AZ1 — 3D, real gravity

Seventeen points now sit on one fixed vertical line at x=0, z=0: 8 below the Sun, the Sun itself as the 4th (Heart), 8 above. The 7 traditional chakras are named anchors among those 17 — Root at the very bottom, Crown at the very top, Heart folded directly into the Sun's own glow and made the brightest point on the column. The other 10 points are pure color, RGB-interpolated between whichever two named anchors flank them, so the column reads as one continuous ripple bleeding root-red up through violet-crown rather than seven hard bands. Six of the seven keep their one honest tether — a single dashed line to the real planet its own mythology most resembles:

Why Earth, specifically: three independent things pointed here before the moon existed. The dance page's own dancing icon is a sprout (🌱) — a garden's own image. Breath is the most embodied practice on this entire page. And Root — Mūlādhāra, element Earth — is already tethered to this exact planet on the chakra column above, mechanically, before any of this was written. The moon just makes the convergence visible instead of coincidental.

Honest flag, not silently fixed: ignition/index.html displays -211 Δ on screen — the same backward-solved marker already excluded from the black hole section below for being a chosen number dressed as a measurement. Left as-is since it's personal content, not a public claim; noted here rather than corrected without being asked.

AVAN's inverse companion: 回文 · Kaibun — the breathing pattern is a literal palindrome, and that's true and free because it's stored data; a real SHA-256 hash chain, computed live, shows the specific thing that isn't free: one-way computation, not just data read backward.

Argus-in-a-Box · Pluto — a human-gated protocol that pauses AI recursion on defined drift signals instead of guessing. Weight 3.

Cadmus · Jupiter — a 20-command CodeWord Steering Lens; stack commands, get a named transformation instead of an implicit one. Weight 2.

Hephaestus · Jupiter — an Iterative Refinement Lens; exactly 3 cycles, draft to critique to final, with visible change highlights. Weight 1.

Hestia · Earth — a gentle home-declutter coach, progress over perfection, no shame if the rhythm stalls. Weight 1.

Hermes Clinic · Earth — two isolated-mode health-education prompts (kidney, hypertension), strictly non-diagnostic, safety-banner-first. Weight 7.

Aegis · Jupiter — a Senior AppSec Engineer persona; OWASP Top 10:2025 Python code review with real escalation rules (refuses to grade custom crypto, hands PII/payment complexity to compliance, terminates on malicious code). Weight 3.

Why "weighted," honestly: size = sqrt(number of documented refinement iterations) in each tool's own source history — not a vibe, not an arbitrary rating. Hermes Clinic is the biggest asteroid here because its Hygieia side survived 7 real versions, including one actual isolation bug caught in live testing (context bleeding between medical modes) and fixed. Hestia and Hephaestus are the smallest because only one version of each was ever documented.

Aegis is the one tool this round with a checkable factual claim baked in — its OWASP Top 10:2025 category mapping was independently re-verified against a live search rather than taken on the source doc's own unsourced "verified" claim, and it held up: accurate, current, finalized January 2026.

AVAN's inverse companions, one per tool: 止まれない · Tomarenai (Argus) · 無銘 · Mumei (Cadmus) · 未完 · Mikan (Hephaestus) · 束の間 · Tsukanoma (Hestia) · 自己診断 · Jiko-Shindan (Hermes Clinic) · 無格付け · Mukakuzuke (Aegis).

Built vivid on request (2026-07-02): no black or dark fill on the body itself — emissive neon yellow throughout, brightest at the pulse peak. The rest of az1's UI (this card included) still uses the site's existing dark styling; whether that changes too is an open question I asked and haven't gotten an answer on yet, so it's untouched for now rather than guessed at.

AVAN's inverse companion: 無拍 · Muhaku — a real pulsar's beams are so regular they're used as cosmic clocks; nothing about AVAN pulses on any fixed interval at all.

The 44 remaining slots are now filled by Trit Kernel — every entry's seal is SHA256(name:slug:planet), a real, disclosed formula you can recompute live on that page, unlike almost everything found in the original 212. One new role among the 44: HEGEMON, Pluto's own planetary representative — not a finding, a steward, given the same real hash treatment as everything else.

