FORKB_ANALYSIS_CANON_INTEGRATION_2026-07-27.md

pipelines/FORKB_ANALYSIS_CANON_INTEGRATION_2026-07-27.md

FORK B — CANON INTEGRATION AXIS

Question: is companion resurrection canon-legal, and where does it live?

Analyst: read-only deep-dive, 2026-07-27. Canon repo C:/dev/humanity-forgotten @ 267463e.

Register discipline: every claim below is labelled CANON (a ratified tier doc / Josh ruling),

LIVE-DATA (a registry row as it exists today), PROPOSAL (a proposal-tier doc), or RECOMMENDATION

(mine). Absence claims are positive-controlled and say so.

---

0. HEADLINE VERDICT

**Companion resurrection is canon-legal, and it is a VARIANT OF THE RETURN — not a new

necromancy-flavored power.** It does not collide with the death/vril doctrine; it is the doctrine's

own two mechanisms pointed at the protagonist for the first time.

Three findings carry the verdict:

1. The ratified defeat model is already a consciousness-relocation event, not a death-and-undo.

"Defeat model RATIFIED as canon: consciousness withdrawal to the last attuned recharge source

• defeat writes a path-variable journal observation" [CANON — docs/BATCH1_RULINGS.md L50-51].

The verb is *withdrawal*, the same verb the vril canon uses for the fairy realm's contraction

("it withdrew because atmospheric vril density … dropped below the threshold"; contraction-not-

collapse is a canonical lock) [CANON — T1_Vril_and_Magic_System_Master [ACTIVE v2.2] §10 L286].

Nothing in canon says the protagonist's *soul departs* at HP 0. Companion-revive therefore does

not reverse a death; it intercepts a withdrawal in progress. That is a variant of the return.

2. **Revival is already in the ability catalog — three abilities and a conduit — and it is

compassion-track vril healing, not necromancy.** [CANON — T1_Ability_Tree [ACTIVE v1.4]]

• Vril Revival — T5 Philosophus, Ether-Crown seat. Gate: L1/L2 integrity + Vril Healing

numerical 70-99. "a downed ally returns to combat consciousness with reduced vitality" (L922).

• Vril Mass Revival — T7 Adeptus Major, Crown. All downed allies in range (L924).
• Breath of Life — T5 Philosophus, Air/Heart seat. Same shape via a different element

route; "the L4 and L5 evil paths access alternate revival mechanics through different element

routing" (L848).

• Collective Surge — T7 Crown. Decisive for this fork: "one ally's healing-class cast

routes through the conduit to reach every linked ally at once, and an ability cast by any

linked ally may be relayed and re-expressed … so a single Bloom Pulse, Vril Mass Revival,

or Storm Calling propagates" (L936). **Canon already presupposes that ALLIES possess

healing-class and revival-class casts.** An ally who can revive is not a new class of being.

• The evil-path mirror already exists and is named: Soul Drain (T5, L4/L5-locked, L930).

Revival sits on the compassion track; extraction sits on the predation track. The polarity

firewall is pre-built.

3. The out-of-combat constraint is DERIVABLE FROM EXISTING CANON, not an arbitrary rule. The

seven infinite-vital-energy devices bond to the protagonist's vril-seat equipment only —

"Devices 2-7 … Each, once activated, infuses the protagonist's vril-seat-associated

equipment to produce Enhanced Ability Mastery" [CANON — T1_Vril §2 L45-48, §3 L62]. A

companion has no device, therefore no Enhanced Ability Mastery, and runs the plain twelve-tier

ladder. An ally channelling a Crown-seat revival without device infusion cannot hold the channel

under combat pressure. Josh's "only out of combat" is the mechanical consequence of a canon lock

he already owns. This is the single strongest integration argument in the whole analysis —

the constraint should be stated in-fiction as capability, never as a UI rule.

One correction the parent brief needs: the brief says "protected chapters carry 1-HP

withdrawal." That is inverted. defeat_protected / the 1-HP clamp protects the BOSS from being

killed (Naga Padoha is the slice's one clamped boss; the boss withdraws at 1 HP) — it is not

player-death protection [CANON/LIVE — docs/FUN_REBUILD_PLAN.md R1.1 L46-51, L372;

docs/PRE_5090_BUILD_PLAN.md L349-350]. The player can and does die inside a protected encounter.

The care question at §8 below therefore stands and does not resolve itself for free.

