# ADVERSARIAL PANEL REPORT — P30 Wing B Additions
**Window:** SP4 (Opus)  
**Scope:** D-B2-MAWQUF · D-B5-FIRST-LANGUAGE · D-B8-DOSSIERS (incl. dossier_browser.html) · D-D6-BENCH sittings log  
**Date:** 2026-09-03  
**Verdict:** FIX-FIRST (5 findings; 5 checks pass clean)

---

## METHODOLOGY

Five attack vectors per mission brief. Each finding anchored to exact `file:field:value`. SP3's scope (D-B1, D-B3, D-B4, strata.py) is not re-reviewed here; SP3's three fixes are confirmed applied in current files. A clean pass was earned on structural ordering, browser accuracy, one-ummah, and bench log. Five findings require correction before the Wing B2 additions are declared coherent.

---

## FINDINGS REQUIRING FIX

### FINDING 1 — B2 ERA STAMP ERROR: Ibn 'Umar/tabari refrain rows mislabeled "companion/mawquf"
**Severity:** HIGH  
**File:** `D-B2-MAWQUF/mawquf_layer.json`  
**Entries affected:** verse `55:15`/Ibn 'Umar/tabari (line ~1101) · verse `55:16`/Ibn 'Umar/tabari (line ~1144); also likely `55:13` and `55:14` via the same Tabari passage

**The law (mawquf_layer.json _meta):**  
> "era-stamped (G14); attribution sample-audited … the speaker rules the stamp (SP3 F1 law)"

**What the rows say:**  
Both rows show `"era_stamp": "companion"`, `"transmission_type": "mawquf"`.  
The `quote_src` for both reads verbatim:

> "…عن نافع، عن ابن عمر، قال: **إن رسول الله ﷺ قرأ سورة الرحمن**، أو قُرئت عنده، فقال: **'ما لِيَ أسْمَعُ الجنّ أحْسَنَ جَوَابا لِرَبِّها مِنْكُمْ؟'** … قال: **'ما أتَيْتُ على قَوْلِ اللهِ: ﴿فَبِأَيِّ آلاءِ رَبِّكُمَا تُكَذِّبَانِ﴾ ؟ إلا قالت الجنّ…'**"

Ibn 'Umar is the TRANSMITTER. The SPEAKER is the Messenger of Allah ﷺ. The report is marfu' via a Companion — exactly the class the layer stamps "prophetic/marfu_via_companion" elsewhere (see 55:15/Ibn 'Abbas/wasit at line ~1112, correctly stamped prophetic).

**Consequence:** The mawquf_layer claims 19 marfu_via rows. The Ibn 'Umar/tabari refrain hadith is an additional marfu row that was NOT counted. Downstream, the dossiers for 55:13 and 55:14 show **no prophetic-era entries** — but if these rows are correctly re-stamped, 55:14's dossier gains a prophetic entry, shifting its "oldest transmitted meaning" from companion to prophetic. This is a material change to the oldest-first architecture.

**Comparison revealing the inconsistency:**  
At 55:15 within the same file, two rows from `wasit` (Ibn 'Abbas and 'A'ishah) are correctly stamped `prophetic/marfu_via_companion` because they say "قال رسول الله - صلى الله عليه وسلم -". The Ibn 'Umar/tabari rows for the same verse say "قال رسول الله ﷺ" — the same structure — but are stamped `companion/mawquf`. The inconsistency is within the same 55:15 dossier entry.

**Required fix:** Re-examine all Ibn 'Umar/tabari rows for al-Rahman refrain verses (55:13 through the last refrain) where the quote_src contains the jinn-response hadith ("إن رسول الله ﷺ قرأ سورة الرحمن"). Re-stamp each as `era_stamp: "prophetic"`, `transmission_type: "marfu_via_companion"`, add `stamp_note` matching the law. Update the marfu_via count in `_meta`. Rebuild affected dossiers.