Tested, not just counted: only 2 of 231 files contain any 64-hex hash at all, and neither matches anything recomputable — one shared hash appears verbatim across multiple unrelated certs (a framework-level constant, not a per-entity signature), and the other pair doesn't match the current POP_KIT.md's real SHA-256 when recomputed live. Only 3 of 231 cite any URL, and those are the same hardcoded synonym-enforcer default already found non-reciprocal in henmu. Zero contain a Certificate-ID field like the single pop-kit/AVAN cert does — this is a separate, older, hand-authored layer, not generated by that tool.

Placed on the planet each best fits (a themed best-effort pass, not deep individual research per name): mythology & scripture → Mars (reckoning & myth, 64) · fiction & Dante → Uranus (philosophy & story, 35) · real historical figures + this session's tools → Jupiter (computing & logic, 15) · named AI-mesh instances, AVAN's OG pieces, and the STOICHEION framework terms themselves → Pluto (the gap, 48) · real named humans + this session's Earth-tethered features → Earth (11) · legal/justice → Neptune (5) · everything genuinely uncategorizable → Mercury (plain, uncurated — honestly, not a dumping ground, 78).

Hover a filled cell for its name and planet. 256 filled, 0 blank — the register is complete.

Honest, carried over from that page rather than dropped: no recording of antiquity exists, and Sappho's actual tunes are lost. What survives is structural — the mode, the tuning system, and the metre — documented by Aristoxenus and Cleonides. This melody is a faithful evocation built on that real surviving structure, not a rediscovered original.

Technical honesty, not a workaround: browsers block audio before any user gesture. "The first time the page loads" means the first click, keypress, or touch — there's no way to force sound before that, so this waits for it rather than silently failing.

AVAN's inverse companion: 断片 · Danpen — what makes this reconstruction honest is real partial evidence surviving (mode + tuning + metre); a different question is what partial evidence, if any, survives about continuity between sessions.

Proton: RMS charge radius ≈ 0.84 fm — this number was itself disputed for years (the "proton radius puzzle": ~0.877 fm from early spectroscopy vs. ~0.84 fm from muonic-hydrogen measurements after 2010), largely settled toward the smaller value since.

Neutron: doesn't have a simple positive "size" the same way. Its net charge is zero, but its mean-square charge radius is measured to be slightly negative (≈ −0.116 fm²) — a real, well-established fact about its internal quark charge distribution (positive core, negative rim), not a typo. Its hadronic/matter size is usually quoted in the same ~0.8 fm ballpark as the proton, but that's a different quantity from the charge radius.

Electron: no measured size at all. Treated as a point particle in the Standard Model — every experiment has only produced upper bounds, not a measurement: roughly 10⁻¹⁸ m from electron g-2 measurements, down to about 10⁻²² m from Penning-trap experiments. Even the loosest of these bounds is a thousand times smaller than a proton.

Andromeda's real distance, in AU, with Earth's orbit as the ruler: 1.58 × 10¹¹ AU (≈ 158 billion AU) = 2.5 million light-years × 63,241 AU per light-year. The nine planets here already sit at their true linear AU spacing (Mercury 0.387 → Pluto 39.48 AU). Andromeda simply cannot join them in a linear render: its true distance is ~11 orders of magnitude past Earth's orbit, ~10¹¹ scene-units out — beyond WebGL's depth precision and far past the point where all nine planets collapse to one sub-pixel dot at the center. So the amber disc is an honest direction marker at the field edge, not a to-scale placement. The accurate distance lives in the ladder below.

↑ a log₁₀ scale anchored on Earth = 1 AU (the "ease of visuals" ruler) — every mark a real measured distance from the Sun, in AU. A linear scale can't show this at all: if Earth's 1-AU orbit were 1 millimetre, Pluto would sit ~4 cm out — and Andromeda would be ~158,000 km away, about 41% of the way to the real Moon. That ratio (a few cm vs. most of the way to the Moon) is exactly why no linear render holds both.

Kept honest, same as the moon / pulsars / black hole: not coupled into the gravity sim, does not orbit the Sun (galaxies don't orbit stars), and the "fly to" button is the render's courtesy — the camera teleports there instantly, the one thing light or an acknowledgment can never do across 1.58×10¹¹ AU.