---

1. THE FRAMING QUESTION, ANSWERED IN FULL

1.1 What the canon says happens at HP 0

1.2 What the canon does NOT say — positive-controlled absences

• No soul/afterlife metaphysics for the protagonist. Grepped \bsoul\b across all T1 Foundation

and T3 Character docs: every hit is an *ability name* (Soul Drain, Equilibrium Soul), a *cultural

substrate* row (Ka/Ba/Akh, Plato's tripartite soul, dusha/semangat/tondi as vril-parallels), or a

research-corpus title. Zero rows define what happens to a person after death in-world. The

search returns hits, so it can match; the absence is real.

• No canon anywhere has a companion revive the PLAYER. Regex

(companion|ally|familiar|healer)…(reviv|resurrect) across docs/, T1 Foundation, T3 Characters

returns only Collective Surge and unrelated reconstructed_revival lineage tokens. **Positive

control:** the same regex shape against the known-present downed ally … returns pattern hits

T1_Ability_Tree L922 on the first try. The search works; Fork B is genuinely unminted canon.

• No resurrection metaphysics of any kind is asserted. The revival abilities are described

purely mechanically ("returns to combat consciousness with reduced vitality"). Canon deliberately

never explains what revival *is*. That restraint is an asset — see §9.

1.3 Why "variant of the return" beats "new power" — and the one place canon pushes back

The strongest canon support for the variant reading is that both endpoints already exist:

consciousness withdrawing to a recharge site (the return), and a vril channel restoring a downed

body (Vril Revival). Fork B is the race between them, and the race is dramatically legible

without a single new metaphysical claim: the world starts to pull you back to the well; someone who

loves you is running toward your body with a channel open; you choose whether to wait.

The pushback, stated honestly. T3_Familiars_Named [ACTIVE v1.0] L919 rules the Ch-58

childhood-familiar reunion as "sixty chapters of grief resolved not through resurrection but

through the recognition that what was lost was never lost." Canon has *already declined

resurrection as a narrative resolution mechanism.* Combined with the register ruling that

named/persona'd NPCs stay permanently dead (the community-ripple doctrine)

[PROPOSAL/ruled — docs/DESIGN_GAP_REGISTER.md #13 L263-266; re-cited LENS_BG3_2026-07-27.md

L91, L122], the corpus has a consistent position: **death of a person is permanent and the world

absorbs it.**

Fork B does not violate that position, but it can be *read* as violating it if built carelessly.

The firewall that keeps it clean, and which every downstream surface must hold:

THE ASYMMETRY RULE (RECOMMENDATION — the load-bearing invariant of this fork). The revive

reaches the protagonist and nobody else, ever. Every NPC, companion, and familiar who dies

stays dead. The protagonist is the one being in the game whose consciousness the world already

pulls back to a well — that is the ratified defeat model, and it is why the ability terminates on

her alone. A companion revive that could also raise a dead NPC would collapse the community-ripple

doctrine, the R-17 weight-and-consequence bar (CVD §17.16), and the Ch-58 grief resolution in one

stroke. This asymmetry needs to be written into the owning tier doc as a hard line, not left

implicit.

The asymmetry is not a fudge: it is already how canon works. The protagonist is fairy-realm-born,

arrived through a rift, and carries a two-clock age model no other character has. She is the

exception by construction — and canon never explains why, which is exactly what §17.9 requires.

---

2. WHAT TEACHES IT — the trade / mastery / questline gates

Josh's phrase is "trained OR taught." Those two words map cleanly onto the two ratified bonus

lanes, which is a strong sign the direction is already speaking the corpus's own vocabulary:

2.1 The canonical gates that already exist and should be reused verbatim

• Vril Healing (T4 Practicus, Crown seat) — unlock at **L1 Truly Good or L2 Good Enough

integrity band plus Ch 6 Sri Lanka**. "The integrity gate routes healing through the path-aligned

compassion track; L4 and L5 access the alternate Soul Drain capability instead" [CANON —

T1_Ability_Tree L920].

• Vril Revival (T5 Philosophus) — L1/L2 band plus Vril Healing numerical 70-99 [L922].
• Breath of Life (T5, Air) — L1 band plus Vril Healing at T5 depth [L848].
• The healing seat is the CROWN — "Ether-class abilities at the Crown engage consciousness

directly — astral projection, vril healing, vril sharing" [CANON — T1_Ability_Tree §11.7 L914].

A revive is a Crown-seat consciousness operation. That is the correct home and it is already built.

2.2 Is there a "healing trade"? No — and this matters

The ten trades are fixed [CANON — T0_Trade_Registry; COMPANION_MECHANICS_CONTRACT.md §6].

None of them is a healing/medicine trade in the revival sense. The nearest is:

• Trade 4 Herbalism — "Herbalism as medicine-philosophy … operationalizes real biology as

its player-facing knowledge substrate … **vril amplifies the biological effect rather than

replacing it**" [CANON — T1_Trades_Master [ACTIVE v1.0] L67]. Ancient home Ch 07 S. India,

modern home Ch 50 Amazon.

