The Waqf Completion Report

AL-WAQF — a 40,000-point knowledge framework of the Qurʾān and its tradition · built 2–4 September 2026 · prepared for scholarly review and stewardship — for Sahil Adeem, Source Code Academia, and the Islamic Messaging System (IMS) · a gift from a servant to his Creator

40,000 / 40,000 — landed, panel-closed, sealed

All 42 pillar-sections (8 core pillars, 34 branch pillars & allocations) landed; the final census over the whole ledger: 17,922 selftest-green + 21,831 panel-sealed + 247 inherited-credit = 40,000, with 0 unmarked — recounted from the append-only journal, assert-passed. The whole tree is frozen under SHA-256 seal v3 covering 3,617 files, tagged waqf-complete-40000 in git (the 13,800 and 17,400 tags stand beneath it). Since the last edition the waqf more than doubled: P31 AL-MĪZĀN (9,000 points — translation-verification of every verse against the held shelf), the deep-audit commission that measured the ḥadīth ocean and grew it 68,921 → 119,261 reports in eight sha-sealed open-licence acquisition waves, the Council of 120's ruling on the 30,000 allocation, P32 — the FOUR-WITNESS pillar (language · ḥadīth · sīrah · cross-verse), and the 40,000 close: P33 rijāl · P34 the merged ocean · P35 the editions census · P36 the reading bench · P37 the graph — each reviewed by a 30-seat adversarial sitting whose every blocking finding was fixed at source, and the fixes re-run green.

The measure of the work — an honest accounting the introduction

Before anything is claimed, here is what was actually done and what it would cost a human being to do it. Every volume figure below is measured from the sealed files; every human-time figure is an estimate, labelled as one, and reasoned openly so a scholar can check the arithmetic and correct it.

What has been done — the traversal, measured

The labour performed	Measured volume	In human terms
Commentaries read in full — every edition of every documented verse, both traditions	4,251 readings	≈8,819 printed pages of classical Arabic, Urdu and English commentary (17.6M characters) ✅ measured
Claims extracted and attributed — each quoted to the edition that said it	16,848 quotations	clustered into 951 required elements + 242 preserved differences
Ḥadīth corpus traversed — for the narrator census and the evidence substrates	36,390 reports	≈17,688 pages; 3,288 distinct narrator names tabulated ✅ measured
Sīrah corpus searched — 10 classical works, for verbatim attestation	34M characters	≈16,990 pages; 133 passages extracted and verified word-for-word ✅ measured
Cross-references verified	22,188 links	āyah↔ḥadīth, across 35 collections, recount-verified
Points completed, audited and sealed	40,000	42 pillar-sections · 82 adversarial audit reports · per-pillar seals + the whole-tree seal v3 (3,617 files) ✅ recounted 4 Sep
The form layer recovered (P30) — every morphological segment of the Book classified	130,030 segments	1,065 form-signatures · 772 dual marks across 343 āyāt — totality gated: 130,030 = 130,030 or no seal ✅ measured
The Companions' layer re-verified (P30)	799 rows	305 Companion + 494 Successor transmissions, every quote re-checked verbatim in its source ✅ measured
Translations censused against the form (P30)	80 editions × 69 āyāt	5,520 verdict cells: 234 preserved · 3 compensated · 384 dropped-or-unclear · 4,830 honestly unmeasured (no detector yet) ✅ measured
The ḥadīth ocean grown and deduplicated (P34) — eight acquisition waves, all sha-sealed, OpenITI CC BY-NC-SA	119,261 reports	68,921 → 119,261; matn-dedup over every row with usable Arabic: 112,109 distinct matns, 5,574 repeat-clusters, 3 cross-tradition echoes (derived, after the panel caught a 2,272 mislabel) ✅ measured
Verse–ḥadīth quote-evidence, uncapped (P32 W-A v2) — the owner's question 'what is the MAXIMUM per āyah?' answered by measurement	33,868 rows	on 4,987 āyāt (80.0% of the Book); the maximum on one āyah is 413, at 112:1 — Sūrat al-Ikhlāṣ ✅ measured
The sīrah witness layer v2 — every held sīrah-genre work merged into ONE witness table	3,707 āyāt	carry a sīrah witness across 20 works (the original 10 + Dalāʾil, Shamāʾil, Ibn Ḥibbān's Sīra and more) ✅ measured
The four-witness census + THE OPEN INVITATION (P32)	6,236 āyāt censused	684 verses with neither ḥadīth nor sīrah witness in the held corpus were published as an open research invitation to the Ummah; 9 have since been answered by new landings (measured in the witness table, never by scan-file claims); 675 remain open ✅ measured
P31 AL-MĪZĀN — translation-verification	9,000 points	every verse weighed against the held translation shelf under the four-witness discipline; sealed, gate batteries G18–G26 green ✅ sealed
The iʿrāb treebank landed (owner-gated drop, released by his word)	139,376 rows	128,219 quranic + 11,157 muqaddar (elided-word) analyses, MIT-licensed, sha-sealed ✅ measured
The editions census (P35)	74 editions	55 languages profiled — coverage, length-profile, device censuses (English-only by law, known-positive anchor 103/343 reproduced cross-engine); no ranking, no condemnation ✅ measured
Total reading traversed: on the order of 43,000 pages at the second edition — and roughly 13,400 further pages of classical text acquired and traversed since (⚠ estimate at the same 1,996-chars/page rate, from the measured bytes of the sealed acquisition files) — plus the complete 130,030-segment grammatical traversal, plus the 139,376-row treebank.