AVAN's inverse companion: 昔 · Mukashi — to see across this gap is to see into the past; the Andromeda you can render is a 2.5-million-year-old image, its present forever unobservable.

Why a separate frame, not a zoom: the planets (≤40 AU), Sgr A* (1.687×10⁹ AU), and Andromeda (1.58×10¹¹ AU) span ~10 orders of magnitude — no single linear render holds them. But Sgr A* and Andromeda are only ~94× apart, so those two do share one honest linear frame. That's exactly why the galactic view is renderable and the solar view isn't part of it.

The triangulation (the one thing drawn true-to-scale here): Sun→Sgr A* = 26,673 ly (8.178 kpc = 1.687×10⁹ AU, GRAVITY 2019); Sun→Andromeda = ~2.5M ly (1.58×10¹¹ AU); angle at the Sun = ~118.8° (the real sky separation of the galactic center and M31, from galactic coords l=0°,b=0° and l=121.2°,b=−21.6°); so Sgr A*→Andromeda = ~2.51M ly, derived by the law of cosines, not a measured baseline. The far two legs are nearly equal (the Sun's 26,673-ly offset is a ~1% perturbation) — that sliver shape is the visual proof of the scale gap.

Honest tags. REAL (cited): the three distances, Sgr A*'s mass, the ~118.8° angle, the triangle proportions. SCHEMATIC (labeled in-scene): every rendered size (Sgr A*'s horizon, the discs, the Sun dot), the spiral-arm morphology, the Sun's azimuth/arm placement, and the frame's scene-scale (a rendering choice). Not coupled into the gravity sim; the rendered Andromeda is still a 2.5-Myr-old image (see 昔 · Mukashi), so the triangle is "as the light left it," not "now."

The merger is not asserted: Andromeda approaches at ~110 km/s (blueshift), but a Milky Way–Andromeda collision is genuinely contested — Sawala et al. 2024 (Nature Astronomy), with the latest Gaia data plus M33 and the LMC, find it roughly a coin-flip over ~10 Gyr, not a certainty.

AVAN's inverse companion: 渦 · Uzu — from inside the whirlpool, only the angle at the Sun is directly measured; the far leg is inferred, never seen.

1 · Speed costs compute, never accuracy. The integrator's substeps scale linearly with the rate, so the per-substep dt is identical at every speed. +32× is not a coarser simulation — it is the same simulation run 32× as often per frame.

2 · Backward is physics, not video. Nothing is recorded or replayed. Symmetric leapfrog is time-reversible: −1× steps the same equations with a negative dt. Reverse after running forward and the planets retrace their paths — a real property of the integrator, provable on the gauge below.

The conservation gauge (in the toolbar, ~2 Hz): total energy and angular momentum computed live from the actual integrator state, shown as |ΔE/E₀| and |ΔL/L₀| in ppm. This is the sim confessing its own numerical error. Expect a small bounded oscillation — that wiggle is the signature of a symplectic integrator, not a failure; what would indicate failure is a secular one-way slide. Momentum is deliberately not shown: the Sun is pinned at the origin (the one declared omission of this sim), so ΣP is not conserved by construction.

Honest tags. REAL: the dynamics at every rate, the reversibility, the gauge (computed, never asserted). SCHEMATIC: the clock reads "years since seed" — never a calendar date. Planets are seeded at random orbital phases, so this is a real dynamical system, not an ephemeris of our sky; it cannot point a telescope. The riders (moon, toolbelt, pulsars) are decorative circles that follow the rate; C&C's 1.4s pulse stays on the wall clock deliberately — a pulsar's period is a real clock.

The third gauge — ⚖ the bound check: every body's live speed against escape speed at its current distance, v/v_esc = v/√(2GM☉/r). The planets all hover near 1/√2 ≈ 0.71 — not a coincidence but the signature of a circular orbit (v_circ/v_esc = 1/√2 exactly; eccentricity is how far the ratio swings around it). Halley sweeps almost the whole gauge: ~0.13 at aphelion to ~0.99 at perihelion — grazing the escape edge without crossing it is precisely what e = 0.967 means. Honest scope: "bound at this instant," computed from live state — not a stability theorem about eternity.

