anti-slop-vision：結合提示詞整理與 VLM 評審準則的視覺品質關卡
anti-slop-vision: Vision quality gates: prompt hygiene plus a VLM judge…
anti-slop-vision 以提示詞整理和 VLM 評審準則設置視覺品質關卡，透過 SHIP / FIX / REGEN 判定，在交付前篩除視覺瑕疵和模型幻覺。工具採用 MIT 授權，優先使用標準程式庫，讀取指定資料，不會將資料傳送到其他地方。檢查規則可能誤判，適合找出需要改寫的內容，不應用來評分作者；若作品重點在於語氣，也不適用。
Most anti-slop-vision write-ups stop at the README: here is what the tool does, here is how you run it. This one covers what it refuses to do, and where it breaks.
What it is
Vision quality gates: prompt hygiene plus a VLM judge rubric with a SHIP / FIX / REGEN verdict — visual slop and hallucinations get cut before delivery.
In Russian: Vision-гейты качества: гигиена промпта плюс рубрика VLM-судьи с вердиктом SHIP / FIX / REGEN — визуальный слоп и галлюцинации отсекаются до отправки.
Install
запуск: python3 scripts/eval.py img/ --prompt "a red guitar" -m <vision-model>
Python. MIT licensed. Standard library first: the tool reads the data you point it at and sends nothing anywhere.
Where it does not fit
Every rule in a checker like this can produce a false positive. Use it to find candidates for a rewrite, not to grade a writer, and not where the voice is the point.
Repository
Source, issue tracker and documentation: https://gitlab.com/ameobius-ai/anti-slop-vision
Other tools from the same author: https://ameobius-ai.gitlab.io
來源：DEV Community · dev.to