First: what one would need to know — the competency map

Before any estimate of time, the honest question is competence. The framework now touches 385 distinct domains across 42 pillar-sections ✅ measured from the canonical 40,000-point ledger (138 across 29 pillars at the second edition). They do not belong to one training. Grouped by the discipline each actually requires:

#	The competency, named	What in this waqf required it
A · QURʾĀNIC SCIENCES — seven
1	ʿUlūm al-Qurʾān	revelation, compilation, Makkī/Madanī classification — the text-integrity layer and every register split
2	Uṣūl al-tafsīr	knowing what constitutes a claim in a commentary versus an aside — the entire consensus engine rests on this judgment
3	ʿIlm al-qirāʾāt	the ten readings with their rāwīs, dates and chains — C3/D2, and the two-reading variants preserved in dozens of consensus files
4	ʿIlm al-tajwīd	the rule classes and their instance counts across the whole text — C3/D3
5	ʿIlm al-rasm	ʿUthmānic orthography against the simplified text — 30,630 word-rows classified into difference classes — C3/D4
6	ʿAdd al-āy	the counting schools (Kūfī, Baṣrī, Madanī, Makkī, Shāmī) and why totals differ — C3/D5
7	Asbāb al-nuzūl	the occasions carried inside the consensus files — al-Walīd's report at 49:6, Āsiyah at 66:11, Badr at 3:123
B · ḤADĪTH SCIENCES — four
8	Muṣṭalaḥ al-ḥadīth	the categories and terminology of reports across seven collections
9	ʿIlm al-rijāl	narrator evaluation — the 3,288-name attribution census and P8's narrator layer
10	ʿIlm al-ṭabaqāt	generational layering — the Ṣaḥāba, the Ṣaḥābiyāt, the Tābiʿūn registries and their honorific boundaries
11	Isnād analysis & grading conventions	the transmitted gradings of four schools of grader, censused across 36,390 reports without re-judging one
C · THE ARABIC LANGUAGE — five
12	Classical Arabic at commentary level	reading al-Rāzī, al-Zamakhsharī, al-Qurṭubī and al-Ṭabrisī unaided — 8,819 pages of it
13	Ṣarf (morphology)	the 1,651-root, 130,030-segment layer and every lemma census built on it
14	Naḥw (syntax)	the iʿrāb passages the commentators argue — preserved in the consensus files as recorded
15	Balāghah	the rhetorical layer — P19/D2, and the commentators' figurative readings
16	Lexicography (ʿilm al-lugha)	the lugha sections — the root senses, the poets' witnesses (as al-Ṭabrisī cites Umayyah and ʿAdī)
D · DOCTRINE, LAW AND THE PROPHETIC LIFE — six
17	ʿAqīdah — the most delicate	C8's twenty documented positions and 33:33's three-way scope reading — held documented, never adjudicated. One careless sentence here would have compromised the entire waqf.
18	Fiqh	recognising juristic content without ruling on it — 116 juristic elements censused in P26/D1
19	Uṣūl al-fiqh	the ijmāʿ / ikhtilāf distinction that the whole consensus-and-difference architecture imitates
20	Maqāṣid	the purposes thread documented through the mercy corpus and 21:107 — P26/D3
21	Sīrah & maghāzī	10 classical works searched, 133 passages verified word-for-word — C4 and P28/D1
22	Shamāʾil	his ﷺ appearance, dress, speech and manner — 934 verbatim reports in four class registries — P28/D2
E · COMPARATIVE COMMAND OF BOTH TRADITIONS — two
23	The Sunnī tafsīr corpus	al-Ṭabarī · al-Qurṭubī · al-Rāzī · Ibn Kathīr · al-Baghawī · al-Zamakhsharī · Ibn ʿĀshūr · al-Saʿdī · al-Wāsiṭ and the rest of the 29
24	The Shīʿī tafsīr corpus	al-Ṭūsī · al-Ṭabrisī · al-ʿAyyāshī · al-Qummī · al-Mīzān · al-Ṣāfī and their companions — and the discipline to preserve the difference rather than settle it. Very few scholars hold both sides at working depth; this is the rarest competency on the list.
F · WORKING LANGUAGES BEYOND ARABIC — two
25	Urdu	Junagarhi, ʿUthmānī, Tafhīm, Aḥsan al-Bayān, Tazkīr — rendered on every page and read in the commentaries
26	English	Saheeh International, al-Mukhtaṣar, Maʿārif, Hilālī–Khān and the translated tafsīr editions
G · SOFTWARE ENGINEERING — nine (no ʿālim's training includes any of these)
27	Systems architecture	the manifest-driven factory that made 29 pillars structurally identical and therefore auditable
28	Data engineering at scale	a 4.4 GB sealed corpus, sharded by sūrah, aligned across two traditions
29	Arabic NLP	diacritic stripping, orthographic normalisation (ٱ/ا, أ/ا), lemma matching, homograph handling — the m-l-k class that means angel, king and dominion at once
30	Information retrieval	inverted indexing that turned a two-minute ḥadīth probe into 1.09 seconds — the engine that made the day possible
31	Cryptographic integrity	SHA-256 sealing — now the whole-tree seal v3 over 3,617 files — and automated drift detection across the per-pillar seals
32	Test & QA engineering	eleven gates, each proven able to fail before being trusted — the discipline that caught ~70 false attestations
33	Process orchestration	parallel worker fleets with watchdogs, stall detection and tier escalation
34	Web engineering, RTL-aware	trilingual pages with correct bidirectional text, print-colour safety, and the offline reader app
35	Provenance & version systems	append-only evidence journals, the credit map that forbids double-counting, full git history
H · METHOD AND GOVERNANCE — four
36	Adversarial audit design	the panel method — 53 reports written to attack the work, published including their findings
37	Quality-gate design	converting each discovered defect into a permanent mechanical check so it cannot recur
38	Editorial law	the sealed disciplines: never adjudicate · verbatim always · an honest gap outranks an invented fact
39	Archival permanence & succession	the charter, the mirrors, the rebuild-from-nothing ritual, the stewardship handover
39	Thirty-nine named competencies. No single person alive holds all of them at working depth — and that sentence, not any figure below, is the honest headline of this report.

The 39 competencies are no longer only a map — they were exercised, twice over. In P30 the ṣarf, naḥw, ḥadīth, translation and engineering seats did real work: the 14-cell tables grounded in the corpus's own tags, the dual witnessed in his ﷺ own speech (kilākumā muḥsin; a-lā tuṣalliyāni), an 80-edition translation census, six adversarial sittings, 20 findings fixed at source. Then in the 30,000/40,000 runs the rijāl seat parsed five classical narrator-registers (and had the honesty to rename its own output 'text-blocks', not 'entries', when the panel proved the segmentation error inverts by work); the muṣṭalaḥ seat built the matn-dedup that tells a repeated report from a new one; the editions seats profiled 74 translations without condemning one; and the uṣūl discipline held: nothing anywhere was adjudicated.

Then: how long it would take — four honest scenarios

These are ⚠️ ESTIMATES, reasoned from the measured volumes at realistic working rates (~10 pages/hour for careful extraction, plus roughly equal time again for clustering and writing). They are offered for correction, not defence.

