not work. It's the roughest edge in this pack, and knowing that up front beats discovering it at 11pm when your workflow won't run.

The page advertises it as a text-to-conditioning node - the registry lists it with a tips input and CONDITIONING, positive, and negative outputs, which is the shape of a node that turns your text into sampler-ready conditioning. That's what the name promises and what the README's screenshots imply. It's not what the shipped code does.

What the code actually does

Open up AC_TextTool.py and the function body is:

def text_ac(self, tips=None):
    file = open('D/:text.text', 'r', encoding='UTF-8')
    str(file.read())
    print(str)
    return (str,)

Three separate problems, any one of which is fatal:

1. It reads a hardcoded path. D/:text.text is a Windows drive path that exists on exactly one machine on Earth - the author's. On your machine open() throws FileNotFoundError the instant the node executes. There's no input for the file path and no way to tell it where your file lives.
2. It ignores what it reads. Even if the file existed, str(file.read()) reads the content and then throws the result away - the return statement doesn't use it.
3. It returns the wrong thing. return (str,) returns the builtin str type, not a string and certainly not conditioning. Whatever you wire to those advertised CONDITIONING/positive/negative outputs gets a Python class object, and the sampler will reject it.

So as shipped this node is a guaranteed red-error node, and even in the author's own environment it would have returned garbage. This isn't a config issue you can fix from the UI.

What to use instead

If a workflow you downloaded has this node in it, delete it and drop in either:

• Nodepad_提示词文本框 - the pack's CLIP text encode wrapper (takes text + clip, outputs proper CONDITIONING). That's the node this one was clearly meant to be.
• Or core's CLIP Text Encode - identical job, no pack needed.

In other words, "embedded text" as advertised is what ComfyUI already does natively, and the pack's own working text node does it better.

Installing the pack

• ComfyUI Manager: search "ComfyUI_AC_FUN" and install.
• Manual:

cd ComfyUI/custom_nodes
git clone https://github.com/A719689614/ComfyUI_AC_FUN-V1.0

Restart ComfyUI. The pack is Cc啊程's (a Bilibili creator and architect by trade) Chinese-labeled suite; no model downloads, no heavy dependencies.

Bottom line

One broken node out of a ~40-node pack isn't a reason to avoid the rest - most of it is a harmless renaming of core functionality, and nodes like Math_数学表达式 are genuinely useful. But this one is a trap, and it's the kind of thing that happens when a hobbyist pack ships features as screenshots before the code is done. Swap it for the pack's working text encoder and move on.