REALMARC_FLAG_PATTERN_2026-07-27.md

supernatural/REALMARC_FLAG_PATTERN_2026-07-27.md

The hard-line-flag pattern for the realm road (Ch 70-77)

Axis owner: the flag-pattern lane of the realm-arc pattern wave (Josh ruling 2026-07-27, fourth

sitting — HL_0048's no-three-consecutive-peaks clause REMOVED for Ch 70-77; "we need a different

pattern for these mythological realm arc leading into the final end arc").

Status: ANALYSIS + RECOMMENDATION, proposal-tier. Read-only pass — no canon file, registry, or

gate was edited by this lane.

Headline: the road's current flag set is a fossil of the pre-re-architecture band. Both of its

TRUE values on the road have lost their original justification, and its FALSE values predate the

living-tradition content that now sits under them. The result is a flag column that is

*anti-correlated with care*: the road's only wholly-invented realm is flagged, and the road's

MAXIMUM-care living-tradition realm is not. The recommended fix is a **reallocation at constant

Josh cost** — 18 flags before, 18 flags after — not an inflation.

---

1. What canonical_hard_line_flag actually DOES

Established by tracing every consumer, not by reading the field name.

1.1 It has ZERO deterministic gate teeth (positive-controlled)

No harness gate reads the column. This is a real zero, not a matcher failure — controlled per the

repo's most-repeated defect class ("a search that CANNOT match reports zero"):

• CONTROL 1: grep -rlF "intensity" harness/*.py → hits check_chapter_topography.py,

rescope_grand_sage_voice.py, stress_test_registry.py.

• CONTROL 2: grep -rlF "mode_variety_check" harness/*.py → hits check_chapter_topography.py,

stress_test_registry.py.

• TEST: grep -rnF "canonical_hard_line_flag" harness/*.py → zero rows.

The same matcher, over the same files, hits two sibling columns of the same registry. The zero is

real. check_chapter_topography.py encodes HL_0048 as R3 and HL_0049 as R1/R2 by reading

intensity and dominant_combat_mode — never the flag. check_grand_sage_silence.py reads

HL_0046 by rule_id off T0_Hard_Lines, never the chapter flag.

**Consequence that governs the whole options analysis: the flag is not a gate. It is a

JOSH-ATTENTION ALLOCATOR and a routing bit.** Every cost it imposes is human-in-loop cost; every

protection it delivers is procedural, not mechanical. Nothing goes red if it is wrong. That is

exactly why it has silently rotted.

1.2 What it does change, mechanically and procedurally

Four real consumers, all downstream of authoring:

1. The spine_status state machine (BPC §9.7, affirmed verbatim in CASCADE_RUNBOOK §6). Seven

states for a non-hard-line node; eight for a flagged node, with canonical_question_review

inserted between pending and authoring. A flagged node literally cannot legally reach

authoring without passing an extra state.

2. /spine-chapter Step 2 (SKILL.md L63; CASCADE_RUNBOOK L84). "For a

canonical_hard_line_flag = TRUE node, pre-draft the SSBL Appendix A canonical-question set

WITH a researched proposed resolution per question, so Josh's pass is a decision batch, not a

research session. Non-hard-line nodes skip this step." This is the single highest-value

consumer: the flag is the *only* mechanism that routes a chapter's canonical questions to Josh

as a batch. Unflagged = the questions are never assembled.

3. Quest-row author_in_loop_required (Contracts §6.5, Decision 5A). `quest_type =

canonical_hard_line carries author_in_loop_required TRUE`; every other quest_type routes to

the Quest-Writer-agent-only track (§6.4) with no author in loop.

4. The §6.8 cultural-authenticity batch audit, historically keyed to a fixed chapter list —

ruled content-driven by Josh at CG-26 (§2.3 below).

