Sources & verification

Every text on this site comes from a named edition, pinned by checksum, with a plain statement of what was verified and how — and of what is not published, and why.

We author no religious content. We take published, established editions, pin the checksum of every file as we received it, and re-check on every build that what you are shown is that same file, character for character. Each source below then states what more was compared against an independent copy, and what could not be: rows no independent copy carries are counted here rather than passed over, and anything we could not stand behind is not shown at all — also counted, with its reason.

What is missing from this site is listed on this site. Held-back material is counted and explained rather than passed over in silence: a source that shows you what it withheld is making a claim you can test.

Quran text — the edition shown by default

Edition: The official Hafs text, version 2.2, published by the King Fahd Glorious Quran Printing Complex in Madinah. This is the text you read by default on every page.KFGQPC UthmanicHafs v2.2, retrieved 2026-07-18
Numbering: The standard Kufan numbering used for Hafs: 114 surahs and 6,236 ayahs, in mushaf order.
What was verified: Every one of the 6,236 ayahs, on every built page, in both languages, is checked codepoint for codepoint against the pinned source file on every build. Before that, the letter skeleton of every ayah is compared against the Tanzil text: the two agree letter for letter except at four places (2:72, 15:7, 27:20 and 36:22), each pinned individually in the build — one difference more or fewer and the build fails.
What is not here: We publish no translation of the Quran in any language. And the check proves that what we ship is what we were given; it does not adjudicate between the two editions, which is not ours to do.
Source file checksum (sha256): UthmanicHafs v22.json — b5f9c469702154ad8764150a874ee3353f014dd84e233d28d6db74e54226275a
Attribution terms: A published permission, not a licenceThe text — and the font it is drawn with — must be credited to the King Fahd Glorious Quran Printing Complex. What the Complex published is a permission for free use in websites and software, not a licence in the legal sense, so we do not advertise a licence over this text that we do not hold.https://github.com/nuqayah/qpc-fonts

Quran text — the alternative offered in settings

Edition: The Tanzil Project's Uthmani text, version 1.1, with Tanzil's default download options (pause marks, sajdah signs, rub-el-hizb signs and superscript alefs). It is baked into every page, so a reader switches to it from settings without a network request.Tanzil Uthmani 1.1 (marks, sajdah, rub, alef, tatweel), retrieved 2026-07-17
Numbering: The same numbering: 114 surahs and 6,236 ayahs.
What was verified: The source file is pinned by its checksum exactly as downloaded, including the copyright block the project wrote into it. All 6,236 ayahs are embedded in every built page and compared character for character against that file on every build.
What is not here: This text is not rendered in the Complex's font but in the face designed for it: a text and its font are a matched pair, and mixing them breaks how the marks are drawn.
Source file checksum (sha256): quran-uthmani.txt — 7f30c647331a61100ebf24a80507dc0fcdd9f2df97f1312b5b2dfcb982a7f326
Attribution terms: A stated licenceCreative Commons Attribution 3.0. It requires verbatim copying with no change to the text, and that the Tanzil Project be clearly credited wherever the text is used, with a link to tanzil.net.https://creativecommons.org/licenses/by/3.0/https://tanzil.net

Mushaf page layout

Edition: The QPC V2 layout: the page and line position of every word in the 15-line Madinah mushaf, 1421H printing, across 604 pages. Taken from quran.com's open API.QPC V2 layout via quran.com API v4, retrieved 2026-07-18
Numbering: Printed page 1 to 604, then the line number within the page.
What was verified: The layout is reconciled against the verified text layer itself, ayah by ayah rather than word by word.
What is not here: There is no pinned checksum for the layout file — the repository holds no checksum manifest for it, and we will not quote a number we do not have. The shapes a mushaf page is drawn with are display-only font codepoints, not text; the readable text of that page is the same verified Quran text as everywhere else.
Attribution terms: No stated termsWord and line positions are facts about a printed mushaf and are not subject to copyright. Attribution: layout via quran.com's open API; the fonts are the Complex's.https://quran.com

