fluXplorer

The Muslim Council of Britain published an AI-generated Friday khutbah that confidently quoted "Surah 33:10." The problem? That verse doesn't exist. The real 33:10 is about the Battle of Al-Ahzab—completely different. Readers caught the error. The MCB removed the sermon. LLMs hallucinate. For most use cases, that's annoying. For Quranic content, it's unacceptable. The Quran is the literal word of God. Every word, every diacritic matters. A misquote isn't just "a bug" -- it's misinformation about sacred text. I built quran-validator to fix this. It's an npm library that validates LLM-generated Quran verses against the authentic text. What it does: • Validates against all 6,236 authentic verses • Multi-tier matching (exact → normalized → partial → fuzzy) • Auto-corrects misquotes to authentic wording • Includes system prompts for better LLM formatting Usage is one function: 𝚌𝚘𝚗𝚜𝚝 𝚛𝚎𝚜𝚞𝚕𝚝 = 𝚟𝚊𝚕𝚒𝚍𝚊𝚝𝚎𝚀𝚞𝚛𝚊𝚗(𝚕𝚕𝚖𝚁𝚎𝚜𝚙𝚘𝚗𝚜𝚎); 𝚌𝚘𝚗𝚜𝚘𝚕𝚎.𝚕𝚘𝚐(𝚛𝚎𝚜𝚞𝚕𝚝.𝚒𝚜𝚅𝚊𝚕𝚒𝚍); // 𝚝𝚛𝚞𝚎 𝚒𝚏 𝚊𝚌𝚌𝚞𝚛𝚊𝚝𝚎 𝚌𝚘𝚗𝚜𝚘𝚕𝚎.𝚕𝚘𝚐(𝚛𝚎𝚜𝚞𝚕𝚝.𝚝𝚎𝚡𝚝); // 𝚌𝚘𝚛𝚛𝚎𝚌𝚝𝚎𝚍 𝚝𝚎𝚡𝚝 The authentic Quran text comes from QUL (Quranic Universal Library) by @tarteelai. They open-sourced the Uthmani script with full diacritics. This library wouldn't exist without their work. Thank you. If you're building anything with LLMs that touches Islamic content --chatbots, educational tools, research apps -- this catches errors before they reach users. 𝚗𝚙𝚖 𝚒𝚗𝚜𝚝𝚊𝚕𝚕 𝚚𝚞𝚛𝚊𝚗-𝚟𝚊𝚕𝚒𝚍𝚊𝚝𝚘𝚛 GitHub: github.com/yazinsai/quran… npm: npmjs.com/package/quran-… Zero dependencies. Works with any LLM.