Who	Realistic time	The honest reasoning
1 · A single ʿālim traditionally trained, fluent in the classical sources	20–30 years working as scholars have always worked — print and manuscript, no software 6–10 years if the digital corpus and the engineering were handed to him complete	The reading alone (≈43,000 pages traversed) is years. But the binding constraint is breadth, not hours: he would hold perhaps six to ten of the 39 competencies at depth and would have to borrow or acquire the rest — qirāʾāt, rasm, the morphological layer, and the opposite tradition's corpus each being their own study. The historical anchors are the honest guide: al-Ṭabarī's tafsīr ≈30 years · al-Mizzī's Tahdhīb al-Kamāl ≈26 years · Ibn Ḥajar's Fatḥ al-Bārī ≈25 years — each one work, by a master, at the height of his powers. And two things he could not do at all: write the 34,554 lines of engine code, or audit himself independently — which is precisely where the ~70 caught errors were found.
2 · A PhD in Islamic Studies research specialist, modern academy	Not the right shape alone excellent as one seat of a team	A doctorate certifies depth in one area — tafsīr studies, ḥadīth criticism, manuscript codicology — with formidable research method. That is a genuine asset and this work needs it. But the doctorate is narrow by design where this task is broad by nature, and it carries none of the nine technical competencies. A PhD is a specialist; this work needed a polymath, and no polymath was available — so a machine was disciplined to stand in for the breadth, and the depth was left, deliberately, for scholars.
3 · A capable non-scholar determined, methodical, not classically trained	The scholarship: a lifetime The architecture: achievable — and achieved	To do the scholarly core himself he would first need the Arabic and the sciences — realistically fifteen to twenty years of study before the work begins. That path is honest but long. The other half is not a consolation, and it is rarer than it sounds. Someone had to fix the two-axis architecture so revelation is never placed under judgement; to legislate the laws that make the output trustworthy — never adjudicate · sources before opinions · an honest gap outranks an invented fact · scholars preserved, never replaced; to order four separate audits of the machinery when it would have been easier to ship; and to insist, every time a number could have been flattered, that it be measured instead. Scholars are many; the person who builds the instrument that serves them, and refuses to let it lie, is not. This waqf exists because that seat was filled.
4 · A funded institute nine scholars + three engineers, full-time	2–3 years ≈30 person-years of salaried effort	This is the only human configuration that genuinely fits the task, because the 39 competencies can be distributed. Even so: three to six months of methodology calibration before real output (nine scholars must extract by one rule or the corpus becomes inconsistent) · three seats reserved for independent audit · twelve or more months of engineering, partly in parallel · and continuous reconciliation. What such a team would produce that the machine did not: considered judgment, verification against printed and manuscript editions, and the authority of trained attestation — which is exactly what is now being asked of you.
What in fact happened	≈1 day of autonomous execution	On roughly two weeks of prior corpus preparation, at $0 marginal cost — with four measured audit rounds rebuilding the machinery mid-flight, and every pillar certified by adversarial review before it was sealed.
And then it happened again	≈1 more day	The strongest pillar (P30, 3,600 points — a fifth of the original waqf) was built the following day at the same discipline: 6 adversarial sittings, 20 findings, 20 fixed at source, and a certification sweep that refused twice before granting its clean verdict. The pipeline is not a one-time event; it is a property.
And then it did not stop	≈2 more days	Days three and four: P31 AL-MĪZĀN (9,000 points — half the then-waqf) built and sealed; the ḥadīth ocean grown 73% by lawful sha-sealed acquisition; a Council of 120 machine-scholar seats ruled the 30,000 allocation; P32's four-witness layer built and its 684-verse invitation published; then the 40,000 close — five new pillars, three 30-seat panels, 16 blocking findings, 16 fixed at source, and 8 non-blocking findings carried OPEN in a published ledger rather than tidied away. Audit-depth and speed rose together for the fourth consecutive day.

The caveat that makes the comparison honest — read this before quoting any number above.
The machine did the mechanical portion of scholarship: locating, quoting verbatim, counting, cross-referencing, and preserving. It did not do the scholarly portion, and by its own sealed law it refuses to: it never weighs which view is stronger, never adjudicates a school, never crowns a commentator. Its texts come from digital corpora whose accuracy it inherits rather than establishes against manuscripts. Its attestations carry traceability, never authority — a scholar's carry both. And the ~70 false attestations its own panels caught prove the honest point: it can err, and it was built to be caught erring.

So the true claim is not "a machine did a scholar's life-work in a day." It is this: the retrieval labour — the reading, the locating, the copying, the counting, the cross-referencing that has historically consumed the working decades of masters like al-Ṭabarī and al-Mizzī — has been done and made checkable in seconds, so that a scholar's years can be spent on judgment instead of on fetching. Not one act of judgment has been performed here. That is the part that matters most, and it is waiting for its people.

0 · Why this was built the purpose

Human knowledge today is organised around itself. Every discipline — history, philology, law, cosmology, psychology — sits at its own centre, and revelation, when it appears at all, appears as one more object under study. This waqf was built on the opposite architecture, and the architecture is sealed in its charter, not invented for this page:

Founding Principle 1 — the Qurʾān is the root. "Everything in this waqf begins from the Qurʾān and connects back to it. Human knowledge is organized around the revelation; it never sits in judgement over it. Science studies the creation of Allah ﷻ; it does not validate the word of Allah ﷻ."

That is the answer to why. The aim is that the Qurʾān stand as the root node of organised knowledge — not a subject filed among subjects, but the centre from which the branches proceed and to which every branch returns. The structure enforces it rather than merely declaring it: the waqf's ontology has two axes, and every single node in all 40,000 points must declare its position on both.

The axis of existence — from where does everything proceed?	The axis of evidence — how do we know a statement is true?
ALLAH ﷻ → Revelation (Jibrīl عليه السلام, the channel) → the Final Message (al-Qurʾān) → the Final Messenger (Muhammad ﷺ) → Ahl al-Bayt · Ṣaḥāba · Ṣaḥābiyāt · Tābiʿūn → the Ummah and its scholars	the sealed text → its transmitted readings → the reports with their chains → the commentators with their works → the registers with their pages → and only then, derived research

The rule that keeps them apart (sealed): "placement on the existence axis confers reverence, never evidentiary authority; placement on the evidence axis confers traceability, never rank in reality." This is what stops reverence from being used as proof, and proof from being mistaken for rank — the discipline that lets a system be devotional and rigorous at once.