al-Tafsir al-Muyassar

Edition: al-Tafsir al-Muyassar, the committee tafsir of the King Fahd Glorious Quran Printing Complex, taken from the QuranEnc encyclopedia as 114 pinned per-surah files.QuranEnc arabic_moyassar, retrieved 2026-07-20
Numbering: One entry per ayah: 6,236 entries across 114 surahs.
What was verified: 6,236 entries of 6,236 are byte-identical to the pinned source. They were compared against two independent copies: the first covers 5,278 ayahs of the 6,236 — because it groups some ayahs under a single entry — and none of those disagree (5,277 are identical once spelling conventions are normalized, and one agrees at the 0.97 similarity bar rather than exactly); the second is an older 2011 edition which differs on 244 ayahs of the 6,236, every one of them listed individually.
What is not here: Those 244 differences are the older edition's own gaps — it drops passages and stops at earlier wording — not errors in what we show. And the tafsir has no pages of its own on this site yet: it appears when you open an ayah, and in the evidence search.
Checked against independent copies: quran.com API — ar-tafsir-muyassar — QUL resource-38 lineage, 114 pinned files
Tanzil ar.muyassar — 2011 snapshot
Attribution terms: A waqf declarationAttribution at the point of display: «التفسير الميسر — مجمع الملك فهد». What the encyclopedia declared is a free waqf for all Muslims, granting copying and republication — a waqf declaration, not a legal licence, so we advertise no licence over it.https://quranenc.com

Sahih al-Bukhari and Sahih Muslim

Edition: Open-Hadith-Data, the vowelled editions, at commit 1515f6cb — descending from the Harf / hadith.al-islam.com digitization.Open-Hadith-Data @1515f6cb (vowelled editions)
Numbering: Fath al-Bari numbering for al-Bukhari and Muhammad Fuad Abdul-Baqi numbering for Muslim (whose book 0 is the muqaddima, in its own sequence). Al-Bukhari holds 7,008 hadith in 97 books; Muslim holds 5,362 in 57 books.
What was verified: In al-Bukhari, 6,927 of the 6,988 hadith that can be compared agree with an independent digitization (99.13% of those that can be compared): 4,649 of them are character-identical once the two digitizations' spelling conventions are normalized, and the other 2,278 agree at the 0.97 similarity-or-coverage bar rather than exactly. In Muslim, 5,294 of 5,334 (99.25% of those that can be compared): 1,480 character-identical, 3,814 at that bar. Beyond that, a second, independently hosted copy of the same digitization matches 7,008 of 7,008 in al-Bukhari and 5,362 of 5,362 in Muslim.
What is not here: 61 hadith of al-Bukhari's 7,008, and 40 of Muslim's 5,362, read differently between the two historical digitizations — each is listed by its number. A further 20 in al-Bukhari and 28 in Muslim have no comparable copy in that lineage at all, so they could not be cross-checked. And no hadith on this site carries a grading of its chain, because the text we were given carries none.
Source file checksum (sha256): sahih_al-bukhari_ahadith_mushakkala_mufassala.utf8.csv — b03f661876fc4089520800b9231b57a1564baff734344ea8914494b86e2eb708sahih_muslim_ahadith_mushakkala_mufassala.utf8.csv — 79232c8d7c27fef161d8f33ca169f9754741cfe14957dde2ac76893374998943
Checked against independent copies: fawazahmed0/hadith-api (Arabic editions) — jsDelivr @1
ceefour/hadith-islamware — @d2cb0588
AhmedBaset/hadith-json — @70b83d6d
Attribution terms: A stated licenceODbL-1.0 for the database and DbCL-1.0 for its contents — and which of the two applies decides what you owe. Quoting individual hadith is a use of the contents, under DbCL, and carries no share-alike. Reusing the structured records as a database — the per-book files, the book structure, and the standard numbers recovered by alignment — is a use of the database: credit Open-Hadith-Data, offer any derived database under ODbL-1.0, and place no technical restriction on it. This site's own derived database is offered under ODbL-1.0 for exactly that reason.https://opendatacommons.org/licenses/odbl/1-0/https://github.com/mhashim6/Open-Hadith-Data

