{
  "id": 16236994,
  "title": "anti-slop-vision: Vision quality gates: prompt hygiene plus a VLM judge…",
  "url": "https://urgent.news/2026/10/23/anti-slop-vision-vision-quality-gates-prompt-hygiene-plus-a-vlm-judge",
  "topic": "tech",
  "section": "Tech",
  "published": "2026-10-23T22:24:25.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/ameobius/anti-slop-vision-vision-quality-gates-prompt-hygiene-plus-a-vlm-judge-4a34"
  },
  "original_language": "en",
  "account": "The anti-slop-vision tool stops after its README, detailing its functionality and usage. However, this report delves into its limitations and shortcomings. Vision quality gates involve prompt hygiene alongside a VLM judge's rubric, leading to a SHIP / FIX / REGEN verdict. The tool eliminates visual slop and hallucinations before delivery. In Russian, it's referred to as \"Vision-gates of quality: hygiene of prompts plus VLM-judge rubric with SHIP / FIX / REGEN verdict - visual slop and hallucinations are excluded before delivery.\"\n\nTo install and run the tool, use Python 3 with the command \"python3 scripts/eval.py img/ --prompt 'a red guitar' -m vision-model.\" It is licensed under MIT.\n\nThe tool reads the data it is given and does not send any information elsewhere. While every rule may produce false positives, it should be used to identify candidates for rewriting, not to grade writing, and not when voice is the main focus. The source code, issue tracker, and documentation can be found at https://gitlab.com/ameobius-ai/anti-slop-vision. The author also has other tools available at https://ameobius-ai.gitlab.io.",
  "summary": "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…",
  "key_points": [
    "Vision quality gates use prompt hygiene and VLM judge rubric",
    "Tool eliminates visual slop and hallucinations before delivery",
    "Install with Python 3 using command python3 scripts/eval.py"
  ],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}