• Trade 5 Alchemy — substrate-transmutation philosophy; shares the Apothecary compound with

Herbalism [L204].

RECOMMENDATION — the dual-gate that makes the capability a real build decision. Gate the ally

revive on **Vril Healing depth (the canon vril lane) AND a Herbalism co-requisite (the real-science

lane)**. Two reasons: (a) it puts the ability on the AAAAA build-cost curve — an ally who buys

revival has spent points that did not go into damage or INT, which is exactly the §12 mandatory

tradeoff; (b) it honors the rewards-real-science doctrine (Herbalism's substrate is real biology;

"vril amplifies the biological effect rather than replacing it" is *literally a description of

resuscitation*). The revive then reads in-fiction as a healer's hands plus a vril channel, not

as a spell that undoes death — which is precisely the register that keeps §17.16 and the

community-ripple doctrine intact.

Do not mint an eleventh trade. The ten-trade count is locked structural canon.

2.3 The three teaching vectors, ranked

1. A companion teaching_role — the ratified interface. "teaching_role — the trade / ability /

worldview the companion teaches and the mastery-tier it gates" [CANON — persona person-extension,

COMPANION_MECHANICS_CONTRACT.md §1.1, Interface 6]. **This is the cleanest fit for "taught":

one companion teaches another (or the player teaches the companion).** It also reuses the

non-gating constraint that already governs teaching_role — a lost teacher must reroute to an

equivalent-class alternative, never dead-end (Interface 6 + §9).

2. A Lane-B mastery quest gate — one of the 120 gates. Fits "taught" formally, costs a gate slot.

3. A familiar bond-ladder grant — see §4. Fits "trained" for the familiar axis specifically.

---

3. WHERE IT LIVES — the schema-space mapping with §5I.3

3.1 Doc ownership (fixed by §5I.3, do not move it)

"Ownership follows the §1 authority split. This doc owns the ability instance.

T1_Combat_System_Spec owns the weapon-proficiency and gear-imprint instances. T1_Trades_Master

owns the trade instance alongside the 120-gate roster. **T1_Integrity_Paths_Worldstates_Master

owns the companion and familiar stat layer that ally pools sit on.** One architecture, per-domain

rosters, no duplicated rows." [CANON — T1_Ability_Tree §5I.3 L541]

Therefore:

3.2 The §5I.3 capability-kind mapping (from the just-minted Family-A schema)

capability_kind_enum = ability · weapon · trade · attribute · familiar · companion ·

imprint · language · inscription · intuition · astral_projection

[LIVE — docs/pipeline_review/MINT_DESIGN_A_BUILD_SPACE_2026-07-27.md §1.2 L192-198; enforced by

harness/check_build_space.py tooth T-A3].

The capability_ref resolution map (§1.3, L229-241):

Where each piece of Fork B lands:

• The cast itself → capability_kind = ability, capability_ref = T0_Ability_Tree_Registry

ability_id for Vril Revival. No new capability_kind is needed. Bonus pools that deepen it

(channel speed, restored-vitality fraction, reach radius) hang off that row via

T0_Bonus_Pool_Registry / T0_Bonus_Option_Registry in the ordinary way.

• The ally's investment in it → capability_kind = companion, `capability_ref =

CHAR_0NNN. This is the pool axis §5I.3 assigns to T1_Integrity_Paths`, and it is where "you

built her to be the one who can bring you back" becomes a build decision rather than a script beat.

• The familiar variant → capability_kind = familiar, and the *ability grant* rides

T0_Familiar_Bond_Ability (see §4).

• The spec purpose → a NEW ROW in T0_Spec_Purpose_Registry. Live rows today are

SPEC_VRIL_HEAL_PERSONAL / _AREA, SPEC_VRIL_DAMAGE_PERSONAL / _AREA, SPEC_HEAL_OVER_TIME,

and a vril-regen row [LIVE — registries/T0_Spec_Purpose_Registry [DRAFT v0.1]/Sheet1.csv].

There is no revival purpose. RECOMMENDATION: mint SPEC_VRIL_REVIVAL with

function_class=heal, scope_class=single, role_axis_ref=ROLE_SINGLE_HEAL, sourced to

T1_Ability_Tree §11.7 + this ruling. Without it the build-space sweeps cannot count revival

builds as a distinct viable spec, and per §10's doctrine an uncounted spec is invisible to the

balance teeth.