1.3 THE THREE-CONCEPT CONFLATION — a live defect this wave must not inherit

Three different things in this repo are all called "hard line," they are used interchangeably in

prose, and their sets disagree:

The 18-flag set is not a function of hard_line_status. It equals: all 9 NARRATIVE_PROTECTED,

∪ 3 of 4 CULTURAL_PROTECTED (Ch 50 Amazon/El Dorado is CULTURAL_PROTECTED and UNFLAGGED — a

residual asymmetry outside this wave's scope but worth a row), ∪ six hard_line_status = NONE

nodes (Ch 8, 55, 66, 67, 68, 74).

This conflation is already causing live confusion: docs/proposals/topography_rows.json L946

carries an open [DECISION NEEDED - JOSH] because CVD §12.2 L662 ties astral mode to "Hard

Line chapter sanctuary preservation per Pillar 3" while CH_66 hard_line_status = NONE — even

though CH_66 *is* canonical_hard_line_flag TRUE and *is* mode astral. The author could not

tell which "hard line" the CVD meant.

Binding consequence for the options below: any option that says "flag ALL road nodes" must

first state which of the three concepts it is setting. If the concepts stay conflated, flagging

all nine realms would import Pillar 3's combat absolute ("no combat against cultural

representatives"; "sanctuary beats remain chapter peak") onto a road that is *ratified as fought* —

"each node EXPLORED and FOUGHT (a genuine ultimate boss, not a read-and-passed door)"

(AAA_REARCHITECTURE_PROPOSAL WS2) — and would collide head-on with HL_0121 ("routing around

fightable mythological material" is the named defect) and with the rated-R don't-defang-monsters

doctrine. It would also, read literally, drag nine realms out of Tier-C MMO scope per CVD L747.

That is the strongest mechanical argument against flag-all, and it is not a vibes objection.

---

2. The asymmetry: what actually happened

2.1 The road's flag column, against the charter's own care and generation tiers

Read the two right-hand columns against the flag column. The correlation is inverted. The one

generation_class = invented realm on the road is flagged. The three living/unsettled realms

in the middle band are not. The charter states the consequence in its own words about Ch 72: "per

the lens audit this is the highest-care realm on the road that has had the LEAST attention."

2.2 Why — the causal history, verified

This is not a judgement call anyone made. It is a column that was never revisited across a

re-architecture. Three verified movements:

(a) Ch 66-69 were flagged for a NAME-LOCK that has since been RETIRED. The four nodes were the

"occult-quartet," carrying the CVD §19.6 four-realm name-lock and the CVD §17.7 naming hard lines

— CASCADE_RUNBOOK L84 records it exactly: "Ch 67-69 questions use the CVD v1.4 names." The

flag was a *naming*-canon routing bit. AAA_REARCHITECTURE_PROPOSAL L1377: "The four-realm

occult-quartet name-lock recorded at §19.6 is replaced by the nine-realm Realm Road spanning Ch

66-74… The three scrubbed coinages are retired in favor of the real tradition-names" (Pohljar →

Pohjola; Tiamarath → the Apsu; Maatherion → the Duat). The lock the flags served is gone.

**(b) Ch 74 was flagged for a reveal that RE-HOMED to Ch 76 — and its own spine entry admits the

retro-fit.** CH_74.md's header, verbatim: "a hard-line node (the last Realm-Road otherworld), its

basis RE-BASED to a STRUCTURAL KEYSTONE … **NOT a reveal; the reveal content re-homes to Ch

76**." The charter concurs: Ch 74 "now leaks NOTHING — no Cassius, no House, no pattern, no

thesis, no Grand-Sage hint, not even an invoke-and-negate." Meanwhile CH_76's registry row reads

"the eighteen-node set: the patriarch reveal and the father encounter." The reason moved; the

flag did not. What Ch 74 got instead was a *structural* rationale ("the last Realm-Road

otherworld") that, applied consistently, would flag Ch 70-73 too — and does not.

(c) Ch 70-73 are UNFLAGGED because when the column was populated they were not these places.

