Anima Prompt Corrector

You pasted a prompt from a Civitai page, it works, but you know the order is wrong - the artist tag is buried mid-prompt and the quality tags are scattered. Anima's tag order convention matters enough that the community made a whole ecosystem of tools around it, and Anima Prompt Corrector is this pack's take: it normalizes a comma-separated prompt into ANIMA order, cleans up syntax, and - the part that makes it worth installing - tells you exactly what it changed.

What it does

One multiline prompt goes in. Out come two strings: corrected_prompt (the reordered, cleaned text) and report (a JSON report covering changed state, unknown tags, duplicate tags, warnings, and the recognized sections). It keeps natural-language casing intact - it's not a tag normalizer that mangles your prose, it reorders known ANIMA sections and leaves the rest of your phrasing alone.

Two more inputs tune artist detection, which is where corrector-style tools usually get it wrong:

• artist_overrides - comma- or newline-separated manual triggers to treat like artist tags. If your artist tag isn't recognized from the bundled list, this is how you tell the node.
• artist_exclusions - triggers that must not be treated as artists, for when something generic is being misclassified.

How it works

The mechanism is straightforward pattern matching against the ANIMA section vocabulary: quality/meta tags, character count, characters, series, artists, and general tags get sorted into their canonical positions; anything unrecognized is flagged as an unknown tag rather than silently dropped. Duplicate tags are reported. The report output is the reason to prefer this over a hand-rolled string sort - you can wire it to a text viewer, check what moved, and build trust in the output before it feeds your sampler.

How to use it

The classic setup is a feedback loop: paste a prompt in, read the report, fix the obvious problems, and let the node reorder the rest. Once you're happy, either feed corrected_prompt straight to a text encode or copy the corrected text back into your editor. For automation, the report JSON is parseable - you could flag prompts with too many unknown tags before they hit the queue.

If you don't need the report or the artist overrides - you just want a clean string out - grab the Simple version instead, which takes one prompt and returns only the corrected text.

Install and gotchas

Pack install:

cd ComfyUI/custom_nodes
git clone https://github.com/n0va39/ComfyUI-EasyUseAnima
cd ComfyUI-EasyUseAnima && pip install -r requirements.txt

Restart ComfyUI; Manager: "ComfyUI EasyUse Anima".

The realistic caveat: this node knows ANIMA's ordering conventions, not every LoRA's personal trigger rituals. A prompt that's "wrong" per the node may still be exactly what a specific checkpoint wants, and the corrector will happily reorder it anyway - the report exists precisely so you can eyeball that. Also, unknown tags are reported, not removed, so don't expect it to scrub your prompt. The translation dependency (googletrans-py) ships with the pack for optional prompt translation; it's only used if you enable it.