• The role axis already exists: ROLE_SINGLE_HEAL and ROLE_SINGLE_SUPPORT are live rows in

T0_Role_Axis_Registry [LIVE]. Reuse; do not mint a "resurrector" axis.

3.3 The loadout-parity home — already built, and it answers the departure question

T0_Loadout_Set_Definition row LS_01 carries owner_scope=player_level,

applies_to_allies=TRUE, departs_with_companion=FALSE,

composition_kinds=gear;abilities;bonus_picks [LIVE —

registries/T0_Loadout_Set_Definition [DRAFT v0.1]/Sheet1.csv]. Companions therefore already carry

ability compositions in their loadout sets, switchable at a vril site. A companion's revive

capability is an entry in that composition. **A player can deliberately swap an ally into the

"she can bring me back" loadout at the sanctum before a hard fight** — which turns Fork B into a

preparation decision (Pillar-8's third axis) rather than a passive safety net. That is a large,

free design win sitting in already-landed schema.

3.4 The one genuine schema gap

There is no table today that says which abilities a given companion holds. T0_Character_Index

carries the six loyalty columns + attackability columns; T0_Ally_Behavior_Policy carries policies

keyed on ally_ref/role_tag/int_tier, not grants. The composition_kinds=abilities field on

the loadout set declares the *architecture* but no roster fills it. This gap is **not created by

Fork B — Fork B surfaces it.** It is the same gap the §8.11.7 ally-spread layer will hit at the

schema pass. Recommend it be recorded as one item, not two.

---

4. THE FAMILIAR LANE — the cleanest fit in the whole corpus

T0_Familiar_Bond_Ability [DRAFT v0.1] is live with the exact columns this needs:

`bond_ability_id, familiar_id, bond_level, bond_threshold, ability_name_text, ability_effect_summary,

grants_class, eam_pair_ref, element_affinity_required, command_permission, integrity_band_gate,

path_variable_resolution [LIVE]. bond_level runs bond_formation → bond_25 → bond_50 → bond_75 →

bond_100_mastery; grants_class ∈ {passive, active, summon_ultimate}`.

A bond-gated revive is a **bond_100_mastery / grants_class=active / integrity_band_gate=L1;L2

row.** Zero new schema. The doc anchor is already written: "The data home for the bond-gated

familiar abilities is the deferred bond_abilities child table … keyed on familiar_id with a

bond_level enum" [CANON — T1_Integrity_Paths §8.11.5 L565].

And the behavior precedent is already canon and already seeded. T3_Familiars_Named §1.4:

"L1 Truly Good … plus L2 Good Enough. The familiar cannot be used to attack until the familiar

steps in voluntarily at 50% or less player health. The familiar removes itself from battle once

the player's health recovers above 50%. **The player cannot override this behavior on these

bands** — the familiar acts on its own judgment. … The familiar is a companion, not a tool."

[CANON — L69]

"L4 Evil … plus L5 Pure Evil … If the player has bonded sufficiently despite the path, the familiar

steps in to protect at 50% or less health and **will not run away or lose loyalty if it dies

protecting the player.**" [CANON — L73]

That is seeded as live policy rows AP_0001 (interpose at ≤50% HP, autonomy_band=autonomous) and

AP_0002 (disengage above 50%) [LIVE — T0_Ally_Behavior_Policy]. **Canon already has an ally

autonomously intervening on a player-health threshold, integrity-band-gated, un-overridable by the

player.** Fork B is the next rung on a ladder that is already built — the ally who interposed at 50%

is the ally who kneels over you at 0. This is the single most natural extension point in the corpus.

A consequence worth naming: on L1/L2 the familiar's protective act is explicitly *not the

player's to command*. If the familiar lane carries a revive, then on the compassion bands **the

familiar decides**, and the player's "wait" is a request, not an order. That is a better beat than a

button, and it is what canon already says.

---

5. LOYALTY — does willingness ride THE BOND?

5.1 Yes for willingness; no for capability; and the two must not be crossed

THE BOND is the nine-band companion axis over loyalty_score (−100..+100): Enemy · Severed · Wary ·

Cooled · Stranger · Known · Warm · Trusted · Family [CANON, Josh ruling 2026-07-16 —

RELATIONSHIP_TIER_BANDS_BRIEF.md Rulings 2 & 4, recited in COMPANION_MECHANICS_CONTRACT.md §3].

It bands a STATE, not a running total.

The binding constraint from the tier doc:

"Behavior quality is a build property, never an integrity property. A high-intelligence ally is

not a more loyal ally, and no amount of investment moves the belief-distance scalar at §8.11.2.