AAA_REARCHITECTURE_PROPOSAL WS2 L102, verbatim: "The built Ch 70-72 fairy content (return +

faceless Realm War Phases 1-2) → folds into the compressed fairy realm Ch 75-77 (WS5). **Ch 70-72

are repurposed as Realm Road nodes** — never added." L1375 adds "the faceless fairy war previously

carried at Ch 70-73." So at flag-population time Ch 70-73 were *invented fairy content with no

living tradition to misrepresent* — correctly unflagged on care grounds. They are now **Hades,

Xibalba, Naraka/Meru, and Annwn**: three living-tradition cosmologies plus one unsettled. The

content changed completely; the flag column did not move.

The asymmetry is therefore NOT deliberate. Option D below ("document it as deliberate") cannot

be honestly executed — there is no defensible design intent to document, and writing one would be

minting a rationalisation into canon, which is the exact defect class the repo has already caught

("a false 'canon doesn't exist' written into the repo").

2.3 The governing precedent is already ruled: CG-26

Josh ruled this class of question on 2026-07-21, and the ruling is directly on point

(BUILD_PLAN_END_TO_END L394, verbatim): *"elevated-care audit routing is **content-driven, not

fixed-chapter-list-driven — but calibrated to the LIGHTER batch-audit tier** outside the fixed

eight, explicitly to prevent care-tier inflation ('this is a rated M action adventure fantasy

RPG… do not go overboard')."*

Both halves bind here. The road's flag set is a fixed list that content has outrun — precisely the

failure CG-26's first half outlaws. And the answer is not "flag everything" — precisely what

CG-26's second half outlaws. CG-26's shape, applied to the road, nearly deduces the answer.

2.4 A second live incoherence the flag fix also closes

The asset lane and the spine lane currently disagree about the same nodes. The charter's §3.3

costume tooth in read_boards.py reads generation_class, never care_tier: for any realm whose

class is living or unsettled, every Panel A image must carry a rights value and Panel A is Lane

R only — no raw generation. Under today's data, Xibalba, Naraka/Meru and Annwn are treated

strictly by the asset gate and as standard nodes by the spine/quest routing. One node,

two lanes, opposite verdicts. Any flag pattern that keys to generation_class closes that seam by

construction.

---

3. The options

Evaluated per: what it costs, what it protects, the care-doctrine read at the *bolder* calibration

(flags are protection where protection is due; flag inflation is the named defect; the §17 floor is

the only absolute), and the factory implication.

Option A — Flag ALL road nodes (Ch 66-77)

*"The whole ascent is hard-line ground: every realm is a mythological tradition's own cosmology."*

• Protects: every living-tradition cosmology on the road, with no judgement call and no list to

maintain. Simple, auditable, impossible to under-apply. Closes Ch 71/72/73 completely.

• Costs: +5 full Stage-2 canonical-question batches to Josh (70, 71, 72, 73, 75) on top of 7 —

a 71% increase in road Josh-batch load, landing squarely in the pre-ship window. Every road quest

leaves the agent-only track (§6.4) onto author-in-loop, which is the load the two-track split

exists to bound. And it flags Atlantis and the fairy nodes — *invented* places with no living

tradition to protect — which is protection billed where none is due.

• The mechanical danger (§1.3): if the three "hard line" concepts stay conflated, this is the

option that can import Pillar 3's combat absolute and the CVD L747 Tier-C exclusion onto nine

realms ratified as fought. Collides with HL_0121 and the don't-defang-monsters bar.

• Care read (bolder calibration): REJECT. This is the textbook named defect — care-tier

inflation, "do not go overboard," CG-26's second half. A flag that fires on every node carries no

information, and a signal that never varies stops being read: the Stage-2 batches degrade into

ceremony, which *weakens* protection for Ch 72 rather than strengthening it. The timid option is

usually the wrong one, and this is the timid option.

• Factory: all twelve nodes take the 8-state machine, author-in-loop quests, §6.8 audit, and

Lane-R-only Panel A. Uniform, and uniformly expensive; the strict asset path is applied to

Atlantis, where it buys nothing (nothing to misrepresent).