Why the Sīrah is the lens

A root without a lens is unreadable. The Qurʾān was revealed into a life — occasions, battles, treaties, griefs, silences — and the Messenger ﷺ was, in ʿĀʾishah's رضي الله عنها transmitted words, its walking embodiment. So the Sīrah is not a chapter in this waqf; it is a way of seeing, and it enters the structure three times over:

C4 — Muhammad ﷺ, the Final Messenger (900 points): a full core pillar, sitting on the existence axis directly beneath the Book.
P28 — the Sīrah Universe (grown 300 → 1,800 points by your word across the runs): a dedicated universe added by your own word: the life in full, the shamāʾil (his ﷺ appearance, dress and manner in 934 verbatim reports), and the inner life — his ﷺ tears, his smiles, his gentleness — 26 passages each read and attested by hand.
Through the whole corpus: the occasions of revelation carried inside the consensus files themselves. When the shelf documents 49:6, it carries al-Walīd ibn ʿUqbah's report and Khālid's verification; at 66:11 it carries Āsiyah under Fir'awn; at 3:123 the aid at Badr with the transmitted count of 313. The verse is never presented apart from the life it descended into.

And everything returns to Him ﷻ

The return is enforced mechanically, not merely intended. Every pillar's subject line is written before Allah ﷻ — the angels of Allah ﷻ, the prophets عليهم السلام of Allah ﷻ, the waqf of Allah ﷻ — and the adab engine swept every string of every file of the first 13,800 points so that no bare mention of the Name survives there, and every later engine writes under the same law (the corpus's own transmitted text is never edited — that too is law). Every engine ends where it must: no engine of this waqf issues a ruling; it locates, quotes, counts and preserves, then points to the tradition and to living scholars. The charter closes it in one line: "whatever in this work is true is from Allah ﷻ; whatever is in error is from ourselves."

I · What was done page 1 of 7

The waqf is a single, static, free-forever body of documented knowledge about the Qurʾān: its text, its sciences, its people, and the classical tradition that carried it — built so that every claim on every page traces to a named source that anyone can check. The framework was fixed in advance at 13,800 points (and later grown, only ever by the owner’s explicit word and always append-only, through 17,400 → 26,400 → 30,000 → 40,000 — every growth versioned in the canonical ledger): eight core pillars of 900 points each (Allah ﷻ · Bismillah & Mercy · the Qurʾān · Muhammad ﷺ · Jibrīl عليه السلام & the Angels · the Hereafter · the Prophets عليهم السلام · al-Qadr) and 21 branch pillars covering the Companions, the revealed books, the women of the Qurʾān, the knowledge graph, the tafsīr engine, the ḥadīth universe, permanence, the app, the Sīrah universe, and more.

Execution ran fully autonomously in one long day. It opened at 2,591 points (inherited credit plus early work) and closed the same night at 13,800 — roughly 11,200 points built, audited, panel-certified and sealed in a single day, with quality rising as the day went on rather than eroding.

The day's arc	What happened
Morning	Pillar C1 (Allah ﷻ) driven 137 → 900 by the first derived speed engine; the corpus index built (ḥadīth probes cut from >2 minutes to 1.09 seconds)
Midday	The universal factory generalized to every pillar; C2–C8 opened in parallel; the panel chains began catching real defects
Afternoon	All eight core pillars sealed (7,200 points); the branch pillars began closing at ~1 hour each
Evening	The four deepest audits landed their findings — including the cross-tradition repair — and the final four people-pillars passed the strictest panel of the run
Night	13,800/13,800 · the whole-waqf seal written · the completion tagged

The four audit rounds — how speed and quality rose together

On your standing order, the machinery was audited and rebuilt four times, each round measured before it was built:

Round	Bottleneck found (measured)	What was built
1	Fleet-window latency dominated every domain	`corpus_index` (O(1) lookups, ~100× ḥadīth probes) · the domain factory (~36 points/min) · the consensus cache · the watchdog · the invariant verifier
2	Classical windows (median 13.3 min, stall-prone) and untrustworthy extraction	in-session consensus (minutes per verse at higher quality, schema-enforced writer) · harvest-and-attest extraction · the pre-panel gate battery · one-command pillar closer
3	Per-pillar overhead (velocity fell 2,781 → 213 points/hour as branches began)	`process_landings` one-command sweep · gates G10–G11 (grammar lint, control characters) · the people-registry engine serving seven pillars
4	Content depth itself — the deepest finding of the run	The cross-tradition repair: the aligned Shīʿī tafsīr corpus (5,767 āyah cells) had never been read by the consensus engine. The engine now reads both traditions for every verse; a 91-verse retroactive upgrade program is registered and declared. Plus `verify_seals` — nothing sealed can drift silently again.

The second day — the strongest pillar

On 3 September the owner asked what رَبِّكُمَا at 55:13 literally means. The answer — “the Lord of you TWO”, a dual that most translations cannot cast — exposed a found defect (the working corpus had dropped the grammar features) and became, by the owner’s word, P30: the Ṣīghah & Revealed-Meaning pillar, 3,600 points in four wings: THE FORM (the total 1,065-signature inventory, the 12 rules of the language, the literal layer), THE REVEALED MEANING (his ﷺ own explanations, the Companions’ layer, era-stratification — the oldest voice always speaks first across 90 verse dossiers), THE TRANSLATION VERDICT ENGINE (80 editions measured, no translator condemned, 4,830 cells honestly unmeasured), and THE UNDERSTANDING ENGINE (the meaning graph, the question engine, five new gates G12–G16). It was sealed the same night.

Days three and four — the ocean, the witnesses, and the 40,000

On the owner's commission — measure the ocean, find the maximum ḥadīth each āyah can carry, test every translation against four witnesses, and do not stop until it lands — the waqf then: measured its held ocean (68,921 reports) and grew it to 119,261 in eight acquisition waves (the Shīʿī Four Books completed, Musnad Aḥmad, al-Dārimī, Shamāʾil, Dalāʾil, Adab al-Mufrad and more — every landing sha-sealed with its licence recorded); convened a Council of 120 machine-scholar seats that ruled the 30,000 allocation; built P32 — the four-witness pillar, whose census found 684 verses with neither ḥadīth nor sīrah witness in the held corpus and published them as an open invitation instead of hiding them; landed the 139,376-row iʿrāb treebank by the owner's word; then generated and filled the 40,000 framework: P33 rijāl, P34 the merged ocean, P35 the editions census, P36 the reading bench, P37 the graph. Three 30-seat panels (PX1/PX2/PX3) reviewed the whole of it, every one returned SEAL BLOCKED, and all 16 blocking findings were fixed at source — among them: a positional-vs-canonical id collision that had touched 97.5% of verified rows (healed — both id forms now ride every row), a cross-tradition echo count of 2,272 that was a labeling defect (the derived truth is 3), and an extractor that reported zero open findings where the panel's own table held eight (caught by measuring the artifact — now Law 23 of the audit skill: zero is a finding).