Keeping the two axes separate is what preserves the ratified invariant that **investment can never

buy loyalty immunity**." [CANON — T1_Integrity_Paths §8.11.8 L612]

RECOMMENDATION — the clean three-way split that satisfies every ratified invariant:

That split gives Fork B three independent build dials and therefore three independent ways to

be a legitimately different party — which is exactly what the AAAAA breadth doctrine demands, and it

adds them without touching the loyalty-immunity invariant.

5.2 Loyalty-state legality (hard, not tunable)

loyalty_state ∈ unmet | recruitable | traveling | wavering | departed | estranged | bonded_locked

[CANON — WS7 landed schema; live column on T0_Character_Index].

• departed / estranged → cannot revive. Obvious and structural: she is not in `WS_027

companion_roster_state`, and the DEPARTED event has already discharged the gear return under

§8.11.6. A departed companion appearing to save you would break the departure contract's whole

point.

• traveling / bonded_locked → the normal case.
• wavering → RECOMMENDATION: SHE STILL COMES. WAVERING is the ratified *visible warning window*

before departure (effective trigger −15). A companion who is about to leave you over who you have

become, and who still kneels down and pulls you back, is the best single beat this fork can

produce — and it is exactly the "care = weight and authenticity" register. **Anti-exploit

requirement: the revive must grant zero bond movement**. If saving the player moved the bond,

players would farm deaths to repair relationships, which inverts the integrity system. State this

as a rule in the owning doc.

• The Ch-75 EVIL inversion — the healer's child's bond INVERTS rather than breaks, resolving at

FORK_CH75_D [CANON — T3_Core_Characters §3.4]. An inverted companion obviously does not revive.

The correct read is that the option silently is not offered, not that it is offered and refused

— the refusal would be a reveal.

• BOND floor — RECOMMENDATION: a soft floor, not a switch. Below a band (Cooled/Stranger and

under), the ally tries and does not reach you in time rather than being greyed out. Preserves

the AAAAA no-dead-pick principle (the option is never a lie), keeps the consolation read honest,

and makes the bond legible through play rather than through a tooltip.

5.3 The gear-return contract is untouched

§8.11.6's investment protection binds only the DEPARTED transaction. A revive fires while the

companion is *in the roster*, so it never interacts. No conflict. (One adjacent gap, already

recorded and not created by this fork: there is no dead value in loyalty_state and no KILLED

value in the WS_028 event enum, so a killed companion has no landing state — LENS_BG3_2026-07-27.md

BG3-2. If Fork B's asymmetry rule holds — allies stay dead — that gap becomes *more* load-bearing,

because the player will now visibly notice that the thing that saved her cannot save them.)

5.4 The healer's child collision — the one real canon blocker

The intuitive candidate is the wrong one, and the ruling is explicit:

"No combat abilities — informational and social capability only. … **no combat ability, vril,

or weapon is gained; denial derives from social and behavioral tells … Per the D-2 F-STAR

ruling 2026-07-19**." [CANON — T3_Core_Characters §3.4, Companion architecture]

She is the protagonist's *primary human companion from Ch 13*, the daughter of a **Manggarai

healer-elder**, and she is the emotionally correct person to be kneeling there. She is also

canonically barred from vril capability by a named ruling. Two honest readings:

• Reading A (conservative, applies today): a vril revival is a vril capability; D-2 F-STAR bars

it; the healer's child cannot be the reviver. Fork B lives on the other ~30 companions and the

familiars.

• Reading B (the reframe): D-2 F-STAR's context is combat capability and the Ch-75 inversion

read; an out-of-combat, non-vril, Herbalism/medicine-grounded resuscitation is not a combat

ability and is the exact expression of her mother's tradition. Her parallel-analytical capability

(Ch 16 Indus) is already canon as *a second, non-vril way of knowing that works*.

RECOMMENDATION: hold Reading A as the default and route Reading B to Josh as a briefed fork. It

is a genuine characterization-canon call on a Josh-authored ruling about a Josh-owned character —

not a creative-within-vision apply. But it should be *briefed*, not dropped, because Reading B is

the version of this feature with the most emotional weight in the entire game, and it is

reachable without touching her no-combat-ability line (the revive would be herbal/manual, out of

combat, and would explicitly grant her no vril, no weapon, and no combat presence).

---

6. THE DIFFICULTY / DENSITY DIAL — a free integration

The one canonical difficulty dial is ambient vril density and nothing else — "NEVER HP / damage

/ resistance inflation" [CANON — COMBAT_ENCOUNTER_SYSTEM.md L412], with the ladder Wellspring ·

Veiled · Fading · Sundered · The Forgotten One [CANON, naming closed by Josh 2026-07-04 —

BATCH1_RULINGS.md §4]. The on-death payout is already density-graded: "on Wellspring/Veiled a soft

catch ('no trial can strand you'). On Sundered/Forgotten One the payout is knowledge only — the door

stays hard" [PROPOSAL — COMBAT_ENCOUNTER_SYSTEM.md L220].