Option B — Flag by TRADITION CLASS (coordinate with generation_class)

*Living/unsettled traditions flagged; invented and historical not.*

• Protects: exactly the realms where a real people's own picture of their own cosmos can be got

wrong at AAA fidelity — Xibalba, Naraka/Meru, Annwn, the Duat, Yggdrasil. Keys the flag to a

factual question ("does this tradition have practitioners today"), which the charter already

ruled is not a care judgement and not reviewer memory but L-R3's named output. Un-flagged nodes

are un-flagged for a *stated, checkable reason*, not by omission.

• Costs: depends on L-R3 landing to settle historical vs living for Pohjola, Hades,

Yggdrasil and Annwn; until then unsettled defaults strict, so the interim set is slightly wider

than the settled one. Requires the proposed generation_class column to actually mint (a Wave-2

item, DIRECTOR call per charter §6.3 — schema, not world canon, so it is not blocked on Josh).

• The gap: a pure tradition-class read under-protects the reveal nodes. Ch 76 and 77 are

invented and carry the arc's single most delicate reveal ceiling and its one sanctioned

surfacing; Ch 75 carries "the arc's most delicate REVEAL-RESTRAINT." Tradition class says

"invented, don't flag." That is wrong, and it is wrong for a reason worth naming: **the flag is

being asked to carry two different jobs.**

• Care read: correct in direction, incomplete in coverage. Adopt its *keying*, not its *scope*.
• Factory: the spine lane and the read_boards.py costume tooth finally agree on the same

node — the §2.4 seam closes by construction.

Option C — Flag by FUNCTION (nodes carrying irreversible story events)

• Protects: the reveal spine — Ch 76 (patriarch + father), Ch 77 (the thesis surfacing),

Ch 75 (the restraint), Ch 69 (the Architect's terminus). Consistent with why 8 of the 18 arc-wide

flags exist at all (Prologue, 1, 8, 38, 55, 58, 65, Epilogue are all narrative anchors).

• Costs: says nothing about cultural exposure. Under a pure function read, Ch 72 — MAXIMUM

care, ~living Hindu/Buddhist/Jain cosmology — stays unflagged, because Vritra's fork is a

*reversible* character beat, not an irreversible story event. That is the current defect

re-derived from a different principle. "Irreversible" is also the softest criterion of the four:

it needs its own definition before it can route anything, and a soft criterion is what let the

current set rot.

• Care read: necessary, insufficient. Same verdict shape as Option B, mirrored.
• Factory: routes reveal-discipline critics and the §17.9 re-verify; does nothing for the

rights/Lane-R asset path.

Option D — Keep the current set, document the asymmetry as deliberate

• Protects: nothing new. Zero cost, zero change, and it is the honest baseline every option

must beat.

• Fatal problem: §2.2 shows the asymmetry is not deliberate. It is a stale column across

the Realm-Road re-architecture, with the receipts in three artifacts (the retired §19.6 name-lock;

CH_74's own "RE-BASED… NOT a reveal" header; WS2 L102's "Ch 70-72 are repurposed"). Documenting

a fossil as an intention writes a false rationale into canon — the exact defect class this repo

has caught before and gated against.

• Care read: REJECT, and reject specifically because the do-nothing option here is *not* the

conservative one. Leaving Naraka/Meru and Xibalba on the agent-only quest track with no Stage-2

batch and no §6.8 audit is a live §17.1 exposure on two living traditions, one of them a major

world religion. Josh's own framing applies in the other direction than usual: erring toward

omission is the more costly error here, because the omission is *protection*, not content.

---

4. RECOMMENDATION — Option B+C, split by job, at constant Josh cost

Split the flag's two jobs and key each to the axis that already exists. This is not a new idea;

it is the charter's own C-4 fix applied one column over. The charter closed C-4 — Atlantis being

simultaneously generation-legal and elevated-duty — as a schema defect, not a disagreement, by

splitting care_tier from generation_class, and stated the principle: *"Two fields, two jobs, one

tooth each, and neither column asked to mean two things."* canonical_hard_line_flag is the next

column asked to mean two things: cultural exposure and narrative irreversibility. Same

defect, same fix.

