If the claim “a derived Standard-Model skeleton” set off an alarm, good — keep it on. This page is the five-minute orientation: what is claimed, what is not, what would falsify it, what is machine-checked, what is still open, and how to reproduce one result yourself. Nothing here is sold; a typed claim stack is dissected.
A two-input discrete compiler whose algebraic kernel derives the Standard-Model skeleton — the gauge group, three families, hypercharge and the flavor invariants — and several dimensionless readouts (α⁻¹, the neutrino angles, β_rad). It is built only from two axioms: the seam constant c₃ = 1/(8π) and the carrier rank g_car = 5. E₈ is the compiler’s internal audit lattice, not a physical gauge group.
It is not an E₈ gauge theory (so the Distler–Garibaldi / Coleman–Mandula no-gos do not apply), and not a certified strict physical theory of everything end-to-end — the keystone SEAM.EQUIV.01 is closed only on its MMST route (SEAM.EQUIV.MMST.01), modulo a cited published theorem (the cited continuum scaling-limit existence, v336; extension leg on the peer-reviewed crossed-product package, realisation at invariant level, v469; the twistor route SEAM.EQUIV.TWISTOR.01 and the unconditional parent stay [O]). The local Einstein equation is parameter-free, and the ambient, non-perturbative quantum-gravity measure is discharged as a redundancy [C] (a certification object, not missing dynamics), not an open hole. The absolute mass scale is a single declared dimensionful anchor (v_geo), and m_p/m_e is explicitly not claimed as a compiler power. No experimental value is ever used as an input — only as a comparison row.
Any single one of: the α fixed-point equation F_U(1)(α) = 0 failing or admitting a second admissible root; a robust neutron-EDM signal (θ_eff = 0 is structural); a second light seam-even Higgs doublet (N_Φ = 1); a robust tensor ratio r ≳ 0.01; a robust w ≠ −1; a JUNO solar angle clearly away from 0.307; or any verification script failing to reproduce its claim. The full kill board →
Every exact identity, lattice/Lie theorem and numerical fixed point is re-derived from the two axioms by 569 Python checks (run_all.py ends ALL CHECKS PASSED), mirrored on an independent Wolfram path, and the carrier algebra is formalised in Lean 4 with 0 sorry. Every claim is typed in a single versioned ledger — if the prose and the ledger disagree, the ledger wins. The verification stack →
Exactly three named interfaces, none hidden: v_geo (one dimensionful scale anchor, the same nature as 1/G, [O]), G_net (the seam-net metric inclusion — algebra discharged to [E]; the seam keystone now [C], closed modulo cited theorems), and F_transfer (the source→pole/relic/cosmology transport — Koide, η_B, the axion relic, m_p/m_e are its four instances, now runnable typed solvers v371–v375 [C]). The whole boundary-QFT layer (Dirac, cutoff, gauging, glue) collapses onto the same G_net premise via the Modular Spectral Closure (v258–v261) — one relative object, no new open item. That premise is now closed modulo cited theorems via the Seam Equivalence Theorem SEAM.EQUIV.01 (the raw RP seam IS the holomorphic (E₈)₁ net at τ=i): the target net is pinned at every computable level by an explicit lattice model (v367/v368) and the S3 closure stack (v376–v379, ground-state witnesses v489/v490), Lean-pinned (FORM.SEAM.MMST.01) to the published MMST/Adamo theorems, leaving [O] = the cited continuum scaling-limit existence only (v336; extension leg on the peer-reviewed crossed-product package, realisation at invariant level, v469; stays [O]); its conformal-deck face QGEO.SYM.01 is a corollary. The step-by-step reduction lives on the changelog. The open residual →
Open the dependency graph and click the α⁻¹ node: the real Python runs in your browser (via Pyodide / WebAssembly) and re-derives α⁻¹ = 137.0359992… as the unique root of the parameter-free cubic — no install, no fit. Run it now → Or clone the repository and run the whole suite locally.