Text Template

The one thing that will trip you up

The default template reads Hello {name}, welcome to {place}! - and if you hit Run without touching it, that's literally what comes out, braces and all. This node only substitutes {var1} through {var5}. {name} and {place} are decorative. The docstring advertises "custom named placeholders" as a feature; the code doesn't implement any. Change the template to {var1} and {var2} (or leave the vars empty) or your "template" gets passed through untouched.

What it actually does

ComfyBros Text Template is a string-formatter node: you write a prompt skeleton with up to five placeholders and fill them from anywhere in the graph - a caption or tag node, a random-prompt generator, a seed counter, even the metadata output of this pack's video nodes if you want to feed numbers back in. It's the ComfyUI equivalent of a Python f-string dropped into the middle of a workflow.

Reach for it when you're assembling prompts from pieces. Swap a subject, a style token, or a location into a fixed skeleton without hand-editing three separate text nodes every time you change one word. The wider ecosystem is full of fancier prompt-tooling nodes, but for "stick five strings into a template," this is about as direct as it gets.

How it works

Under the hood it's a plain .replace() pass, in order, var1 through var5. Only non-empty variables get substituted: leave var4 empty and a literal {var4} survives into the output and straight into your sampler's positive prompt, where the model will read it as garbage text. There's also no escaping - if a value you feed in contains {var2}, the later replacement fires on the freshly inserted text too. For a five-slot formatter that's a fair trade; just know what you're getting.

The inputs and the output

• template - the multiline string skeleton, with {var1}…{var5} placeholders.
• var1 … var5 - optional strings, default "". The only things that actually get replaced.
• Output: text (STRING) - wire it into any prompt input, e.g. the positive_prompt of ComfyBros Text to Video.

Installing

Easiest via ComfyUI Manager - search for "ComfyBros". Or manually:

cd ComfyUI/custom_nodes
git clone https://github.com/turnbros/ComfyBros

Then restart ComfyUI. One wrinkle: the README tells you to run pip install -r requirements.txt, but no requirements.txt exists in the repo - the command just fails. It doesn't matter for this node. It's pure string operations and needs nothing beyond what ComfyUI already ships; the declared dependencies (pillow, numpy, requests) are all present in any working ComfyUI install.

Troubleshooting

• Output still contains braces - you used {name}-style placeholders, or you left a var empty. Both cases pass the template through unmodified.
• Only five slots - need more variables than that? Chain two templates, or reach for a purpose-built text-utility node; this one is deliberately minimal.
• The dependable node in the pack - worth saying: unlike its siblings, this one needs no API key, no instance, no cloud account. The video nodes in ComfyBros all demand a configured RunPod endpoint. This one just works, and that's kind of the point.