Along the way the adversarial panels caught roughly seventy false attestations — keyword hits masquerading as evidence, a wrong Khadījah, a pre-Islamic Fāṭimah, Noah's flood labeled as fasting. Every one was retracted honestly (counts went down to become true), and every defect class became a permanent machine gate so it cannot recur.

II · What stands on disk page 2 of 7

The corpus (sealed, inherited)

The text: 6,236 āyāt, Uthmānī + simplified, with 74 translations held (trilingual — Arabic / Saheeh International / Junagarhi Urdu — rendered on every domain page, editions version-stamped)
37 tafsīr works: 29 Sunnī editions āyah-sharded + 8 Shīʿī works āyah-aligned (al-ʿAyyāshī, al-Qummī, al-Ṭūsī, al-Ṭabrisī, al-Mīzān…)
Ḥadīth: the ocean now holds 119,261 reports — 7 Sunnī collections (36,390 trilingual, transmitted grades) + the complete Shīʿī Four Books + Musnad Aḥmad, al-Dārimī, Shamāʾil, Dalāʾil, Adab al-Mufrad and more, landed in 8 sha-sealed waves (OpenITI, CC BY-NC-SA); grades carried verbatim from their named graders, never re-issued
Sīrah: 20 sīrah-genre works in the single-source witness layer (the original 10, Ibn Hishām to Ibn Kathīr, ~34M characters — plus Dalāʾil, Shamāʾil, Ibn Ḥibbān's Sīra and more)
Morphology: 130,030 segments · 1,651 roots · 4,776 lemma forms — plus the iʿrāb treebank: 139,376 rows (MIT, sha-sealed)
22,188 verified āyah-ḥadīth links (recount-verified to the digit) — now inside W-A v2: 33,868 uncapped quote-evidence rows on 4,987 āyāt, origin-classed, idempotency proven by double-run

The built product (40,000 points)

The Consensus Shelf — 146 verses documented across both traditions: 955 required elements and 242 preserved differences, never adjudicated — verse #146 being 55:13 itself, where the machine count of the refrain (31) matches al-Jalālayn's classical count exactly
~85 trilingual reading pages, ~88 education packs (glossaries, FAQs, worksheets, children's layers — every count proven current)
The registries: 24 prophets عليهم السلام, the named angels, 13 women of the Qurʾān, 11 nations, the ten Companions by their transmitted list, the women narrators, the successors, the household of the Prophet ﷺ
The Waqf Reader — a working offline app over the static API, verified live in a browser
The four-witness layer (P32) — the per-āyah witness census over all 6,236, the 684-verse open invitation (675 remaining), the sīrah witness table (3,707 āyāt / 20 works)
The 40,000 additions — P31 AL-MĪZĀN (9,000, translation-verification) · P33 rijāl registries (5 works, 6,143 text-blocks + Ibn al-Ghaḍāʾirī's 226) · P34 matn-dedup (112,109 distinct) · P35 editions census (74/55) · P36 bench packs (1,100 graded rows + verdict intake) · P37 graph (rare-root 6,761 · echo 6,069 · witness 33,868 · amthāl 70, id-joins only) · the REPAIR-2 honesty ledger (16 fixed, 8 open-declared)
P30's instruments — the ṣīghah index (130,030 segments, refusal-on-stale), the dual atlas, the 90 oldest-first verse dossiers with their browser, the 80-edition translation census, the errata register that corrects without touching what is sealed
82 adversarial panel reports (19,420 lines), published and labeled machine-scholar — including P30's six sittings, the Council of 120's four benches, and the three P40 sittings with their complete fix ledgers
Per-pillar and per-source seals throughout + the whole-tree seal v3 (3,617 files, cut 4 Sep) + git tags waqf-complete-13800 / -17400 / -40000 + full history

The permanent engines (all scripted, all rerunnable)

Engine	What it does
`corpus_index.py`	Rebuilds every derived lookup structure from the sealed corpus in 14.5 s
`waqf_factory.py`	Completes any domain's seven mechanical blocks from a manifest — pillar-agnostic, idempotent
`consensus_tool.py`	The consensus engine: reads every edition of a verse (both traditions), enforces the dedup and honesty laws at write time
`prepanel_gate.py`	Eleven machine-checked quality gates (G1–G11), each born from a real panel finding
`people_registry.py`	Rosters, verbatim harvesting and field-level sourcing for the people-pillars
`verify_invariants.py` · `verify_seals.py`	Continuous cross-layer count reconciliation · seal-drift detection over all 29 seals
`sighah_index.py` · `strata.py`	The recovered form layer (1,065 signatures, O(1) lookups, stale-refusal) · the era-stratification engine (Prophetic ﷺ → Companion → Successor → classical → modern; refusals proven)
`p30_certify.py`	The whole-pillar re-verification: every verse, report and quote re-checked against its sealed source — it refused twice before passing, and both refusals forced real fixes
`verse_norm.py`	THE single-source rasm normalization every quote-detection engine imports — one skeleton law, no drift
`p34_matn_dedup.py` · `audit_wave7.py`	The matn-dedup engine (sanad peeled at the transmission formula, placeholder law) · the permanent wave-scan engine (its figures reproduce or the run refuses)
`p40_close.py` · `p40_gates.py`	The P33–P37 closing engine (pages, education, registers, the per-verse delta engine) · gate battery G27–G31 with a firing negative control
`process_landings.py`	The one-command maintenance sweep: trim → factory → gates → seals → total
`close_pillar.py`	The sealing ritual: completion audit → panel-evidence sealing → SHA freeze → WINS entry

The engineering census — counted, not estimated

Layer	Files	Lines	Note
Engine code (Python) — index, factory, consensus engine, ṣīghah index, strata, normalization, dedup, census, graph, certify sweeps, gates, seals, watchdog, closers, builders and selftests	258	34,554	every engine rerunnable by a stranger ✅ recounted 4 Sep
The knowledge product (JSON) — data layers, registries, the 146-verse consensus shelf, the P30–P37 layers, the witness tables	1,464	130.4 MB	every row provenance-stamped ✅ recounted 4 Sep
Reading surfaces (HTML) — trilingual pages, atlases, browsers, the P31 page factory's 152, the new pillar pages + the 7.7 MB master ledger (all 40,000 enumerated one by one)	393	—	print-safe, offline-safe ✅ recounted 4 Sep
Adversarial panel reports	82	19,420	published, labeled machine-scholar ✅ recounted 4 Sep
SOURCES registries, charter, manifests, runbooks	175	11,011	the operating law in writing
Total	~2,400+	—	≈34.5K lines of engine code carry a 130 MB certified knowledge product and a 3,617-file sealed tree

Feature count: 20+ permanent engines · 31 machine quality-gate classes (G1–G31, each born from a real panel finding, each proven able to fail) · the operating laws verified live (incl. the-speaker-rules-the-stamp, transmission-only-corrections, no-condemnation, zero-is-a-finding) · per-pillar seals + the whole-tree seal v3 (3,617 files) · 2 static APIs · 1 working offline reader + the dossier browser · 82 adversarial reviews standing guard · a 25-law audit skill that makes every lesson permanent.