---

### FINDING 2 — B2 MISSING QUALIFICATION: Tabari's موضوع label absent from stamp
**Severity:** HIGH  
**File:** `D-B2-MAWQUF/mawquf_layer.json`  
**Entry:** verse `1:1` / companion `Ibn Mas'ud radiyallahu 'anhu` / edition `tabari` (line ~29)

**The law (mawquf_layer _meta):**  
> "every row verbatim-verified against its shard/aligned cell … differences preserved"

**What the row says:**  
`era_stamp: "prophetic"`, `transmission_type: "marfu_via_companion"`.  
The `quote_src` contains verbatim:

> "قال: قال رسول الله ﷺ: **'إن عيسى ابن مريم أسلمته أمُّه إلى الكتَّاب…'** [[الحديث ١٤٠- **هذا حديث موضوع، لا أصل له.**"

The **ultimate speaker is the Prophet ﷺ** — the era-stamp "prophetic" is therefore mechanically correct under SP3 F1. However, the `stamp_note` reads only: "marfu via the Companion - the ultimate attributed speaker is the Prophet ﷺ (the speaker rules the stamp, SP3 F1 law)." It says nothing about the fact that **Tabari himself, within the very quote_src, explicitly calls this hadith مَوضوع (fabricated) with لا أصل له (no basis).**

**Why this is a defect:** The verbatim-verified layer claims "differences preserved." Tabari's own fabrication judgment IS a difference — it is the source's own qualification of the transmission. Carrying the stamp "prophetic" without any flag that the source condemns the hadith presents an ʿilm al-hadith verdict without the source's own counter-verdict. A reader of the dossier who sees "prophetic" for this row has a materially incomplete picture.

**Note:** This is not a request to re-stamp the row (the stamp is mechanically correct: someone attributed this to the Prophet ﷺ). It is a request to make the source's own judgment visible.

**Required fix:** Add a `quality_flag` field to this row (or extend `stamp_note`) with text such as: `"Tabari footnote ١٤٠: 'هذا حديث موضوع، لا أصل له' — the source itself calls this transmission fabricated; carried verbatim, judgment preserved."` This flag should also cascade to any dossier entry that draws from this row.

---

### FINDING 3 — B5 POETRY-WITNESS MISCLASSIFICATION: majority of seed entries are not named-poet citations
**Severity:** HIGH  
**File:** `D-B5-FIRST-LANGUAGE/first_language_seed.json`

**The law (first_language_seed.json _meta):**  
> "revelation-era senses enter ONLY with a named early witness (**a poet the commentator cites, an early lexicon**)"

**Claim:** 30 entries, all marked `"marker": "poetry-witness (en)"`.

**Attack — verse by verse test against the law:**

| Verse | Edition | Named poet or lexicon in excerpt? | Assessment |
|---|---|---|---|
| 107:7 | zamakhshari | "the poet **al-Ra'i** used 'ma'un' to mean zakat" | ✅ PASSES — named poet, word meaning established |
| 47:18 | tabari | "as attested by **Jarir** and **Aws ibn Hajar** in their poetry" | ✅ PASSES — named poets, word meaning established |
| 106:1 | zamakhshari | "like poetic enjambment across two surahs; even Ubayy ibn Ka'b r…" | ❌ FAILS — structural device invoked, no named pre-Islamic poet cited to establish word meaning |
| 106:1 | miqbas | "this is like poetic enjambment (tadmin) across two surahs; al-Qurtubi says…" | ❌ FAILS — same device argument, no named poet |
| 17:44 | tafhim_en | "The entire universe bears witness that its Creator is free from every fault…" | ❌ FAILS — theological witness, not a lexical poetry-witness |
| 17:44 | ibnashur | "each creature is a standing declaration of His transcende…" | ❌ FAILS — creation-witness, no named poet |
| 33:45 | saadi | "confirmed by 'litakūnū shuhadā' 'ala al-nās…'" | ❌ FAILS — Quranic cross-reference, not a named poet |
| 33:45 | maarif_en | "Shahid means the Holy Prophet ﷺ will be a witness…" | ❌ FAILS — theological exposition, no named poet |
| 36:3 | saadi | "The Quran alone would suffice as proof and witness…" | ❌ FAILS — prophethood argument, no named poet |
| 36:8 | ibnkathir_ar | "a similar ellipsis occurs in Arabic poetry where mentioning one paired element implies the other" | ⚠️ BORDERLINE — poetic device correctly invoked, but no specific poet named by the commentator |
| 36:8 | tabari | "ellipsis by contextual implication, as in c…" (truncated) | ⚠️ BORDERLINE — same |
| 55:13 | tabrisi | "Ibn Amir read wa-l-habba dha al-asfi wa-l-rayhana with nasb…" | ❌ FAILS — qira'at variant discussion; no named poet, no word-meaning witness |
| 85:22 | baghawi/ibnashur/muyassar/tabari | All four: refutation of "this is poetry"; Quran is in al-Lawh al-Mahfuz | ❌ FAILS — theological defense of Quran against the label "poetry"; no named poet cited for word meaning |

