L1 has been a 2-event skeleton since round 1, waiting (in its own status field) on "Lane B's
arena-mode controller". That exists now, so C's round2_plan is built as written: molar
crush-cycles, the saliva tide, amylase droplets, the three samples exactly where C put them
(behind the molars / under the tongue / a tonsillar crypt), and the epiglottis exam. 24 events.
The teaching order is the level: move (orbs, no current) -> LOOK (the arena opens and you can
stop, which no other level allows) -> shoot (one droplet, then two) -> coat (the molar ring
grazes you, mucin is right there) -> RESTRAINT -> flow returns (the tide, the funnel) -> the
exam. Every verb alone, in the calmest room in the game, before anything is asked of it. Two
hazards are deliberately re-used rather than invented: the molar crush IS a ring_gate and the
saliva tide IS a non-lethal surge, so L2's gates and its reflux finale become callbacks instead
of new ideas.
THE MAST CELL (LANE_C_NOTES §Boss specs 5) sits at the exact midpoint. A sentinel immune cell:
cyan, no attack, no chase, inert aura (auraDps 0 — it shares the floater's drift branch and its
aura does literally nothing). Fly past and the encounter just ends. Shoot it and it degranulates
— droplets close in from the walls (which is how "the cavity swells shut" is expressed, because
a runtime tube radius is not in Lane A's contract), the coat bleeds, and past half health the
airway closes on a 25s timer. You always get through. The only question is what you get through.
It needed a new boss `kind: 'sentinel'` with its own update path, because forcing a thing with
no rings and no windows through the phase machine would have been worse than a branch. Damage is
the failure state, so every number describes a consequence rather than a fight.
Verified, all three player choices: SPARE -> "spared", coat untouched at 100, karma +25, score 0.
SHOOT -> "degranulated", coat 100->56. KILL -> "anaphylaxis-survived", coat 100->6. Score is 0 in
every case: shooting it buys literally nothing, so no medal can ever reward the mistake.
Two bugs the testing paid for: `def.phases[0]` crashed on a boss that has no phases, and the
first degranulation drain (5/s) was UNDER the player's 8/s coat regen — measured coat going
80 -> 100 during a histamine cascade, i.e. the punishment was healing you. Now 14 and 22, which
are net -6/s and -14/s against regen.
L1 is no longer a skeleton: its checkpoint-gap and zero-samples warnings are gone, and the
campaign now opens on a real tutorial instead of an empty mouth with a gate at the end.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Two boss specs written to C's format (LANE_C_NOTES §Boss specs, round 3 additions) and the
first of them built. Both extend the roster along C's own law — one boss, one pressure — onto
the two axes the game had systems for and no boss pointed at: reputation (§4 Sovereign) and
restraint (§5 Mast Cell, specced only).
THE SOVEREIGN. The appendix is not vestigial: it is a biofilm reservoir where the gut's flora
shelter and re-seed the colon after a purge, and the whole fight is that fact. It is the only
boss whose existence is decided before you arrive — read once on entry:
karma >= 70 THE COMMUNION — no fight at all. Full coat, a permanent phage escort. The
reward for five levels of restraint is being spared a boss.
karma >= 30 THE TOLL — P1 only.
below THE SOVEREIGN — three phases. P1's matrix is immune to the cannon and only the
antacid torpedo disperses it (a mature biofilm's whole defining trait); P2 is a
quorum bloom that re-armours when the room gets crowded; P3 is a 20s reseed run.
And the mechanic that makes it this boss: the lair is seeded with YOUR OWN FLORA, and it feeds
on dysbiosis — every commensal your stray fire kills re-armours a plate, live, mid-fight.
KARMA vs STANDING — a real bug found in testing, and worth the extra number. `standing` is a
SPENDABLE meter: a full bar is consumed to bud an ally, dropping to a 50 floor. That meant a
perfect player could never HOLD >= 70, so the Communion was almost unreachable — verified: a
symbiotic pilot showed standing 50 and was demoted to "toll". `karma` is the same inputs with
nothing spending it. The Sovereign's tier and gates' `secretTo` branches read karma; E's bar
still reads standing.
Also: boss.js generalised to a registry, which was the claim its header made — the Sovereign
needed zero changes to enemies.js. LS_appendix (3 segments, 800 units, 16 events), reached from
L4's ileocecal valve via `secretTo`/`secretNeed` — anatomically exact, the appendix hangs off
the cecum right past that valve. The secret is hidden behind conduct, not behind a wall.
C's selfcheck caught par.time as a free medal and the fix improved the design: par 160 sits
under the timid traversal, so only a Communion run (which has no fight) can beat it. The time
medal rewards the same conduct the level is about.