4.1 The proposed road pattern

Two flag *reasons*, one flag *column* (no schema migration required — the reason rides the existing

spine-header prose and the extension rail, exactly as CH_74's header already carries its basis):

• CULTURAL basis → fires when generation_class ∈ {living, unsettled}. Factual, checkable,

L-R3-settled, not reviewer memory.

• NARRATIVE basis → fires on a node carrying an irreversible reveal event (the §17.9 ceiling

nodes).

• LIGHT tier (CG-26's calibrated second half) → generation_class = historical, or a node

carrying reveal *restraint* without the reveal itself. Gets the §6.8 batch cultural audit and the

reveal re-verify critic, not the full Stage-2 Josh batch and not author-in-loop quests.

4.2 Why this is the recommendation

• It costs Josh nothing. Road full-tier flags: **7 before (66,67,68,69,74,76,77), 7 after

(69,71,72,73,74,76,77). Arc-wide total: 18 before, 18 after.** This is a *reallocation*, not

an inflation — the single most important property, because it satisfies the bolder calibration

and CG-26's anti-inflation half simultaneously and provably. The attention moves off Atlantis

(nothing to misrepresent) and the Apsu (no practitioners) and onto Xibalba, Naraka/Meru and Annwn.

• It answers the strongest objection to un-flagging. Un-flagging Atlantis, Pohjola and the Apsu

does not weaken their anti-fringe firewalls, because those firewalls were never carried by

this flag. They are carried by care_tier (which the charter ratified as the critic-routing and

review-depth driver) and by harness/check_layer_register.py, which runs inside run_gates.py

with 0 TIER-A as a hard exit condition. Those are *teeth*; the flag is not. The protection

that actually holds Donnelly/Blavatsky/Ariosophy off the page is untouched.

• It keys to a fact, not a vibe. generation_class answers "does this tradition have

practitioners today" — the charter explicitly ruled that assignment "is not a care judgement and

not reviewer memory." A flag keyed to a fact cannot rot the way a fixed list rots. That is the

structural repair, and it is the reason this fix will still be correct after the next

re-architecture.

• It closes the §2.4 two-lane disagreement between the asset gate and the spine routing at zero

extra cost.

• It repairs Ch 74 rather than changing it. The flag stays; only its stale basis is replaced

with a true one. No visible churn, no re-cascade.

4.3 The guardrail this wave must state out loud

THE FLAG IS NOT A PACING INSTRUMENT. With HL_0048's peak clause lifted for Ch 70-77, the road

is free to ascend continuously, and the flag pattern must not quietly re-impose a valley. Three

specific traps:

• Pillar 3's "transformation/sanctuary/learning beats remain chapter peak" and HL_0050's relief

language can be misread as making a flagged node *calmer*. On the final ascent it must not.

Ch 71 and Ch 72 are both PEAK today and stay PEAK under this recommendation — they gain a

Josh batch and a rights-strict asset path, not a lower intensity ordinal.

• A flagged node is not a less-fought node. HL_0121 governs: Vritra is fought, the two Lords of

Xibalba are fought, Níðhöggr is fought. The flag routes *review*, never *combat presence*. Any

drafting of this pattern that reads "flagged → sanctuary" has re-imported the confusion of §1.3.

• The mode-variety clause (HL_0049, both clauses) is untouched by everything here and still binds

all 79 nodes.

The flag pattern is orthogonal to the intensity string the pacing lane designs. It should be

authored so that swapping either one does not perturb the other — the same separation-of-instruments

discipline the charter applied to power/DEMAND/awe at §1.4.

4.4 Strongest objection to the recommendation

*Splitting the flag's basis without splitting the column leaves the two jobs sharing one boolean, so

a future reader still cannot tell why a node is flagged without reading prose — which is how the

current set rotted in the first place.* This is a fair hit. The honest answer is that a second

column is the cleaner fix and should be the Wave-2 schema proposal (a hard_line_basis enum:

cultural / narrative / both / light, riding the registry-extension rail beside

generation_class, DIRECTOR-callable as schema rather than world canon). The single-column version

recommended here is the *no-migration* form that can land immediately; it should be explicitly

recorded as an interim, with the enum as the durable form. Landing the interim without recording

that intent would reproduce the defect one generation later.