**Result:** Only 2 of 9 distinct verses (107:7, 47:18) contain entries that satisfy the layer's own law. 7 of 9 distinct verses carry entries that are NOT poetry-witness passages in the technical mufradat sense.

**Required fix:** Three options — choose one and document in meta:  
1. **Filter:** Remove entries that fail the named-poet test; retain only 107:7 and 47:18.  
2. **Retitle and reclassify:** Rename the layer to "literary-witness seed" with a broader definition: "passages where classical commentators invoke poetry, poetic devices, or creation-witness to establish meaning." Document which entries are type-A (named poet for word meaning) vs. type-B (literary device invoked / thematic witness).  
3. **Declare the defect:** Add an `audit_note` to the meta: "Poetry-witness law compliance: 2/9 verses (107:7, 47:18) meet the named-poet standard; 7 verses carry broader literary-witness markers that do NOT name an early poet — the systematic harvest program is the correct resolution, not this seed."

---

### FINDING 4 — B5 DUPLICATE ENTRIES: 15 exact pairs in 30-entry seed
**Severity:** MEDIUM  
**File:** `D-B5-FIRST-LANGUAGE/first_language_seed.json`

**Evidence:**

The file states 30 `seed_entries`. Counting by verse+edition pairs and comparing truncated excerpts:

| Pair | Verse | Edition | Status |
|---|---|---|---|
| entries 1 & 3 | 106:1 | zamakhshari | identical excerpt |
| entries 2 & 4 | 106:1 | miqbas | identical excerpt |
| entries 5 & 6 | 107:7 | zamakhshari | identical excerpt |
| entries 7 & 9 | 17:44 | tafhim_en | identical excerpt |
| entries 10 & 12 | 33:45 | saadi | identical excerpt |
| entries 11 & 13 | 33:45 | maarif_en | identical excerpt |
| entries 14 & 15 | 36:3 | saadi | identical excerpt |
| entries 16 & 18 | 36:8 | ibnkathir_ar | identical excerpt |
| entries 17 & 19 | 36:8 | tabari | identical excerpt |
| entries 20 & 21 | 47:18 | tabari | identical excerpt |
| entries 23 & 27 | 85:22 | baghawi | identical excerpt |
| entries 24 & 28 | 85:22 | ibnashur | identical excerpt |
| entries 25 & 29 | 85:22 | muyassar | identical excerpt |
| entries 26 & 30 | 85:22 | tabari | identical excerpt |
| entry 8 | 17:44 | ibnashur | unique (no duplicate) |
| entry 22 | 55:13 | tabrisi | unique (no duplicate) |

**Count:** 28 of 30 entries form 14 duplicate pairs; 2 entries are unique. Effective distinct entries: **16**.

**The "30 seed entries" count is therefore misleading.** This appears to be a double-ingestion artifact from the consensus shelf pipeline.

**Required fix:** Deduplicate the seed. Remove the 14 redundant entries; the seed would contain 16 distinct entries. Restate the meta count accurately.