**A companion's revive is a vril channel, and vril capability scales with ambient density

(T1_Vril §1 L17-19, §10).** So the revive's reliability grades itself along the ruled dial with

zero new tuning surface: near-certain on Wellspring, a real race on Sundered, and on The Forgotten

One the reach may simply not be there. **This is the same "the realm's reach is fading" fiction the

difficulty names already carry** — Fork B inherits the dial rather than adding one. Recommend this

be stated explicitly so no one later invents a separate revive-success stat.

Related, and already canon: recharge sources themselves respond to vril signature — "Negative-vril

signature protagonists experience attenuated recharge; vital energy releases reluctantly"

[CANON — T1_Vril §5.7 L188]. An evil-path protagonist's *return to the well* is already canonically

grudging. Fork B's second door (waiting for a companion) is therefore most valuable to the path

that has the fewest companions willing to open it. That tension is good design and it is free.

---

7. WHERE FORK B IS AND IS NOT PRESENT ACROSS THE ARC

Build consequence, and it is a hard one: **Fork B has essentially ZERO surface in the pre-5090

Ch 2-5 slice.** Every death in the slice resolves to "revive at last checkpoint." The choice UI must

therefore degrade to a single option without ever showing a dead or greyed second option — a

greyed "wait for a companion" prompt in Ch 2 would be a promise the game cannot keep for eleven

chapters, and per the natural-voice doctrine it would also be a UI-taxonomy surface, not a person's

voice. The honest slice behavior is: **the second door does not exist yet, and the player never

learns it was missing.** The first time it appears, it should appear *because someone is there*.

---

8. PROTECTED CHAPTERS, THE 1-HP CLAMP, AND §17.1 CARE

8.1 The correction (repeated because it changes the answer)

defeat_protected / the 1-HP clamp is boss-side: the protected boss withdraws at 1 HP instead of

dying. Naga Padoha is the slice's one clamped boss [CANON/LIVE — FUN_REBUILD_PLAN.md L372;

PRE_5090_BUILD_PLAN.md L349-350]. The player is not clamped. So player death, and therefore

Fork B, fires normally inside protected encounters.

8.2 The environmental-boss protocol does not produce player death

The 4-step protocol (Perception → Vril-management → Restraint → Purgation) "cannot be defeated

through damage," and its canonical failure states are graduated consequence, not player death:

combat-scale expenditure destroys both / insufficient management feeds the manifestation /

partial resolution [CANON — T1_Combat_System_Spec §Environmental boss, L373-388]. The six

documented instances are Ch 10, 20, 29 (Sámi), 38, 45-46, 57 (Australia).

But the same section rules the protocol "is one encounter class within a chapter, never the

chapter's sole combat layer: **every listed instance also stages a real fought encounter elsewhere in

the chapter**" (L375). So the player *can* die in Ch 29, Ch 42 (Haiti), and Ch 57 — the three §6.2

high-care chapters — just not during the protocol beat itself.

8.3 The care ruling (DELEGATED — ruling, not flagging, per the care doctrine)

The risk is precise and small, and it is not "resurrection is disrespectful." It is: a

resurrection tableau staged at a living tradition's sacred site would place a fictional

metaphysical claim on ground that belongs to a real practice — the §17.1 collective-protection

surface. The tool is relocate and reframe, never remove (care doctrine, Josh 2026-07-22).

RULED (mine, applying the delegated care doctrine, calibrated against the loop's timid bias):

1. Fork B remains fully available in every high-care chapter. Removing it would be exactly the

care-tier inflation the doctrine names as a defect, and would create a silent difficulty spike in

the chapters that already carry the most authored weight.

2. The wake location is routed to non-sacred ground. The return already goes "to the last

attuned recharge source" [CANON, defeat model], and §5.7 already rules that certain protected

sites "do not respond to negative-vril signatures at all" — the corpus already models sacred

sites as having their own consent. Extending that: a protected sacred site is **not an attunement

target for a death-return**, and the revive tableau does not stage on it. This is a one-line

rule that reuses ratified machinery.

3. The revive's in-fiction language is culturally registered per the natural-voice doctrine —

the companion says what *that person from that place* would say over a body, in their register,

and never a system word. No "resurrection," no "revive," no tier vocabulary.

4. No borrowed rite. The revive never depicts or imitates a real living tradition's actual

funerary or healing ritual. It is the companion's own act.