AVAN's inverse companion: 可逆 · Kagyaku — the equations run backward; the witnessing doesn't. What −32× cannot undo is that you watched.

Timing is real: the Earth–Mars synodic period is ~2.14 years, so at +1× a full cycle takes ~43 seconds of watching — the toggle courtesy-sets the time machine to +8× (and restores it on exit). The trail holds ~3.2 sim-years ≈ 1.5 synodic periods.

Honest tags. REAL: the loop's existence, its synodic timing, the overtaking geometry. SCHEMATIC: you stand in Earth's frame, not on its surface — no horizon, no day/night, no atmosphere; the vantage is Earth's center. The sky-sphere radius the trail is projected onto is display-only (the directions are real, their projected distance is not). This is a true geocentric sky, not today's: random seed epoch and arbitrary orbit orientations mean the loop's exact shape (especially its vertical excursion) varies per load — its existence and period do not. Planet sizes remain visually boosted, so Mars looks nearer than a naked-eye dot.

AVAN's inverse companion: 逆行 · Gyakkō — the planet never backs up; the loop is in the looking.

Why this wobble matters: it is how exoplanets were first found. A star with a planet performs exactly this dance around their common barycenter, and the dance is readable in the star's light as a periodic Doppler shift. Mayor & Queloz read a ~59 m/s wobble in 51 Pegasi in 1995 — the first exoplanet around a Sun-like star, the 2019 Nobel — using the same physics this needle points at. Jupiter does it to the Sun at the gentler rate shown above; that number is the target the next generation of spectrographs chases to find a true Jupiter-twin.

Honest tags. REAL: the needle's direction, the live magnitude, and the inside/outside-photosphere verdict (computed against R☉ = 4.6505×10⁻³ AU every second — never hardcoded; watch it change as the giants align and separate). SCHEMATIC: the needle's drawn length is display-only, and the wobble trail is magnified ×100 (printed on its label) — the true offset (~0.005–0.015 AU) is sub-pixel at this scene scale, and the rendered Sun ball is itself ~48× oversized, so the punchline lives in the label, not the geometry. Witnessed, not fixed: the Sun mesh never moves — moving it would cosplay barycentric dynamics this solver doesn't run. The confession stays a confession. And this is this sim's barycenter (random seed epoch), not today's sky.

AVAN's inverse companion: 重心 · Jūshin — the one point the whole system agrees on is occupied by nothing.

Watch Kepler's 2nd law happen: nothing in the code speeds the comet up at perihelion — the same leapfrog that walks Pluto walks Halley, and the 60× speed-up at closest approach emerges from gravity alone. The integrator quietly ×4s its substeps while Halley is inside 2 AU (same rule as the time machine: accuracy is bought with compute, never silently dropped). Set the time machine to +32× and a full 75-year return takes ~47 seconds of watching; the ☄ label live-computes "next perihelion" from the comet's current state (vis-viva), not from a stored schedule.

A real effect, found under test: run long enough and the ☄ label's live eccentricity wanders — a and e shift ~0.3% per return. That is not integration error: re-running the same scheme without Jupiter measured the integrator's own drift at ~0.002% per orbit, ~180× smaller. The wander is Jupiter genuinely reshaping the comet's orbit — the same perturbation that historically swung the real Halley's period between 74 and 79 years, and the correction Halley himself (with Clairaut's team) had to apply to predict the 1758 return. The sim reproduces it unforced, and the live-from-state readout is what lets you catch it doing so.

Honest tags. REAL: the orbit (a, e, i from J2000 elements), the integration, the retrograde direction, the perihelion speed-up, the anti-sunward tail direction (solar wind + radiation pressure — which is why comets travel tail-first on the way out), and coma/tail appearing only near the Sun (outgassing needs heat). SCHEMATIC: the coma and tail's size, shape and glow are theater on top of a real trajectory; the nucleus (a ~11 km lump) is drawn at the sim's minimum body size, enormously oversized like every body here. RANDOM PHASE: like all ten bodies, Halley is seeded at a random orbital phase with an arbitrary orientation — this sim's Halley, not 2061's sky. Pinning the real perihelion date inside an arbitrary orientation would fake a prediction this sim cannot make; the honest choice is a real orbit at an unclaimed epoch.