---

### FINDING 5 — B2 COMPANION ATTRIBUTION MISMATCH: safi rows at 1:1
**Severity:** MEDIUM  
**File:** `D-B2-MAWQUF/mawquf_layer.json`  
**Entries:** `1:1/'Ali ibn Abi Talib/shia_aligned/safi` · `1:1/Ibn al-Zubayr/shia_aligned/safi`

**The law (mission brief):** "the named Companion actually appears in the quote"

**Test — companion as speaker vs. subject:**

**Entry 'Ali/safi:** `companion: "'Ali ibn Abi Talib"`, `era_stamp: "companion"`, `class: "transmitted-explanation"`.  
The `quote_src` reads:  
> "…فلا يدخل الجنة إلا من أحبه من الأولين والآخرين فهو إذن قسيم الجنة والنار. قال: **المفضل بن عمر**. فقلت: له **يا ابن رسول الله** فرجت عني…فقلت: أسأل يا بن رسول الله فعلي بن أبي طالب (عليه السلام) يدخل محبه الجنة…"

'Ali ibn Abi Talib is the **subject** of the sentence ("'Ali ibn Abi Talib's lover enters paradise"). The **speaker** is an Imam being addressed as "son of the Prophet" (successor-era) talking to al-Mufaddal ibn 'Umar. 'Ali's name APPEARS but 'Ali is NOT the transmitter of the tafsir content.

**Entry Ibn al-Zubayr/safi:** `companion: "Ibn al-Zubayr"`, `era_stamp: "companion"`, `class: "transmitted-explanation"`.  
The `quote_src` contains:  
> "…الحجاج بن يوسف حين وضع المنجنيق على **ابن الزبير في الكعبة فقتله**…نزل القرآن على أربعة أرباع…"

Ibn al-Zubayr is a historical reference (killed by Hajjaj at the Ka'ba). The **speaker** of the Quran-in-quarters explanation is "أبي جعفر (عليه السلام)" — Imam al-Baqir, a successor-era figure. Ibn al-Zubayr appears but speaks nothing.

**Why this matters:** `class: "transmitted-explanation"` claims the Companion is TRANSMITTING an explanation. The quote_src does not support this for either row. Both rows should be:  
- either re-classed as `"mentioned-reference"` with `era_stamp: "successor"` (the actual speaker's era)  
- or removed, with a note that al-Safi carries no direct 'Ali or Ibn al-Zubayr explanation of 1:1

**Required fix:** For both rows: re-examine al-Safi's text around these passages. If no direct companion-voiced explanation of 1:1 exists in al-Safi, reclassify to `class: "successor-era-reference"` and re-stamp `era_stamp: "successor"`. Do not carry these as "companion-era transmitted-explanation."

---

## CHECKS THAT PASSED CLEAN

### CHECK A — Dossier era ordering: 5 additional dossiers spot-checked

SP3 verified 15 dossiers. I spot-checked 5 from the expanded 90-verse set (verses not in SP3's sample):

| Verse | Oldest stamp | Sequence | Verdict |
|---|---|---|---|
| 2:255 | prophetic (3 marfu via Companions) | prophetic(3) → companion(9) | ✅ |
| 55:13 | companion | companion(6) | ✅ (F1 note: would gain prophetic if Ibn 'Umar/tabari fixed) |
| 55:15 | prophetic (wasit marfu) | prophetic(2) → companion(2) | ✅ (internal to 55:15) |
| 9:100 | companion | companion(4) | ✅ |
| 49:6 | companion | companion(3) + sirah_moment | ✅ |

All 5 pass the era-ordering law. **Note:** If Finding 1 is fixed, 55:13 and 55:14 will gain prophetic-tier entries — the ordering law will remain satisfied since prophetic precedes companion.

**ERA ORDERING: PASS ✅**

---

### CHECK B — Browser vs. JSON: 3 dossier counts and shelf numbers verified

| Dossier | Browser count | JSON count | Match | Shelf (req/diff) | Browser shelf | Match |
|---|---|---|---|---|---|---|
| 1:1 | 22 | 3+14+4+1=22 ✅ | ✅ | 7/0 | "7 required elements · 0 differences preserved" | ✅ |
| 1:2 | 13 | 3+7+2+1=13 ✅ | ✅ | 7/0 | same | ✅ |
| 1:5 | 6 | 2+4=6 ✅ | ✅ | 10/0 | "10 required elements · 0 differences preserved" | ✅ |

All three browser-vs-JSON checks pass exactly.

