CAST_INVENTORY.md

npcs/CAST_INVENTORY.md

FOUNDATION A — The Staged Cast Inventory

Measured from all 79 docs/spine/CH_*.md nodes (Ch 1-77 + Prologue + Epilogue), the two

connective-web docs, and the companion-fold fields. Read-only pass. This is an INVENTORY of what

the spine already stages, not a design and not a recommendation — the casting/loyalty calls are Josh's.

---

0 · Method and the disclosed filters

Three different measurements are used below; they disagree on purpose, and the disagreement is the

single most important finding. Every count states which filter produced it.

• FILTER-1 — actor-beat count. Occurrences of a CHAR_NNNN id inside a beat actors: [...] line.

This is STAGED in-scene presence (the character is physically in a played beat). This is the

"Ntala-191 method": Ntala = 191 actor-beats. Measured by grep "actors:" CH_*.md | grep -c CHAR_id.

• FILTER-2 — any-mention chapter count. Number of CH_*.md files containing the id ANYWHERE

(prose, end-state folds, seam asserts, "carried lightly" bookkeeping). This OVER-counts presence

because a departed/rooted ally is still named in later folds. Measured by grep -l.

• FILTER-3 — multi-chapter actor recurrence. CHAR ids that appear as actors: in 2+ DISTINCT

chapters. This is the tightest test of "travels/recurs on-screen."

FILTER-3 result — the whole recurring-on-screen cast is seven ids:

The load-bearing consequence: of 161 distinct CHAR ids in the spine, exactly TWO

(Ntala, Ana) are staged as traveling companions. Everyone the folds call "found-family carried"

is, by FILTER-1/3, a single-home-chapter figure who is named forward but never staged present again.

The modern arc's large nominal "found-family" (CHAR_0140-0162, ~23 ids) is **bookkeeping, not

companionship** — see Tier 3. Any measured "companion" number that leans on FILTER-2 would be

inflated by exactly this; the lane's first burned failure mode (undisclosed filters behind numbers)

lives here, so it is disclosed up front.

---

TIER 1 — The staged travelers (the only two with companion energy across arcs)

Ntala — CHAR_0002 — "the healer's child" (PRIMARY companion, journal-keeper)

• Culture/region: origin deliberately soft in the early arc — first met in the African arc

("the marsh where they first met"; the Ch-3 fold plants the silent-observation arc beginning at

Ch 3, the Ch-13 body says "followed her route unsigned" from the Ch 10-12 marsh — a minor

Ch-3-vs-Ch-10/12 origin tension worth a ruling). Gender is NOT independently set: it inherits from

the player's Ch-1 best-friend selection (the design-resonance echo, HL_0116, zero forward valence).

• Presence (measured): FILTER-1 = 191 actor-beats; FILTER-2 = 78/79 chapters (absent only from

Ch 2 and Prologue). Silent unsigned watcher Ch 3-12 → formally joins and signs her name "Ntala" +

opens the evidence-journal at Ch 13 (the canonical join-and-journal beat, HL_0070) → continuous

primary companion Ch 13-77 + Epilogue.

• Role as staged: witness-and-keeper. She reads script-STRUCTURE by documented analysis where the

protagonist reads vril-imprint (Ch 16 "parallel intelligence"); holds the evidence-journal that

becomes the arc's spine (Ch 13→38 transfer→48 deliberation→56 decision→57 public release→77);

companion-assisted deeper reads (Ch 13 serpentine/Vimana clue). Fights beside occasionally but her

verb is record/witness, not blade.

• Staged join/leave/return: join Ch 13; NO leave — at the Ch-65 D-23 Departure/Continuation lock

she is "sticky either way" (the record stays open regardless), i.e. explicitly NOT an integrity

leave. Present into the Realm Road on Continuation, held-in-world on Departure (both non-gating).

• Trade/craft: the journal / documentation itself (an authored system, not one of the 10 trades).
• House entanglement: her journal is the object the House hunts (Ch 17 "made the protagonist a

marked target"; Ch 18 the journal a hunted plot object); her record is what answers Cassius at Ch 76.

• Companion energy: MAXIMAL and continuous. The spine's one true co-protagonist.
• Connective web: CF-2 (the Ch-1 best-friend gender-seed → Ntala's build → Ch-75 homecoming

reunion is the OPEN cardinal — recommendation in the web is "keep the echo, reunion available").

Ana — CHAR_0017 — "the melee blade"

• Culture/region: Manggarai (Flores) — a caci stick-fighter (the Flores caci martial tradition),

making her the ancient-arc origin culture's returning daughter.