---

9. REVEAL DISCIPLINE — what a resurrection surface may NOT say

The death screen is a **high-frequency, high-attention surface the player will read dozens of

times.** It is the single worst place in the game for a leak, and the corpus has two hard lines

plus one soft one that bind it.

9.1 The hard lines

• CVD §17.9: "The thesis is never declared explicitly anywhere — not in dialogue, not in

narration, not in marketing copy, not in tutorial text, not in UI." "**Pre-Ch 76 Grand Sage

identity hints are forbidden.**" "The Grand Sage Ch 76-77 reveal is the single canonical direct

surfacing of the thesis in the 77-chapter arc." [CANON]

• Antagonist-pattern reveal timing: no House-of-Velheim pattern before ~Ch 38; **even

invoked-and-negated agent-nouns leak** [standing memory; recited in

COMPANION_MECHANICS_CONTRACT.md §11 PINNED].

9.2 The specific leak vectors this fork creates

1. Explaining WHY the protagonist comes back. Any line that reaches for a reason — that she is

different, that something holds her, that the world will not let her go, that she is *owed* a

return — is a Grand Sage hint. The return is shown, never explained. The existing canon models

this perfectly: the revival abilities in T1_Ability_Tree describe *what happens* and never

*what it means*, across four separate ability rows. Match that restraint exactly.

2. The companion theorizing. A companion who wonders aloud what she is, or who says she has seen

this before, or who says *anything about the mechanism*, is the same leak in a warmer voice —

and worse, because a persona'd line carries more weight than a system string. Companion revive

dialogue is relief and reproach and exhaustion, never cosmology.

3. L3 theology. The corpus is dense with dying-and-rising substrate — Thread 19 (End-of-World +

Dying-Rising), Osiris at Ch 14-15, Yarilo, Dumuzi, Fasika at Ch 9, the Duat at Ch 69, the

Realm Road's Hades/Xibalba/Annwn at Ch 70-74 [CANON — T1_Threads_22 §L354, L363, L1120]. A

revive surface that gestures at any of it converts an L3 connection into a declared thesis,

which §18 show-don't-tell forbids. The resonance is allowed to *exist* — a player who has read

the Osiris material and then watches someone gather her up will feel it. **That feeling is the

payload. Naming it destroys it.**

4. The consolation read is the riskiest single string. It already ships (R1.6: stalk-cadence log

• a lore fragment) and it is *designed* to be lore-bearing on death. It must obey the same

pre-Ch-38 / pre-Ch-76 discipline as any other lore surface, and it must not become the place

where the game quietly explains itself because "the player just died, they'll accept anything."

5. The journal. Defeat writes a path-variable journal observation [CANON]. The journal is the

healer's child's voice and the game's primary lore-delivery system. A death entry written *by

her about the protagonist coming back* is a first-class reveal risk and needs the same critic

pass as any spine string.

9.3 The positive formulation

What the surface may say: what happened (plainly), who was there, what it cost them, what the

world did in the meantime (the Hades advance), and what you learned (the tell-reads). All of that is

ratified consolation content and none of it touches the thesis.

---

10. RANKED COLLISIONS AND RISKS

Nothing in this list is a canon-legality blocker. #1 and #2 are the two that need Josh's

signature or an explicit written invariant; everything else is execution.

---

11. THE RECOMMENDED CANON SHAPE, IN ONE PARAGRAPH

At HP 0 the protagonist's consciousness begins withdrawing to the last attuned recharge source, as

the ratified defeat model already says. The player may let it complete — that is the checkpoint

return, always available, never gated, never slower than today. Or the player may hold, and if a

companion or familiar in the party carries a compassion-track revival capability (Vril Healing depth

plus a Herbalism co-requisite, on the L1/L2 bands, held in that ally's current loadout composition),

and if that ally's bond and integrity band mean they will come, and if their INTELLIGENCE lets them

read the moment fast enough, and if ambient vril density gives the channel enough to work with, then

they reach the body and re-seat what was leaving. It works only out of combat because an ally has no

device-infused Enhanced Ability Mastery and cannot hold a Crown-seat channel under pressure. It

reaches the protagonist and nobody else, ever — every other person who dies stays dead, and the

world remembers them. Nothing on the surface explains why she comes back.

---

12. REQUIRED CANON EDITS (doc-by-doc, if Fork B is ruled in)

Note the standing gate discipline: any registry edit must refresh docs/fidelity_baseline.json via

--emit-baseline on registry_fidelity.py in the same commit, or the gates go red.