The Forty Hadith of an-Nawawi

Edition: The open hadith-api edition (ara-nawawi), pinned by checksum. These are the 42: the forty and the two an-Nawawi appended.fawazahmed0 ara-nawawi (jsDelivr @1), retrieved 2026-07-20
Numbering: The book's own sequence, 1 to 42. There is no external numbering to map to.
What was verified: Each of the 42 matns is re-derived from the pinned source byte for byte on every check, and each was traced to our own collections wherever it occurs in them.
What is not here: 5 of the 42 carry no citation from us, because their sources lie outside the collections we hold. For those, an-Nawawi's own takhrij line inside the matn remains the only attribution, and we add nothing to it that we have not verified.
Source file checksum (sha256): ara-nawawi.json — 91f559b90422005b768bb449426d87d2e6fabbb43ac3a31e5c24c50ca77465e6
Checked against independent copies: AhmedBaset/hadith-json — forties/nawawi40 — @70b83d6d
Attribution terms: No stated termsThe matn is classical and in the public domain, including an-Nawawi's own takhrij wording. The repository we took it from claims an open licence — a claim, not a clearance — so we advertise no licence over it.https://github.com/fawazahmed0/hadith-api

Ibn Rajab's eight additions

Edition: The eight hadith al-Hafiz Ibn Rajab al-Hanbali added in «جامع العلوم والحكم» to complete the classical fifty. They ship from the printed edition — al-'Abbad al-Badr's «فتح القوي المتين في شرح الأربعين وتتمة الخمسين» on turath.io (book 11325) — from eight individually checksum-pinned pages.turath.io book 11325 — «فتح القوي المتين», 8 pinned pages, retrieved 2026-07-21
Numbering: Numbers 43 to 50, kept separate from an-Nawawi's 42 and never blended into them.
What was verified: Each of the eight matns is re-derived from its pinned page byte for byte. All eight were found in our own collections, so all eight carry a citation with our own standard numbers — fuller coverage than the forty-two have.
What is not here: The source edition's own footnote numbers are not shipped; the citation you see is worked out by us. An auto-paginated, non-printed edition (book 147927) was the original source and was demoted to witness after a misreading was found in hadith 45 — what you read today is the printed text.
Checked against independent copies: turath.io book 147927 — «الأربعون النووية مع زيادات ابن رجب» — auto-paginated matn edition, 8 pinned pages
Attribution terms: No stated termsClassical public-domain matn. The source site publishes no terms, so we attribute no licence to it.https://turath.io

Adhkar — Hisn al-Muslim

Edition: «حصن المسلم» by Sa'id ibn Ali ibn Wahf al-Qahtani, from a pinned open dataset holding 134 categories and 298 texts.rn0x/hisn_almuslim_json @0405ee17 (2022-10-19)
Numbering: The book's own order of categories.
What was verified: 131 texts of the 298 are shown, across 68 categories of the 134. Every text shown is byte-identical to the source file, and none was shown until we found it in our own collections: Quran against the Complex's text, hadith against the two Sahihs and the collections we hold for checking.
What is not here: 167 texts of the 298 are held back, each with a stated reason: 63 did not reach the matching bar, 61 name no explicit authority in the book's own citation, 20 have a citation line the matcher does not accept and are left unclassified — five of them the book's own introduction paragraphs, which carry no citation line at all, 13 have no open collection we could check them against, 5 await comparison against a printed copy, 3 use wording common enough that no single narration can be pinned to it, and 2 were rejected on review. And not one citation reproduces the book's own footnote wording: every citation you see is our own phrasing, built from numbers we verified.
Source file checksum (sha256): hisn_almuslim.json — baaa876a36069a80b6eb22db41713cb5d2f60de63807029df36ff13ac3dbd072
Attribution terms: No stated termsThe dataset states no licence: there was no licence file at the commit we took it from, and the repository interface reports none. The adhkar texts themselves are Quran and Sunnah.https://github.com/rn0x/hisn_almuslim_json