• Presence (measured): FILTER-1 = 44 actor-beats; FILTER-2 = 55 chapters — BUT the FILTER-2

count is the textbook over-count: her real in-party windows are (a) Ch 3-5 as a rooted Manggarai

age-peer, non-traveling; (b) Ch 27-28 the Mediterranean-bodyguard reunion; (c) Ch 41-77 rejoined as

the party's melee blade. The Ch 29-40 mentions are all fold lines saying "Ana departed after Rome,

her next canonical node Ch 41" — she is ABSENT-in-party for 12 chapters but still named. Reporting

"Ana in 55 chapters" without this filter would be false presence.

• Role as staged: melee/bodyguard; "astral body-guard" (Ch 44/46); at Ch 61 she takes the wound

from the Blind-Spot's hunt (marked/hunted/nearly-taken, survives with her own agency — the bond

deepens, the personal cost carried).

• Staged join/leave/return: recurring age-peer Ch 3-5 → reunion Ch 27 → **departs after Rome

(Ch 28), out Ch 29-40 → rejoins Ch 41** (the triad-climax reunion) → continuous to Ch 77.

Then the ONE built integrity-leave in the whole spine: at Ch 55-56 she WITHDRAWS on the

sustained-corrosive "forcing-seizure" branch (FORK_CH55_A / ws.companion_loyalty_transition_events),

her melee slot covered by Wenceslao (CHAR_0148) so no beat walls — this is the sole staged instance

of companion-loyalty-IS-integrity (see Gaps).

• Trade/craft: martial (caci → melee blade); no trade-home.
• House entanglement: hunted/wounded by the House's Blind-Spot operative at Ch 61.
• Companion energy: HIGH but discontinuous — a come-and-go/for-life hybrid with two big gaps.
• Connective web: named as THE candidate for IR-3/CL-02 (the first integrity-driven leave/draw);

Ch 55-56 appears to be the node that answered that call.

---

TIER 2 — Recurring come-and-go allies (real on-screen recurrence, NOT travelers)

Nakhoda Binti Hassan — CHAR_0056

• Culture/region: Swahili coast (a rare female dhow captain of the historical tradition).
• Presence: FILTER-2 = 11 chapters (Ch 8, 9, 14, 16, 17, 33, 34-38). Roots the Trade-10

Fishing/Navigation home at Ch 8; recurs on sea-crossings Ch 9 and Ch 38 (and offers the

ally-status-conditional dhow passages at Ch 16/17), each explicitly "an opportunity, never a gate."

• Role/craft: Trade-10 (Fishing & Navigation) teacher; the mtepe sewn-plank craft; a positive

cross-cultural recognition. Modern home base carries to Ch 56.

• Loyalty shape as staged: come_and_go on her OWN arc, ally-status-conditional, never in-party.
• Companion energy: the clearest staged "come-and-go" template in the corpus — but a teacher/

contact, not a traveler.

Kxao — CHAR_0071 (San/Rift-Valley shaman-elder)

• Rooted teacher recurring Ch 10 → 13 → 41 (the serpentine/Vimana read across the African arc). A

recurrence template for a rooted-mentor-who-reappears, not a companion.

The Elven King / father — CHAR_0005

• Endgame reveal figure (Prologue, Ch 1, 75, 76, 77). NAME-FORBIDDEN (HL_0099). Not a companion —

listed only to keep FILTER-3 honest (he is the 3rd-most recurring actor id and must not be mistaken

for ally-cast).

---

TIER 3 — Single-home-chapter allies "carried lightly" in folds (the candidate light-ally tier)

Each of these is staged as an actor ONLY within its home chapter (FILTER-1 small, all in one

chapter), then named forward in companions_present folds as "carried lightly" / "found-family

carried." None recurs as an on-screen actor (FILTER-3 = 1 chapter each). They are the material

Josh flagged as the light-ally candidate tier.

The four named modern-arc light allies (Josh's candidate tier)

The accreted "found-family" — CHAR_0140 through CHAR_0162 (~23 modern-arc regional ties)

All rooted to their region, each staged only in its home chapter, then aggregated as

"the accreted found-family carried." FILTER-1 for each is 1-6 beats in one chapter; FILTER-3 = 0

recurrence. Roster as the folds name them:

Tension to flag: Ch 64/65 folds say these ties are "carried aboard the Vimana … rooted to their

regions" — carried-aboard and region-rooted at once. Mechanically they are region-rooted (never

actors after their home chapter); the "aboard the Vimana / found-family carried" language reads as

warmth-bookkeeping, not staged travel. This is the gap between the mothership-hub travel model (which

implies companions travel ABOARD) and what the spine actually stages (nobody but Ntala/Ana is aboard).