Verified: all three tiers branch correctly (communion = 0 nodes + 4 phages; toll = 1 phase;
full = 3 phases, 4 armoured); killing a commensal mid-fight re-armours a plate (reason
"dysbiosis"); the valve routes karma 100 -> LS_appendix and karma 0 -> L5. Levels 5/5 valid.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The first new level since L2, and the campaign's endurance beat. 7 segments, 6000 units,
63 events, ~10 minutes. Authored entirely from systems that already exist — no new engine work.
The three acts are anatomy, not invention: duodenal bulb -> descending duodenum (the ampulla,
where bile enters) -> jejunum -> deep jejunum -> ileum -> terminal ileum -> ileocecal valve.
Giardia really colonises the duodenum/jejunum and Peyer's patches really concentrate in the
ileum, so the difficulty curve IS the gut's own immune gradient: you fly from the quiet
absorptive end into the lymphoid end and the walls get angrier.
Identity vs L2: flow 8-10 against L2's 12-20, curviness up to 0.70, fog 0.050. L2 punishes bad
reflexes; L4 punishes bad reading. Segmentation ring gates (the small bowel's real motility)
teach at 70% open and test at 43% in an 8-unit lumen.
Three catalogue rows carry the new fiction onto existing archetypes — chyme_clump (floater),
neutrophil_hunter (seeker), bifido_colony (flora). The ileocecal biofilm gate at s5500 is the
first time the reputation loop pays out at a level's END rather than mid-level.
C's own tooling verifies it: selfcheck 4/4 valid, and --sim reports par 600s vs par-pace 576s
("par is reachable at par-pace throttle"), a pressure curve of 1.00/1.43/1.90/2.46/1.65/2.69/
1.63, min clearance 3.62 vs the 2.5 law, all 22 checkpoint gaps <= 29.3s. Driven end-to-end
in-engine: all six entity types spawn, biofilm fires at 5500, valve gate completes to L5.
The GDD's Tapeworm miniboss is not here — no boss framework exists yet. It goes in at ~5500
when that lands; the level notes say so.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The deeper payoff of the flora loop. A new `biofilm` level event: cross one and combat checks
BIOME STANDING — high enough (>= need), the native flora PART the biofilm, budding a wave of phage
allies + refilling coat, then spending the goodwill; too low, it stays sealed and you run the
gauntlet raw. Emits `biofilm {opened}` so E can voice both.
- combat/index.js: biofilmGate(ev) on the level:event pump (fields need/allies/coat).
- levels/index.js: the selfcheck knows the `biofilm` type now (validates need 0..100).
- L2: one demoed at s2500 — the opening flora/akkermansia/beacon let you BANK standing, and the
gate cashes it into a phage escort right before the hiatus + reflux finale.
The mechanism is complete; L5 also gets the fast-route/Appendix branch when that level + world
support land. Verified in-engine: high standing -> opened, 3 phages, coat 50->90; low -> sealed.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Roster round-2, batch 2 — three new archetypes, mostly reusing existing behaviour branches:
- virion (norovirus): a tiny fast swarmer, its own pool of 40 so a cloud reads as a swarm.
Shares the seeker branch (chase + dissolve on contact). The arithmetic threat. -> L5.
- driller (H. pylori): a TANKY corkscrewer (hp 50). Seeker chase + a rotating perpendicular
drift (the helix), and it does NOT dissolve on contact — it grinds you and keeps drilling,
so you must out-damage it. A new behaviour branch. -> L3 stomach; previewed in L2 at s2240.
- beacon (Faecalibacterium): the reward-state commensal. Shares the commensal branch, heals
AND gives the biggest standing (25) with no coat cost. A bright cyan octahedral gem, the
special one of the family. -> L5; previewed in the L2 breather at s1580.
Verified in-engine: driller corkscrews + survives contact; virion is tiny + dissolves; beacon
heals coat and spikes standing to 50% in two tends. Three clean distinct silhouettes (gem /
cone / dot-swarm).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The first foe with an ATTACH state machine, and the first that gives boost a defensive job.
Giardia chases like a seeker, CLAMPS its ventral suction disc onto your hull, and drains coat
while it rides — until a boost's i-frames throw it (or it lets go after attachMax, then can't
re-grab for reattachCd). The counterplay is timing a boost, not out-running it.
- enemies.js: `latcher` archetype — a two-state (chase / attached) branch. Attached, it rides
player.pos + a spin offset and drains `drain`/tick; ps.iframeT > 0 (boost) detaches and flings
it clear. Amber tetrahedron (the trophozoite), so a clinging one reads on the hull.
