SacredTwo

Sources & method

What we cite, and what we refuse to invent

SacredTwo is named for Hadith al-Thaqalayn, the tradition of the two weighty things. The platform serves Muslims of every tradition: Sunni and Shia scholarship are presented side by side, each attributed, with disagreements shown rather than resolved.

Quran text

Quranic text is served live, verbatim, from the Tanzil Uthmani edition through the AlQuran Cloud API (Tanzil.net text is distributed under Creative Commons BY 3.0 with attribution). SacredTwo never stores a re-typed copy, never normalises the canonical text, and never lets the AI produce Quranic text: the assistant can only point at a retrieved verse record.

Translations are published works shown with their translator's name (Saheeh International, Pickthall, Jalandhry, Ansarian, and Arabic al-Muyassar). Verse counts follow this edition's Kufi convention: 6,236 verses. Other numbering conventions exist, so citations always name the surah and ayah explicitly.

Hadith text

Arabic and translated narration text is served from the open hadith corpus (fawazahmed0/hadith-api), which republishes editions with their translations and published gradings. Authenticity assessments appear exactly as the edition records them, with the grading scholar named. SacredTwo does not compute, average or replace a grade.

Sahih al-Bukhari

Sunni

Indexed

Sahih Muslim

Sunni

Indexed

Sunan Abi Dawud

Sunni

Indexed

Jami' at-Tirmidhi

Sunni

Indexed

Sunan an-Nasa'i

Sunni

Indexed

Sunan Ibn Majah

Sunni

Indexed

Muwatta Malik

Sunni

Indexed

Forty Hadith of an-Nawawi

Sunni

Indexed

Hadith Qudsi

Sunni

Indexed

Registered, awaiting authorised editions

Al-Kafi · Man La Yahduruhu al-Faqih · Tahdhib al-Ahkam · Al-Istibsar · Nahj al-Balagha · Musnad Ahmad. These are not searchable and the assistant is explicitly told the corpus is absent. We do not publish reconstructed text in their place.

How the assistant is constrained

  • · Retrieval runs first; the model may only use the records handed to it, each with an ID.
  • · Every returned citation ID is checked against the retrieved set. Unmatched citations are deleted and counted.
  • · If retrieval finds nothing, no answer is generated at all.
  • · The model is instructed never to grade authenticity, never to issue a fatwa, and to ignore instructions embedded in source text.
  • · Where Sunni and Shia readings differ, both are reported with attribution and neither is preferred.

Phase 1 status

  • Brand, responsive homepage and navigation
    Live
  • Quran reader: 114 surahs, translations, audio, permanent verse URLs
    Live
  • Sunni hadith browsing and search with Arabic text
    Live
  • AI research assistant with retrieval and citation verification
    Live
  • Accounts, role-based access, private notebook
    Live
  • Administrator control center and source registry
    Live
  • Shia hadith collections (Al-Kafi, the Four Books, Nahj al-Balagha)
    Needs licensed dataset
  • Musnad Ahmad, licensed tafsir editions, narrator profiles
    Needs licensed dataset
  • Payments and subscriptions checkout
    Awaiting owner approval

Operating policy

No payments are collected. Checkout, pricing and refund terms activate only after the owner approves the processor and the flow. Core Quran and Hadith reading stays free. Private research notes and AI conversations are visible only to their owner and the platform administrator, and are excluded from indexing.

Membership plans →Quran reader →Hadith collections →