---

TIER 4 — One-chapter teachers / hosts / honored figures (the bulk, ~130 ids)

The remaining ~130 CHAR ids (CHAR_0008-0016 Flores, the region-teachers of every ancient node, the

honored figures Shomer/Emunah/Senami/Raghavendra, etc.) are single-chapter, single-region texture:

knowledge-holders, trade-tutors, honored NPCs, boss/antagonist operatives. They are inventoried here

as a CATEGORY, not rowed individually (rowing 130 with 13 columns each would bury the signal and is

not what the mandate needs). They matter to companion design only as the SOURCE POOL from which a

future rooted-mentor-who-recurs (the Binti Hassan / Kxao pattern) could be promoted. Representative

companion-adjacent examples already carrying that latent energy: Elder Rato (CHAR_0009), Priestess

Senami (Trade-1 home, Ouidah), the waganga healer-allies (Ch 8).

---

THE GAPS — where the story structure wants a companion and none is staged

Measured against the arc structure (ancient Ch 2-38 region-per-chapter · modern 39-65 · Realm Road

66-77). Each gap is a fact about the current spine, not a proposed fix.

1. The entire ancient-arc first half (Ch 2-12) is SOLO. Eleven consecutive region-per-chapter

nodes (Flores → Sri Lanka → East Africa → Congo → West Africa) have NO active traveling companion.

Ntala is a silent unsigned watcher (no actor-beats until Ch 13); Ana is a rooted Manggarai age-peer

confined to Ch 3-5. The protagonist witnesses the arc's foundational assimilation-and-integrity

choices with no companion to witness THEM. This is the single largest companion gap in the arc and

directly answers the mandate's question "does any companion witness the ancient arc's first half?"

— answer: no.

2. The Ch 29-40 mid-ancient stretch has ONE companion (Ntala only). Ana is departed; Binti Hassan

is conditional/at-sea. Twelve chapters of single-companion travel.

3. The Realm Road (Ch 66-77) roots NO new companion. Every realm node explicitly states "no new

ally NPC introduced" — the otherworlds have no living communities to ally with. The Realm Road is

ridden by Ntala + Ana + the reunited childhood familiar (the 22nd-slot, Ch 58/65) + the endgame

father/King. CF-2 (re-meeting the grown best friend as a realm-NPC at Ch 70/75) is the OPEN cardinal

that could populate this — currently unbuilt. Answers "who rides the Realm Road?" — the two

travelers, the familiar, and (endgame) the father; no realm-native companion.

4. Path-flavor gap — the companion-loyalty-IS-integrity mechanic has ~ONE staged instance. The

connective web's own grep found "ZERO built instance of a companion leaving or being drawn on

integrity grounds"; the Ch 55-56 Ana WITHDRAWN-on-corrosive branch (covered by Wenceslao) is the

one node that answers it. The loyalty-model variety Josh ratified (for_life / invested_sticky /

come_and_go / drawn_conditional / not_companion) is NOT exercised across a roster — there is no

staged for-life companion besides Ntala (who is "sticky either way," i.e. not integrity-reactive),

no staged DRAWN-in companion, and only the single Ana WITHDRAWN instance.

5. No companion is culturally rooted outside the Flores origin + the modern Americas. The two

travelers are Flores-adjacent (Ana Manggarai; Ntala African-marsh origin, soft). The whole ancient

Eurasian/Mediterranean spine (Ch 14-38: Kush, Egypt, Indus, Mesopotamia, Levant, Persia, China,

Rome, the Atlantic) stages zero traveler-tier companion. Any culture whose companion the design

wants there would need substrate the corpus already has (those nodes are richly built) but a

traveler is not staged.

6. Substrate-thinness dependency flags (where faking depth would violate §17.1): the two

travelers' home substrate is solid — Ana's Manggarai/caci tradition is anchored in the Ch 2-5

Flores build (Forth, Allerton); Ntala's is deliberately soft/African-marsh and would need a ruling

on origin (Ch-3 vs Ch-10/12 tension above) before a culture is assigned. If Josh promotes any Tier-3

regional tie to traveler status, the ones with the thinnest current on-page substrate are the

deferred/unnamed ones (CHAR_0155 Ainu "name deferred," the un-idd Tunde/Amboko) — these are the

book-processing-enrichment dependencies to flag rather than fake.

---

Summary counts (all filters disclosed)

• Distinct CHAR ids in the spine: 161.
• Staged traveling companions (FILTER-3, recurring on-screen across arcs): 2 (Ntala, Ana).
• Recurring on-screen allies total (FILTER-3, 2+ chapters): 7 ids, of which 2 are travelers,

1 come-and-go teacher (Binti Hassan), 1 rooted recurring mentor (Kxao), 1 rooted elder (Rato),

and 2 are non-ally (the father CHAR_0005, the Architect CHAR_0003).

• "Carried-lightly" light-ally candidates + accreted found-family (single-home-chapter, fold-named):

~27 ids (Tunde, Amboko, Oupa Gert/CHAR_0125, + CHAR_0140-0162).

• One-chapter texture cast (Tier 4): the remaining ~125 ids.
• Ntala actor-beats: 191 · Ana actor-beats: 44 · next-highest ally actor-beats: single digits.

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