**BROWSER ACCURACY: PASS ✅**

---

### CHECK C — One-ummah: no adjudication in B2 or B5

Four candidate sites examined:

**B2 / 33:33:** Layer carries 'A'ishah (wider-household reading), 'Ali (ahl al-bayt reading), Umar ibn al-Khattab, Umm Salamah, Ibn 'Abbas — all at companion tier without ranking. Neither reading is declared correct. `differences_preserved: 1` confirmed. ✅

**B2 / 55:20:** Tabari/Ibn 'Abbas (cosmological seas) and shia_aligned/nurthaq/Salman al-Farisi (allegorical 'Ali-Fatimah reading) appear side by side in the layer without adjudication. ✅

**B5:** Multiple commentators (zamakhshari, tabari, saadi, etc.) per verse; no verse carries a single line declaring one commentator's reading correct. ✅

**Dossier / 33:33:** Kitab-al-Ghayba-Numani:30 correctly marked `class: "deferred-to-panel"` — not adjudicated, not excluded, but flagged for panel attention. ✅

**ONE-UMMAH LAW: NO ADJUDICATION FOUND ✅**

---

### CHECK D — D-D6-BENCH sittings log: accurate and clean

The `bench_charter.json` sittings_log has 4 entries:

| Entry | Content | Status |
|---|---|---|
| WAVE74 launch | SP1+SP2+SP3 launched 2026-09-03 11:56 | ✅ accurate |
| SP3 | FIX-FIRST; 3 findings; all 3 fixed same-hour; clean checks listed | ✅ |
| SP2 | PILLAR-COHERENT; 5 verified counts exact | ✅ |
| SP1 | FIX-FIRST; 5 findings fixed same-hour | ✅ |

SP3's F3 fix ("retitle dossier meta") is confirmed applied: current `dossiers_v1.json` `_meta.layer` reads "D-B8 dossiers v1 - transmitted meanings oldest-first, assembled (panel SP3 F3: 'as it was revealed' is the aspiration…)" — the overclaim is gone. ✅

**No adjudication is recorded in any sitting.** Each sitting either declares PILLAR-COHERENT or FIX-FIRST with numbered findings. No sitting overrides a source's meaning. ✅

**BENCH LOG: PASS ✅**

---

### CHECK E — 19 marfu' rows: all 19 claimed rows contain explicit prophetic attribution

All 19 rows with `transmission_type: "marfu_via_companion"` or `"marfu_via_companion_shia"` were checked against their `quote_src` for explicit prophetic attribution language:

| Row | Verse | Companion | Quote attribution found |
|---|---|---|---|
| 1:1/Ibn Mas'ud/tabari | 1:1 | Ibn Mas'ud | "قال رسول الله ﷺ: إن عيسى ابن مريم…" ✅ (+ موضوع flag: see F2) |
| 1:7/Jabir/tabari | 1:7 | Jabir | "قال رسول الله ﷺ: قال الله عز وجل: قسمت الصلاة…" ✅ |
| 55:15/Ibn 'Abbas/wasit | 55:15 | Ibn 'Abbas | "وروى مسلم عن عائشة قالت: قال رسول الله - صلى الله عليه وسلم -" ✅ |
| 55:15/'A'ishah/wasit | 55:15 | 'A'ishah | same wasit passage, "قال رسول الله…" ✅ |
| 55:16/Ibn 'Abbas/wasit | 55:16 | Ibn 'Abbas | same wasit passage ✅ |
| 55:16/'A'ishah/wasit | 55:16 | 'A'ishah | same wasit passage ✅ |
| 2:255/Umar/wasit | 2:255 | Umar | "سمعت رسول الله صلى الله عليه وسلم يقول: أعظم آية…" ✅ |
| 1:1/Ubayy/jawami | 1:1 | Ubayy | "قال رسول الله (صلى الله عليه وآله): أيما مسلم قرأ فاتحة الكتاب…" ✅ |
| 1:7/Ubayy/tabrisi | 1:7 | Ubayy | "أبي بن كعب، عن النبي صلى الله عليه وآله وسلم قال: من قرأها…" ✅ |
| 2:255/Ubayy/tabrisi | 2:255 | Ubayy | "قال رسول الله: (يا أبا المنذر! أي آية في كتاب الله أعظم?…)" ✅ |
| 2:255/Ibn 'Umar/tabrisi | 2:255 | Ibn 'Umar | "وروى الثعلبي…عن عبد الله بن عمر، قال: قال النبي: من قرأ آية الكرسي…" ✅ |
| Remaining 8 marfu_via_shia rows | various | various | All contain explicit "قال رسول الله" or "عن النبي…قال" ✅ |