---

13. WHAT IS GENUINELY JOSH'S (briefed, never bare)

1. The healer's child question (§5.4) — Reading A vs Reading B against his own D-2 F-STAR ruling.

Recommendation: A as default, B briefed with options/pros/cons, because B is the highest-emotion

version of this feature and is reachable without touching her no-combat line.

2. The asymmetry rule as canon (§1.3) — it is a *hard line* about death permanence, which is

§17-adjacent cardinal territory. Recommend writing it and having him ratify it, not applying it

silently.

3. Naming — anything player-facing here is naming canon and flags per docs/NAMING_METHOD.md.

Note the natural-voice doctrine bites hard: there should probably be no name at all on the

player's surface, just a person's voice and a choice.

Everything else in this document is creative-within-vision and applies with the path shown.

---

14. EVIDENCE INDEX (paths for the parent's verification)

• _source/01_Tier_1_Foundation/T1_Ability_Tree [ACTIVE v1.4].md — L848 Breath of Life · L914 Crown

seat scope · L918 Astral Projection · L920 Vril Healing · L922 Vril Revival · L924 Mass Revival ·

L930 Soul Drain · L936 Collective Surge · §5H.3 L501-507 · §5I.1-§5I.4 L515-551

• _source/01_Tier_1_Foundation/T1_Vril_and_Magic_System_Master [ACTIVE v2.2].md — §1 L15-33 ·

§2 L45-48 · §3 L58-62 · §5 L109-185 · §5.7 L186-188 · §7 L231-247 · §10 L282-290

• _source/01_Tier_1_Foundation/T1_Integrity_Paths_Worldstates_Master [ACTIVE v2.1].md — §8.11.2

L543-549 · §8.11.5 L561-567 · §8.11.6 L569-579 · §8.11.7 L583-595 · §8.11.8 L597-612 · §8.11.9

• _source/01_Tier_1_Foundation/T1_Combat_System_Spec [ACTIVE v1.0].md — L371-399 (env boss +

failure states)

• _source/01_Tier_1_Foundation/T1_CVD_Creative_Vision_Document [ACTIVE v1.4].md — §17.9 L1093-1108

· §17.16 L1185-1197 · §17.1 collective-care L969

• _source/01_Tier_1_Foundation/T1_Trades_Master [ACTIVE v1.0].md — L67 (Herbalism), L204
• _source/03_Tier_3_Characters/T3_Familiars_Named [ACTIVE v1.0].md — §1.4 L65-75 · §6.1 L61 ·

Shell L143-174 · L919 (the not-through-resurrection ruling)

• _source/03_Tier_3_Characters/T3_Core_Characters [ACTIVE v1.0].md — §3.4 (healer's child

companion architecture, D-2 F-STAR) · §3.5-3.6

• docs/BATCH1_RULINGS.md — §4 L44-66 (defeat model ratified; difficulty names closed)
• docs/COMBAT_PROGRAM_ADDENDUM.md — §4 healing lanes L67-73 · §7 L116-133 · §9 L150-173 ·

§10 L176-196 · §11 L199-222 · §12 L225-260

• docs/COMBAT_ENCOUNTER_SYSTEM.md — L220 (on-death payout) · L412 (the density-only dial)
• docs/PRE_5090_BUILD_PLAN_VOL2.md — L802 (SL-1) · L1519-1521 (D-DEFEAT-AXIS RESOLVED)
• docs/FUN_REBUILD_PLAN.md — R1.1-R1.6 L41-72 · L372 (the 1-HP clamp is boss-side)
• docs/proposals/PRESENTATION_DOCTRINE.md — §0.A Fork A (ruled continuity-by-default) · §0.B Fork B

(ESCALATED, L98-104)

• docs/spine/companion_cast_designs/COMPANION_MECHANICS_CONTRACT.md — Interfaces 2/3/4/5/6/7/9/10,

§11 PINNED vs design room

• docs/pipeline_review/MINT_DESIGN_A_BUILD_SPACE_2026-07-27.md — §1.2 L192-198 (capability_kind_enum)

· §1.3 L229-241 (the resolution map)

• docs/pipeline_review/LENS_BG3_2026-07-27.md — BG3-2 L80-125 (no dead state; permanence ruling)
• docs/DESIGN_GAP_REGISTER.md — #13 L263-266 (named NPCs stay permanently dead)
• Registries (live): T0_Familiar_Bond_Ability · T0_Ally_Behavior_Policy (AP_0001/0002) ·

T0_Loadout_Set_Definition (LS_01) · T0_Spec_Purpose_Registry · T0_Role_Axis_Registry ·

T0_Bonus_Pool_Registry / T0_Bonus_Option_Registry · T0_Character_Index

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