A second, weaker objection: L-R3 has not landed, so generation_class is partly unsettled and the

interim set is wider than the settled one. That is by design — the charter ruled the asymmetry

argument ("a wrongly restrictive call costs a wave of reference assembly, a wrongly permissive one

ships a model's guess about a living people's cosmology") — and Annwn is the only node it currently

affects.

---

5. Factory implication — what changes per flagged realm node

For a full-tier (TRUE) realm node, the per-chapter factory does five things it does not do

otherwise:

1. The 8-state machine. canonical_question_review before authoring — the node cannot reach

authoring without it (BPC §9.7).

2. The Stage-2 batch. SSBL Appendix A question set pre-drafted **with a researched proposed

resolution per question**, so Josh's pass is a decision batch, not a research session. For Ch 71,

72 and 73 this is the deliverable that does not exist today.

3. Quests leave the agent-only track. author_in_loop_required TRUE propagates to the quest

rows (§6.5 / §6.4), plus the §6.8 cultural-authenticity batch audit.

4. The rights-strict asset path (already specified by charter §3.3, and already true for these

nodes via generation_class — the flag change makes the spine lane agree rather than adding a

new burden): Panel A is Lane R only; every Panel A image carries a rights value; the

never-raw-generate rule extends from structures to iconographic systems — the Duat's

registers and gates, Naraka's strata, Xibalba's houses, Meru's axis are living traditions' own

pictures of their own cosmos and are exactly what a model will confidently hallucinate at AAA

fidelity.

5. Critic routing at full depth per care_tier, plus the §17.9 reveal re-verify on the

narrative-basis nodes (76, 77).

For a LIGHT-tier node (66, 67, 68, 70, 75): the §6.8 batch cultural audit and the anti-fringe /

reveal critics run; the Stage-2 Josh batch and author-in-loop quests do not. This is CG-26's

calibrated tier, applied verbatim.

For the road's substrate debt, which the flag change makes newly urgent: every one of the three

newly-full-tier nodes has a named RCM substrate gap in the charter — Xibalba carries "a blunt

do-not-use instruction on the displaced body"; Naraka/Meru has a substrate gap and the *least*

attention on the road; Annwn is MISSING ENTIRELY from the RCM Acquisition Priority Master List

(positive-controlled in the charter). A Stage-2 batch cannot be pre-drafted with researched

resolutions over a substrate gap. **Sequencing consequence: the Wave-3 substrate repair for Ch

71/72/73 is a hard predecessor to their Stage-2 batches**, and should be scheduled as such rather

than discovered at the factory.

---

6. Residuals surfaced, outside this axis

• Ch 50 (Amazon / Colombia / El Dorado) is hard_line_status = CULTURAL_PROTECTED and

canonical_hard_line_flag = FALSE — the only CULTURAL_PROTECTED node in the arc that is

unflagged. Same defect class as Ch 71/72, outside the road. Worth a row in the arc-wide pass.

• The three-concept conflation (§1.3) deserves a one-paragraph disambiguation in

CASCADE_RUNBOOK and the spine-chapter skill, and would close the open

[DECISION NEEDED - JOSH] at topography_rows.json L946 (CH_66 astral vs

hard_line_status = NONE) as a definitional answer rather than a canon ruling.

• CVD L747 ("Tier C MMO does not extend into" hard-line chapters, naming Ch 38/57/69) needs its

set stated explicitly against whichever concept it means, before the Tier-C lane reads it and

excludes nine realms by accident.

• Ch 68's live narrative debt, charter-named: the Body prose is a KEEP-VERBATIM block still

reading to the pre-J10 non-lethal framing while the typed layer reads to the fought-and-broken

host — "a narrative-pass debt sitting inside a hard-line node." It survives the flag change (Ch 68

drops to LIGHT) and still needs the pass.

Generated by harness/site/structure_site.py — the URL path is the repo path. review root