replace

The text in a ComfyUI workflow is never final. You want to swap && for and, clean a token out of a prompt, replace a placeholder with an actual value, or fix a spacing quirk that showed up in every line you loaded from a file. replace is the node that does it: swap every occurrence of one substring for another, or limit the surgery to the first N occurrences.

It's the string node you'll use on almost every workflow that processes text - prompt templating (swap {subject} for the actual subject), sanitizing output, converting separators between formats. If you've ever done find-and-replace in an editor, you already know this node; it's the same operation, in the graph.

How it works

Python's str.replace(old, new, count). With the default count=-1, every occurrence of old is replaced with new. Set count to a positive number and only the first that many occurrences change, scanning left to right.

• "a b a b".replace("a", "X") → "X b X b"
• "a b a b".replace("a", "X", 1) → "X b a b"
• "hello".replace("l", "L") → "heLLo"

One Python quirk to know: if old is an empty string, replace inserts new between every character - "hi".replace("", "-") → "-h-i-". It's technically correct Python, and it's almost never what you wanted.

Inputs and outputs

Three required inputs:

• string - the text to edit.
• old - the substring to find.
• new - the replacement.

Plus one optional:

• count - INT, default -1. Limit to the first N replacements; -1 means replace all.

Output is one STRING - the edited copy.

Installing it

Ships in the Basic data handling pack by StableLlama. ComfyUI Manager, search "Basic data handling", install, restart. Manual route:

cd ComfyUI/custom_nodes
git clone https://github.com/StableLlama/ComfyUI-basic_data_handling

Pure Python, zero dependencies, no downloads. Under Basic/STRING in the menu.

Common issues

Two traps. First, replace is literal, not a pattern: it matches the exact substring, so replace("a", "X") on "A" does nothing - case matters, and there's no wildcard. If you need regex, this pack has a separate regex module; if you need case-insensitive matching, lowercase both sides first. Second, the empty-old behavior above will silently insert characters everywhere if you leave old blank. And remember count counts from the left - with count=1, the first occurrence changes, so if you actually wanted the last one replaced, that's a job for splitting from the right with rsplit and rejoining.