Collections we hold for checking and never show

Edition: Five further collections — Jami' at-Tirmidhi, Sunan Abi Dawud, Sunan an-Nasa'i, Sunan Ibn Majah and Musnad Ahmad — from the same open hadith database at the same pinned commit.Open-Hadith-Data @1515f6cb (vowelled editions)
Numbering: Standard numbering where it could be derived honestly; rows where it could not stay unnumbered. Musnad Ahmad has no standard numbering at all.
What was verified: The four numbered Sunan are checked exactly as the two Sahihs are, and clear the same bar: 3,798 of at-Tirmidhi's 3,868 comparable rows agree with an independent digitization, 4,573 of Sunan Abi Dawud's 4,585, 5,657 of an-Nasa'i's 5,660, and 4,321 of Ibn Majah's 4,331. Musnad Ahmad is not held to that bar, and we do not claim it is: the other lineage carries no comparable copy of 24,971 of its 26,363 rows, so only 1,392 could be compared at all — 1,376 of those agree — and the rest are guarded only by a same-lineage copy that matches 26,363 of 26,363. All five have their own separate report.
What is not here: These collections are never shown and never linked from any page. That is enforced by the build itself: if anything in the published site so much as points at where they live, the build fails. We hold them so that a citation to at-Tirmidhi in the adhkar section can be checked against a text we actually have, rather than against memory.
Attribution terms: A stated licenceODbL-1.0 for the database and DbCL-1.0 for its contents, with the same distinction between them as for the two Sahihs.https://opendatacommons.org/licenses/odbl/1-0/https://github.com/mhashim6/Open-Hadith-Data

Ayah timings

Edition: Segment data from the Quranic Universal Library (QUL) by the Tarteel team, stored as retrieved, with checksums. This is the origin of the timings for 19 of the 26 published reciters; the other seven are the next section.QUL surah_segments preview API, harvested from 2026-07-17 — 19 of the 26 published reciters
Numbering: One timing per ayah, and per word where the data provides it.
What was verified: Every file records the exact recording its segments were made against, and the source responses are kept as received for audit.
What is not here: A timing set is only valid for the recording it was made for. Different sources carry different recordings of the same reciter, so a mismatch looks correct while being wrong — we therefore never pair a timing set with any other audio.
Attribution terms: No stated termsAttribution to the Quranic Universal Library (QUL) — Tarteel — wherever this data is used.https://qul.tarteel.ai

Ayah timings — generated here by forced alignment

Edition: For 7 of the 26 published reciters there is no upstream segment file at all: the ayah onsets are our own, produced by CTC forced alignment against the very audio bytes this site serves, using the open jonatasgrosman/wav2vec2-large-xlsr-53-arabic model. 114 surahs of 114 for each of the seven.CTC forced alignment, model jonatasgrosman/wav2vec2-large-xlsr-53-arabic — 7 of the 26 published reciters, 114/114 surahs each
Numbering: One timing per ayah and per word, across all 6,236 ayahs of each recitation.
What was verified: Every file records the model, the alignment parameters and the recording it was aligned against. The alignment was run on the bytes this site itself serves, so the bond between a timing and its recording is correct by construction rather than by trust. These are machine-produced onsets: they drive the listen-along highlight, and nothing else on this site rests on them.
What is not here: No such set is published for a reciter whose audio is not cleared, and no set is ever paired with a different recording. They are not offered as a reading of how the recitation should be performed, nor as a measurement of tajwid timing — they are onsets for a highlight.
Attribution terms: A published permission, not a licenceThese sets are this project's own work and fall under the standing permission this site gives for its own layer: quote them, republish them, build on them, provided maqraati.com is named as the source and an altered version is not presented as this site's. The model is a third party's, published openly; the recordings themselves grant no reuse right — see the About page.

How to quote this page

Quote the figure with its denominator exactly as written here — "6,927 of 6,988 hadith", not "99%" — name the edition and its version, and link to the source's own section. Every source on this page has a permanent link that does not change.

Recitations

The recordings themselves, and the terms on which they are published, are set out on the About page — the one meta page, where the whole rights statement lives. About