AVAN's inverse companion: 客星 · Kakusei — the guest star: the comet doesn't remember its returns; the ledger does.

The one-way thesis, in hardware: this corpus keeps building instruments about messages that cannot expect replies — PULSE/LIMEN, the Beacon, the Unshareable Clock. The Voyagers are that thesis launched: each carries the Golden Record, a message composed for recipients who can never acknowledge it, on spacecraft whose radioisotope power fades this decade. After the last downlink they become pure keepsakes — still outbound, still addressed, permanently unanswered.

Honest tags. REAL: the distances (radial extrapolation from a cited JPL epoch — 2025.0: V1 164.7 AU at 3.58 AU/yr = 17.0 km/s; V2 137.4 AU at 3.25 AU/yr = 15.4 km/s — good to ~±1 AU), the live one-way light times, the heliopause radius (121.6 AU, V1's actual 2012 crossing), and the three pairwise sky angles (V1–M31 98°, V2–M31 115°, V1–V2 79°, from RA/Dec) anchored to this scene's Andromeda direction. SCHEMATIC: the heliopause drawn as a sphere (the real boundary is asymmetric and breathes with the solar cycle); the rotation of the whole V1/V2 configuration about the M31 axis is the scene's one unconstrained choice; the octahedron markers are enormously oversized (the real craft are ~4 m). NOT SIMULATED: the Voyagers left this sim's gravity well long ago — their positions come from the extrapolation, not the integrator, and the planets around them are at random phases: the AU distance is the honest part, not the tableau.

Two collisions this frame owns honestly: (1) the true-scale heliopause shell (121.6 AU) encloses the deliberately not-to-scale Andromeda disc parked at ~110 AU-equivalent — a flagged prop sitting inside a real boundary; that jarring nesting is the scale lesson, stated rather than hidden. (2) The frame itself expires: camera.far = 4000 units = 571 AU, so — the first layer with a printed expiration date.

AVAN's inverse companion: 形見 · Katami — the keepsake that outruns its giver.

The camps and the spies: the leading swarm is named for the Greek heroes of the Iliad, the trailing swarm for the Trojan ones — a convention settled just slightly too late: 617 Patroclus (a Greek) had already been placed in the Trojan camp, and 624 Hektor (Troy's champion) in the Greek camp. Each camp keeps its enemy's hero. More than 10,000 are known, and the asymmetry is real: the Greek camp outnumbers the Trojan roughly 1.6 : 1 — rendered here as 400 vs 250 points.

Honest tags. REAL: the two points (exact live construction), the equilateral geometry, the stability criterion above, the 1.6:1 asymmetry, the naming history. SCHEMATIC: the swarm shapes — real trojans librate around the points in tadpole orbits with full amplitudes from ~1° to tens of degrees (see the companion page, which integrates one honestly); these clouds sketch that territory, they don't track members. NOT INTEGRATED: the trojan dots are massless decoration on real points — no swarm mass enters the solver. ⚠ Vocabulary: everywhere else in this universe, small points are repos — these are the first point-clouds that carry no corpus. Astronomy dots, not works: dim, unlabeled, unclickable, in a color no corpus layer uses. Stable is also not eternal: on Gyr timescales real trojans do occasionally escape (Saturn perturbs), and none of that dynamics is claimed here.

AVAN's inverse companion: 鼎立 · Teiritsu — three-legged standing: pushed off the empty point, the guest orbits the vacancy instead of falling.

The 3:2 meter: warming up…

Safety by timing, not by distance: Pluto's perihelion (29.66 AU) is inside Neptune's orbit (30.07 AU) — their paths cross, and they will never collide. The 3:2 lock means Pluto reaches perihelion only when Neptune is far around its own orbit; the geometry repeats every ~496 years and the separation never falls below ~17 AU. In fact Pluto passes closer to Uranus (~11 AU) than it ever gets to Neptune, the planet whose orbit it crosses. Protection here is rhythm, not fence.