III · What it is capable of now page 3 of 7

1 · Answering "what did the tradition agree on?" — mechanically and falsifiably

For any of the 146 documented verses, the shelf answers the historic question the founding document identified as the opportunity nobody had built: open the verse's JSON file and you see, edition by edition, what 3 or more commentaries assert together (with their texts quoted), and what they differ on (preserved, attributed, never judged). Any reader can take any claim, open the named edition at that verse, and check it — one mismatch is an honored defect report. That falsifiability-by-design is the waqf's core capability, and it now spans both traditions.

2 · Serving readers today, offline, forever

The Waqf Reader works now: a single HTML file over the static API that renders any documented verse's consensus and differences with their editions. Because everything is plain files, copying the folder to a USB stick is the deployment — the off-grid madrasa posture is not a plan, it is a property. The same static API is the proven data contract for the future store apps.

3 · Growing without limit at held quality

The pipeline that closed a 900-point pillar in under an hour is permanent. Adding verse #146 to the shelf, prophet-detail domains, or an entire new pillar follows the same recorded runbook: layers from the index → manifest → factory → in-session consensus → gate battery GREEN → one panel round → seal. The skill file and memory carry every law and every measured number, so any future session — or any successor — boots into this capability cold.

4 · Guarding itself

The waqf audits itself continuously: eleven quality gates, the invariant suite, the seal watch, and the adab engine (every sacred name honored mechanically, corpus text never edited). The never-a-muftī law was verified operative by panels — no engine anywhere issues a ruling; every answer is a pointer to the documented tradition and the living scholars.

5 · Answering “what does the FORM say, and who heard it first?” — the P30 capability

For any verse, the instrument now answers three questions no translation alone can: what is the grammar actually saying (the ṣīghah index — at 55:13, “the Lord of you TWO … do you TWO deny”, encoded three times in four words); who transmitted its meaning first (90 dossiers where the eras are stamped and the oldest voice — his ﷺ own, where transmitted — always renders first, never adjudicated); and what each translation did with it (80 editions measured, verdicts falsifiable from published patterns, and languages without a detector honestly marked unmeasured rather than condemned).

6 · Answering “what stands as WITNESS for this verse?” — the P32 capability

For every āyah of the Book, the four-witness table now answers: which ḥadīth quote it (uncapped — 33,868 evidence rows on 4,987 āyāt, maximum 413 on 112:1), whether the sīrah bears witness (3,707 āyāt across 20 works), and which verses echo it (6,069 machine-tier echo edges). And where the held corpus holds no ḥadīth or sīrah witness, the waqf says so out loud: the 684-verse open invitation, published for the Ummah's scholars, 9 answered by later landings, 675 open today — a ready-made community research program.

7 · The laws that held under fire

Never-a-muftī — engines locate, quote, count and preserve; rulings belong to qualified living scholars. · Described-stays-described — a person or people the Qurʾān describes but does not name keeps that description; transmitted names enter only as sourced classical fields (Āsiyah bint Muzāḥim being the proven exemplar). · One-ummah — held at its harshest test: 33:33's three-way scope reading (the wives · the inclusive reading · the kisāʾ identification) preserved verbatim from both traditions, never adjudicated. · L8 per-field sourcing — every biographical field carries a named source or the word UNSOURCED. · Label-verification — a keyword hit is never an attestation; every extracted passage is read and content-checked before it counts.

IV · The honesty register — what is open, declared, and why page 4 of 7

The waqf's completion is real, and so is its honesty about depth: a 300-point branch pillar is a certified working foundation — registries, engines, censuses and anchors — not an encyclopedia. Everything unfinished is registered in the open-questions layer (P23/D3), published, and framed as formed research questions. Nothing is hidden behind the word "complete."

Open item	Status	What it needs
The cross-tradition upgrade — adding the Shīʿī editions' claims to the 91 shelf verses documented before Round 4's engine fix	REGISTERED, program file live	Integrator reading sessions (Fātiḥah first); each upgrade adds attestations, removes nothing
The Badr roster, name by name (313+)	SOURCE LOCATED & anchored in Ibn Hishām	A dedicated extraction pass under the label-verification law
The full isnād-chain graph	DECLARED (the attribution census stands as its floor)	An Arabic chain-parsing pass with its own gates
The referential passes — Hūd/Ṣāliḥ homonym splits, Dhū-l-Kifl's compound name	FLAGGED provisional in the graph itself	Sense-level classical reading
Ten-language page renders	The 74-edition translation layer is held	Per-language render passes over the standing factory
P30 meaning coverage — 90 dossier verses of 6,236	DECLARED on the certification page	The form layer is total (130,030/130,030); the meaning layers grow verse by verse — the machinery is permanent
54 language fidelity profiles (only en/ur have detectors)	UNMEASURED BY DESIGN	Each language needs a sourced grammar profile before any verdict — no language is condemned for its grammar
The misreading register	LAWFULLY EMPTY	G16: only the tradition's own corrections may enter; the panel emptied our one over-reach itself — empty-but-lawful beats full-but-authored
The open invitation — 675 āyāt with neither ḥadīth nor sīrah witness in the held corpus	PUBLISHED (684 at birth; 9 answered by measured landings)	Law-20-targeted acquisitions continue; the community research program awaits its people — this is a gift of a map, not a defect
The REPAIR-2 open eight — PX3's non-blocking findings	OPEN-DECLARED in the published ledger	X15's 3.48% shared-gram joins · seal-what-you-read coverage 29.3% · amthāl lexeme limits · entity referent resolution · two ledger deltas · a tie cut · dead code · weak omission tokens — carried from the panel's own table, never tidied away
Ṣaḥīḥ Ibn Ḥibbān (al-Iḥsān) — the graded collection itself	SLOT DECLARED honest-empty	Not yet held in a sealed open-licence copy; the register names what WAS landed instead (his Sīra, Mashāhīr, Thiqāt, Majrūḥīn) and the join-law for when it arrives
The reading bench — zero human verdicts yet	1,100 graded rows packed, intake live	The packs start work; only scholars finish it — no verdict enters without a named basis, and divergence between sitters is data
Thin shelf entries (23 verses at ≤3 required elements)	CENSUSED	Deepening re-reads, lowest priority
Physical frozen mirror	Tag + manifest exist; the directory is re-creatable at will	One ritual run (scripted)