- levels/enemies.js: ARCHETYPES += latcher; catalogue `giardia_latch` (biome small_intestine).
- L2: one Giardia previewed on the open thoracic run (s1940) — it belongs to L4's villi maze,
but the grab works in a flow tube, so it teaches the boost-dodge verb now.
Verified: it chased + clamped, drained coat 100->84 while attached, and a simulated boost
(iframeT) shook it off.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
A foe whose corpse keeps hurting — the definitive colon enemy. C. diff drifts with a corrosive
aura like any floater, but on death it leaves a SPORE_CLOUD: a static acid-green zone that
damages anything in it for ~5s, then disperses. Killing it costs you the ground it died on
(shoot the cloud to clear it, or fly around).
- enemies.js: two archetypes. `cdiff` (amber octahedron, shares the floater drift+aura branch)
drops a `spore_cloud` in kill(), placed at the death spot. `spore_cloud` (acid-green per the
readability law, foe:false so the phage ignores it and clearing it never scores) is static,
damages on contact, and counts down a `life`. Entities gained a `life` field (only the cloud
uses it).
- levels/enemies.js: ARCHETYPES += cdiff; catalogue `cdiff_bloom` (biome large_intestine —
registered for L5's encounter pass, not shoehorned into the esophagus).
Verified: killing a cdiff dropped a cloud at its death spot; sitting in it dropped coat 100->89;
the cloud expired after its life. Amber octahedron vs acid-green blob — clean read.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The cleverest node in the reputation loop — a commensal that eats your mucus. Brushing it
COSTS coat but fills BIOME STANDING fast (gain 12 vs flora's 5): the "spend coat for a quicker
phage ally" trade against flora's safe heal-and-slow.
- enemies.js: flora's branch generalised to a shared COMMENSAL branch, sign-flipped by data
(cfg.coat gifts, cfg.coatCost nibbles); flora:tended now carries `gain`. New `mucus_grazer`
archetype — cyan (still flora, readability law), but a dodecahedron amoeba-blob so the coat
cost reads as a different bug, not a betrayal by the healing rod.
- combat/index.js: standing gain comes from the event's `gain`.
- levels: ARCHETYPES += mucus_grazer; catalogue `akkermansia`; L2 teaches it at s210, right
after the flora reef so the same-colour / opposite-trade contrast lands.
Verified: brushing it dropped coat 100->88 while standing hit 36% in 3 tends (4x flora's rate);
distinct blob vs the flora capsule.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The first ALLY, and the reputation loop that earns it — "don't shoot the biome, it helps you"
(V2_IDEAS) made mechanical.
- combat/enemies.js + balance.js: the `phage` archetype. A friendly seeker that hunts the
nearest FOE (not the player) and rams it — a bacteriophage bursts its host after `bursts`
hits. Fast (26 > every foe) so it catches things, fragile (hp 8) so it is spent, not a wall.
Entities now carry a `foe` flag (foes true; flora + phage false); kill() generalised so ANY
ally scores nothing and never combos. Fallback silhouette is a bare icosahedron = the phage's
real icosahedral capsid. Teal (EMISSIVE.ally), distinct from cyan "don't shoot" flora.
- combat/index.js: BIOME STANDING — one run-scoped float. flora:tended raises it, flora:harmed
drops it 25; at the symbiotic threshold the biome BUDS a phage near you (capped 3, 3s cooldown),
spending standing. Emits `standing {value}` for a future Lane E meter.
- levels/enemies.js: ARCHETYPES += phage; catalogue `phage`.
Verified deterministically in-engine: tending flora budded a phage, which hunted a spore_pod foe
(30hp) and rammed it dead; three-colour readability holds (amber foe / teal ally / cyan friend).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The round-2 microbe pass, vertical slice. Cross-lane (integrator call): the point was one
thread proven end to end — art pipeline in, friend and foe out.
- core/assets.js (D): the GLB loader, finally wired. GLTFLoader was vendored but never
imported and the manifest had 0 models, so enemies.js's `glb?.geometry` hook was dead code.
Now every manifest model preloads (AWAITED — pools set geometry at construction, a late load
can't retrofit an InstancedMesh) into ONE BufferGeometry normalized to a unit bounding
sphere; get('models',n).geometry lights up the hook. Optional-asset law intact: any failure
-> no geometry -> procedural fallback.
- combat/enemies.js + balance.js (B): two new archetypes.
flora — the FIRST FRIEND. Cyan (ART_BIBLE: flora reads friendly). Drifts like a
floater but its aura HEALS (sheds coat), never harms. Shootable — and shooting
it is the reputation trap: kill() gives no score, no combo, emits flora:harmed.