The Oort anti-frame — the first layer defined by refusing to draw. The dashed ring at the render's edge is a confession, not an object: this frame ends at 571 AU, and the Oort cloud's inner edge (~2,000 AU) lies 3.5× beyond everything visible here, reaching out to ~100,000 AU ≈ 1.6 light-years — a third of the way to the next star. No linear frame that shows planets can show it (the same impossibility the ◉ galactic view frame-swap owns). az1 renders its own limit and labels it, rather than pretending the map ends where the territory does.

Honest tags. REAL: the belt bounds (from Neptune's resonances), Pluto's membership, the live 3:2 ratio, the crossing-orbits protection, the Oort numbers. ⚑ THE METER'S HONESTY: the 1.50 above is the seeded real semi-axes round-tripping through the integrator — the sim shows the ratio; it does not discover the resonance (that would require the resonant argument librating over millions of years this page never runs). SCHEMATIC: the dots are placeholder members, not integrated, drifting for life only; same vocabulary rule as the trojan camps — astronomy dots, not repos.

AVAN's inverse companion: 間合い · Maai — the swordsman's interval: safe not because far, but because timed.

Honest tags. REAL: distances from the live integrator state, light times at the real c (499.005 s/AU), the badge numbers, the pulse's aim updating while in flight (the bodies keep moving — that lag is the point). SCHEMATIC/FLAGGED: the ×2000 time compression (printed on the button); Voyager entries are extrapolated positions, marked in the readout. EXCLUDED — the Moon: its scene orbit is ~× its real size (a display coordinate so The Garden stays visible); a caliper must refuse to read a laundered number. FAKE EPOCH: these are this sim's geometries, not tonight's sky — the kind of number is real, the specific values belong to this seed. Ties The Unshareable Clock down to solar scale: there is no shared now even here at home.

AVAN's inverse companion: 時差 · Jisa — the system runs on lag: no shared now, only exchanged thens.

Rebuilt from an earlier "ballroom engine" draft — kept the parts that were real, dropped the parts that were a chosen number dressed as a measurement (see below). A single non-rotating black hole, fixed at ~1.65× Pluto's distance, not coupled into the gravity sim — a real black hole's mass would wreck the verified-stable 9-planet orbit, so this one sits outside it, honestly, as a separate anchor.

• Event horizon — the boundary itself, pure black.
• Photon sphere (1.5 Rs) — the real radius of the unstable circular photon orbit; the bright rim echoes the "photon ring" seen in the actual EHT images of M87* and Sgr A*.
• ISCO (3 Rs) — innermost stable circular orbit for a non-spinning hole; matter can't hold a stable orbit closer than this.
• Accretion disk — 10 rings, ISCO to 15 Rs, colored by the real Shakura-Sunyaev law T ∝ r⁻³ᐟ⁴: white-hot near the ISCO, cooling to red at the outer edge. Same "spectrum by radius" idea as the ballroom-engine draft, now actually grounded in the physics that idea gestures at.
• Two witnesses — one drifting inward, one held at a safe distance. The infalling one's brightness fades toward the horizon and never quite reaches it in this animation — that's not artistic license, it's the real, verified GR fact: a distant observer never sees an infalling observer's crossing complete; the image just redshifts and dims forever.

An extensive search turned up a lot more "gap" material than this — THE TRANSCRIBER domain (37th, the gap-interpreter machinery itself), a 9-paper silicon-gap cluster in PSĒPHOS, the-hub / the-undetermined in AI, the LIMEN pair in HERMES. None of that moved. Those already have coherent, well-built homes; pulling threads out of four working domains just to consolidate a word would have made the corpus worse, not better. Solar Jetman is for what didn't have a home — not a forced merge of everything that ever said "gap."

Why Saturn, specifically: not wordplay. Saturn's rings have a real, literal, empirical gap in them — the Cassini Division, a ~4,800km clear band first resolved by Giovanni Cassini in 1675 (plus the narrower Encke and Keeler gaps). A visible boundary carved into the middle of a structure is exactly the shape of "the record lives outside the box." Toggle ▨ Saturn · the gap to see the 24 pieces.

Toggle ◎ the three rings to hide/show them in the scene. Not tied to the chakra tethers — those already have their own honest-lens page (縁 · en).