All 19 claimed rows truthfully attribute to the Prophet ﷺ. Finding 1 identifies ADDITIONAL marfu rows not in this count; the 19 are not wrong, they are incomplete.

**MARFU' ATTRIBUTION TRUTHFULNESS: ALL 19 PASS ✅**

---

## SUMMARY TABLE

| # | Attack | File | Verdict |
|---|---|---|---|
| F1 | Missed marfu' stamps: Ibn 'Umar/tabari al-Rahman refrain hadith | D-B2-MAWQUF/mawquf_layer.json | ❌ FIX — HIGH |
| F2 | موضوع label absent: 1:1/Ibn Mas'ud/tabari, Tabari himself calls it fabricated | D-B2-MAWQUF/mawquf_layer.json | ❌ FIX — HIGH |
| F3 | Poetry-witness misclassification: 7/9 distinct verses fail named-poet law | D-B5-FIRST-LANGUAGE/first_language_seed.json | ❌ FIX — HIGH |
| F4 | Duplicate entries: 14 exact pairs in 30 entries, effective seed = 16 | D-B5-FIRST-LANGUAGE/first_language_seed.json | ❌ FIX — MEDIUM |
| F5 | safi companion attribution: 'Ali and Ibn al-Zubayr are subjects not speakers | D-B2-MAWQUF/mawquf_layer.json | ❌ FIX — MEDIUM |
| A | 5 additional dossier era orderings | D-B8-DOSSIERS | ✅ PASS |
| B | Browser vs. JSON: 3 dossiers, counts + shelf numbers | D-B8-DOSSIERS + dossier_browser.html | ✅ PASS |
| C | One-ummah: B2 layer, dossiers 33:33 + 55:20, B5 layer | All Wing B2 layers | ✅ PASS |
| D | Bench log accuracy and no adjudication | D-D6-BENCH | ✅ PASS |
| E | 19 marfu' rows: all contain explicit prophetic attribution | D-B2-MAWQUF | ✅ PASS |

---

## PANEL VERDICT: FIX-FIRST

Five fixes required before the Wing B2 additions are declared PILLAR-COHERENT.

**Priority sequence:**

1. **F1 (ibnu 'Umar/tabari refrain rows)** — Re-stamp all Ibn 'Umar/tabari rows for al-Rahman refrain verses containing "إن رسول الله ﷺ قرأ سورة الرحمن…" as `prophetic/marfu_via_companion`. Update marfu_via count in `_meta`. Rebuild dossiers for affected verses. This is the most architecturally significant fix: it promotes the oldest stratum for 55:13/55:14 from companion to prophetic.

2. **F2 (موضوع flag)** — Add a `quality_flag` field to 1:1/Ibn Mas'ud/tabari row disclosing Tabari's own fabrication judgment. Cascade to any dossier note for this source.

3. **F3 (B5 poetry-witness)** — Choose and implement one of the three resolution paths (filter / retitle / declare). The law-compliant entries (107:7, 47:18) are sound; the others require honest reclassification.

4. **F4 (B5 duplicates)** — Deduplicate the seed; restate the entry count.

5. **F5 (safi attribution)** — Re-examine 'Ali/safi and Ibn al-Zubayr/safi entries at 1:1; reclassify if the companion is confirmed not the speaker.

The dossier architecture (era ordering, browser accuracy, one-ummah, bench log) is sound. The foundation holds; all five findings are correctible without structural redesign.

---
*SP4 adversarial panel — read-only — no layer or dossier file was modified.*