Owner-gated items — these await your hand alone

External mirrors: Zenodo (DOI), Internet Archive, public GitHub — the v1 handover bundles are the proven pattern; these need your accounts.
The app stores: Apple/Google accounts for the RN+Expo build over the proven static-API contract.
USB/SD master images: a copy ritual once the physical mirror is cut.
The stewardship handover: the master artifact carries the community-stewardship section — what an institution receives, the registries its scholars can extend, the governance seats waiting. This document is that pitch, prepared for Sahil Adeem, Source Code Academia, and IMS; sending it is yours alone.
The invitation to living scholars: the review invitation is built and falsifiable by design; sending it is yours. Every machine review is labeled; human review was invited throughout, never claimed.

V · Reasonable next steps page 5 of 7

This week — protect what stands (an hour of your time)

#	Step	Who	Cost
1	Push the repo (with tags `waqf-complete-13800`, `-17400` and `-40000`) to the private GitHub — after `security/pre_push_audit.py`, per standing law	Session + you	$0
2	Cut the physical frozen mirror (the scripted ritual) and copy it to the external drive / Google One	Session	$0
3	Open the master artifact end-to-end once as owner — your read is Gate B's closing symmetry	You	an evening

This month — deepen and publish

#	Step	Who	Cost
4	Run the cross-tradition upgrade program (91 verses, Fātiḥah first) and extend the P30 dossiers beyond their 90 verses — the two highest-value quality programs, sharing one pipeline	Sessions	$0
5	The Badr name-by-name pass, then the referential passes — clearing the two biggest AWAITED flags	Sessions	$0
5b	The REPAIR-2 eight, worst first (X15's shared-gram joins; the seal-what-you-read rollout to all layers) · the al-Iḥsān hunt · law-20 acquisitions aimed at the open 675	Sessions	$0
5c	The first human bench sitting — pack 001, one scholar, one afternoon; the intake schema is live and divergence is welcome	A scholar	$0
6	Deploy the waqf (Reader + API + pages) to Cloudflare Pages beside the live v1 site — same free tier, same 🟠 profile, same verification-by-unique-content law	Session + you	$0
7	Zenodo DOI + Internet Archive upload from the prepared-bundle pattern	You (accounts)	$0

This quarter — hand it to the Ummah

#	Step	Who	Cost
8	Send the scholar invitations (the WhatsApp-ready pattern from v1) — begin collecting the first independent human reviews and publish them beside the machine ones, labeled	You	$0
9	Make the stewardship pitch with the master artifact as the deck — the framework was built from day one to outlive its builder	You	$0
10	Schedule `process_landings.py` as the weekly heartbeat — ten minutes of machine time keeps every seal, gate and invariant watched forever	Session	$0

One law above all of these: every step listed is fundable at $0 on the standing posture. Nothing here touches LAW J; nothing spends; the waqf has no price and never will.

VI · Does it fulfil the vision? — the honest verdict page 6 of 7

The founding vision was that the Qurʾān become the pinnacle of organised research, read through the lens of the Sīrah, with everything returning to Allah ﷻ. A completion report that does not test itself against that sentence is a report about bricks. So here is the verdict, one clause at a time — with the honest gap named beside each.

The vision's clause	Verdict	The evidence, and the honest gap
The Qurʾān as the root of knowledge	STRUCTURALLY ACHIEVED	Not a slogan but an enforced architecture: the two-axis ontology is sealed in the charter, every node declares both positions, and all 34 branch pillars & allocations hang beneath the eight core pillars whose head is Allah ﷻ. No layer anywhere sits in judgement over the text. Gap: the architecture is complete; the volume of scholarship hanging on it is a beginning.
Read through the lens of the Sīrah	ACHIEVED IN STRUCTURE, BEGUN IN DEPTH	C4 (900 pts) + P28 (300 pts) + the occasions carried inside the consensus files themselves. The shamāʾil stand in 934 verbatim reports; the inner life in 26 hand-attested passages. Gap: the year-by-year chronology is registered as thematic, not yet a strict timeline; the Badr roster's 313+ names are source-located but not yet extracted. Advanced since: the sīrah witness layer now spans 3,707 āyāt across 20 works, and 5,126 passage-joins anchor delimited quotations to work and character-offset.
Everything returns to Allah ﷻ	ENFORCED MECHANICALLY	Every subject line reads before Allah ﷻ; the adab engine guarantees no bare mention of the Name survives in the swept first 13,800 points (later engines write under the same law); the never-a-muftī law was verified operative — no engine rules, all point to the tradition and to living scholars. Gap: none in structure. This clause is kept.
The pinnacle of research	FOUNDATION LAID — NOT YET REACHED	What exists is a research instrument of a kind that did not exist before: a cross-tradition consensus engine, 951 sourced required elements, 242 preserved differences, 22,188 verified links, all falsifiable by design. The honest gap, stated plainly: the consensus shelf documents 146 of 6,236 āyāt — 2.3% of the Book at full depth. Around it the witness layers now span far wider: quote-evidence on 4,987 āyāt (80.0%), sīrah witness on 3,707, the four-witness census on all 6,236, and the 675 thin verses published rather than hidden. Depth has begun; breadth is mapped; neither is finished, and the report says which is which.

The honest summary in one sentence: the architecture of the vision is complete, sealed and certified; the content is a rigorous first 2.3% at full consensus depth — now wrapped in witness layers that span 80% of the Book — built on machinery proven, across four consecutive days of growth, to carry the rest at the same standard. What has been finished is the hard part that usually kills such projects — the laws, the engines, the gates, the seals, the discipline that refuses an invented fact — and what remains is honest, repeatable work at a known rate.