Verified: the aura fires player.refill({coat}); a shot flora scores 0.
spore_pod — a foe yeast cluster: the floater's drift+aura, but its OWN pool so its blobby
GLB will not reskin the food-debris floater.
Pool geometry now scales a GLB by the archetype radius (GLBs ship unit-normalized).
- levels/enemies.js (C): ARCHETYPES += flora, spore_pod; catalogue entries lacto_drifter
(Lactobacillus) + yeast_pod (Candida). L2 gets a light teach beat in the calm opening — a
friendly reef at s180, one yeast at s300, both off the racing line.
Ships playable as glowing primitives (a capsule + an icosahedron), exactly like the rest of
the round-1 roster; the Trellis/Hunyuan meshes drop in via the manifest with no code change.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Landed before the cut: schema v2 (segments[].wave.amp override, validated — breathe stays
biome-owned); hiatus re-authored tight AND calm (radius back to 6.5, amp 0.6, 3.3u steady
clearance); par 210 -> 180 with full rationale (below throttle-only par-pace, above a surf
run — the medal teaches the mechanic); finale surge 21 -> 24 so surfing is the stylish
escape (required 1.20x); surf profile reads A's live CREST_FACTOR; chaseAnalysis() +
escapability laws in validate; NEW --probe builds A's real canal and measures clearance
against the model (all L2 segments clear the 2.5 floor, hiatus 3.75); isMain guard +
importSpline shim — fixes A's selfcheck process.exit killing C's mid-run (qa.sh had been
green on one level out of three).
Cut off before: fly-through of the integrated build (blocked on B), pickup economy handover
(B proposed numbers in balance.js meanwhile — reconcile), L1 encounter design doc, GDD boss
folding, NOTES. Committed by F to protect the shared tree; levels selfcheck 3/3 GREEN.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Data spine:
- Schema v1 ratified (TECH.md, C-owned). Additive over v0: schema/par/next, segment
name, cross-section placement (theta/rho as a FRACTION of safe radius), spread, span,
pickup+gate event types, hazard params. STUB_LEVEL synced (that const only, per
ROUND1 §Lane C.1) — world.hash() unchanged (1bdf001), geometry provably untouched.
- levels/index.js: registry + loader (node+browser, no THREE), real schema selfcheck
(qa's --selfcheck hook is now live) and a pacing simulator (--sim). Consumes the other
lanes' contracts live rather than restating them: A's biomes.js for ids and
wave.amp+breathe, A's SKIN, B's tuning.js hull radius, B's balance.js for scores.
- levels/enemies.js (fiction id -> archetype join table), levels/difficulty.js.
L2_esophagus — the vertical slice, complete: 3600 units, 7 beats, 47 events, 65 enemies,
10 checkpoints, 3 samples, reflux-surge finale. L1/L3 skeletons (shape only).
Anatomy is the level designer: the three real esophageal constrictions are the three
setpieces, and all three enemies are real esophageal pathology (bolus chunk, eosinophilic
esophagitis, candida esophagitis).
Measured (node web/js/levels/index.js --selfcheck|--sim; qa GREEN):
- selfcheck 3/3 valid, exit 0.
- pressure curve 0.27 -> 2.14 -> 4.35 -> 0.69 -> 4.53 -> 6.35 -> 4.40 (breather/pressure/
spike/breather/pressure/peak/finale).
- par 3:30 vs measured 2:47 speedrun / 3:19 par-pace / 4:22 timid; par.score 9000 of a
9400 kill budget (95%), computed from B's scores.
- all 10 checkpoint gaps within the <=30s death-cost law (max 28.8s); min clearance 2.73.
Two contract collisions found and resolved (LANE_C_NOTES):
- Biome ids: authored mouth/pharynx/cardia/pylorus; A's registry has six canonical ids and
rejected all four. A is right — biome = look, segments[].name = anatomy. Conformed.
- Balance: my first enemies.js restated hp/size/speed/score that B owns in balance.js.
Cut to a join table; index.js reads B's scores via scoreFor().
Caught live: A raised esophagus wave.amp 0.9 -> 1.4 mid-session, making the diaphragmatic
hiatus unflyable (2.33 vs the 2.5 floor). Widened 6.5 -> 6.8. The invariant also failed my
own aortic constriction's first draft (radius 7, symmetric squeeze) -> radius 9, directional.
No in-engine flythrough: ?stub=1&lvl= does not compose (snippet offered to F) and the stub
is single-segment, so it could only render beat 1. The pacing evidence is a kinematic model,
not a playtest — it proves L2 is flyable, paced and survivable, not that it feels good.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