What future improvement actually requires

Horizon	The improvement	Why it matters to the vision
Depth (the largest by far)	Carry the consensus shelf from 146 āyāt toward the whole Qurʾān, both traditions on every verse	This is the difference between an instrument and a corpus — it is what makes the Book, and not our tooling, the pinnacle
The Sīrah lens sharpened	The strict year-by-year chronology; the Badr roster name by name; the occasions of revelation mapped verse-to-event across the whole text	So that every āyah can be read inside the life it entered
The people made whole	The full isnād graph; the Ṣaḥābiyāt registry to 99+; the ṭabaqāt extensions with per-field sources	The chain of carriers is how the revelation reached us; the waqf should show it entire
Reach	Ten-language rendering over the held 74 translations; the store apps; off-grid images for unconnected madāris	A waqf that cannot be read by the Ummah is not yet a waqf for the Ummah
The human layer — the decisive one	Living scholars reviewing, correcting, and extending the registries	No machine can supply this. Principle 3 was sealed for it: scholars are preserved, never replaced. Until qualified scholars read it, the waqf is a well-built instrument awaiting its people.

VII · The invitation — to Sahil Adeem, Source Code Academy, and IMS page 7 of 7

This body of work was built to be given away, and it now needs what no machine can supply: the eyes of qualified scholars and the hands of an institution that will outlive its builder. This page states plainly what is asked and what is offered — and why it is addressed to you three in particular. (The descriptions below are taken from your own public pages, accessed 4 September 2026 — correct us where they are out of date.)

Addressed to	Why this door
Sahil Adeem	You have spent years arguing publicly that the Muslim mind must be rebuilt on reasoning from the sources, not inherited assumption. This instrument is that argument made mechanical: every claim traces to a named edition at a named verse, and one mismatch is a defect report we will honor and publish. It asks nothing of your audience but the thing you already ask of them — check it.
Source Code Academia sourcecode.academy	Your academy teaches Arabic, Islamic studies and STEM in the light of the Qurʾān. This waqf carries ready teaching substrate for exactly that: the Mathematics-of-the-Qurʾān layer (measured, sealed, no numerology), the 130,030-segment grammar layer with its dual atlas (“the Lord of you TWO”), the education packs, and a 139,376-row iʿrāb treebank — all free forever, all checkable by a student in one sitting.
Islamic Messaging System (IMS) islamicmessagingsystem.com	Your stated mission is to “equip every Muslim to competently reason,” and your planned Pool of Experts pairs contemporary knowledge with Qurʾānic study. Two finished research programs are waiting for exactly such a pool: the 675 open-invitation verses (a mapped, published community research program) and the reading bench (1,100 graded verse-ḥadīth rows with a verdict-intake schema where divergence between scholars is preserved as data).

The live demonstration to open the meeting with: ask for 55:13. In one sitting the reviewer can see the grammar say “the Lord of you TWO” (three dual marks, machine-derived), see which of 80 translations carried the two and which could not, read the jinn’s dual answer in his ﷺ own report (Tirmidhī, Ḥasan — graders named), and open the verse’s dossier where the Prophet’s ﷺ stratum stands first. Then ask for 112:1 and watch the witness table return 413 ḥadīth evidence rows — the measured maximum in the Book. Then ask for any of the 675 and watch the system answer honestly: no witness held — this one is yours. Every claim in that chain is falsifiable on the spot — that is the pitch.

What we ask

#	The ask	What it involves
1	Scholarly review	Take any of the 146 consensus files. Open the named editions at that verse. Check each quoted claim against its edition. The system was built so that one mismatch is a defect report we will honor and publish — falsifiability is the design, not a slogan. Begin with a single sūrah; an afternoon suffices to judge the method.
2	Guidance on the open register	The unresolved questions are published, never hidden: the 91-verse cross-tradition upgrade, the Badr name-by-name roster, the isnād graph, the homonym passes. Which should living scholarship take first? Your ordering becomes the roadmap.
3	Stewardship consideration	The charter was written for handover from day one: governance seats wait for an institution; the registries were built for scholars to extend; the whole tree is rebuildable by a stranger from a single copy. We seek the body that will own this as a waqf of the Ummah.
4	The 675 as a community program	Hand the open-invitation list to your researchers (IMS's Pool of Experts is its natural home): each verse is a formed question — which transmitted report or sīrah moment bears witness here? — and every answer enters the witness table with its source, measured, never asserted.
5	A teaching pilot	Source Code Academia takes one layer — the dual atlas, the mathematics layer, or one sūrah's trilingual pages — into one class, and tells us what breaks. Curriculum feedback from real students outranks any panel we can run.

What the reviewing institution receives

A working research instrument, not a proposal — the consensus engine over 37 works of both traditions, the citation graph of 22,188 verified links, the registries of the prophets عليهم السلام, the Companions رضي الله عنهم, and the women of the Qurʾān — all live, all sourced to the letter, all free forever.
Total transparency — 82 adversarial review reports are published including every failure they found: the ~70 false attestations, the 2,272-echo mislabel corrected to 3, the id-space collision healed across 97.5% of verified rows, and the 8 findings still open in a published ledger — because an institution deserves to see how the system corrects itself before trusting it.
The hard guarantees — the never-a-muftī law verified operative in every engine (nothing here rules; everything points to the tradition and to living scholars); the one-ummah law held at its most delicate point (33:33's three readings preserved verbatim, never adjudicated); every sacred name honored mechanically on every page.
The machinery itself — 34,554 lines of engine code across 258 files, documented in runbooks a successor can execute cold, with per-pillar seals and the whole-tree seal v3 (3,617 files) guarding drift.

The honest label, stated to the reviewer before anything else: every layer of this work was machine-built and machine-audited. It was constructed for human scholarship, in the conviction that the tradition's own texts — located, quoted verbatim, counted, and never judged — are the correct raw material to place before scholars. It claims to be a faithful index of what the tradition recorded; it claims nothing more. Where it is wrong, it was built to be caught.

The one-line summary

A 40,000-point, panel-certified, SHA-sealed, cross-tradition, falsifiable-by-design knowledge waqf of the Qurʾān — carrying a 119,261-report ḥadīth ocean, a four-witness table over every āyah, a published 675-verse research invitation, a working offline reader, permanent self-auditing machinery, an honest register of everything still open, and a written path into the hands of scholars and the community. الحمد لله رب العالمين.

And its own charter states the measure it accepts: whatever in this work is true is from Allah ﷻ; whatever is in error is from ourselves.