v1.4.0

Added

• A third sealed L3 request now binds only projectile_or_impact_icf to exact
  SCPN-ICF-IMPACT-CORE fixture and reproducible wheel custody. Impact timing,
  derived asymmetry, noncyclic trajectory/outcome, and numerical model phase
  remain distinct, while every other ICF configuration inherits no evidence or
  authority.
• A host-independent materialisation CLI now reconstructs sealed device
  physical-evidence requests from exact local producer fixtures without
  importing producer code. The second L3 instance binds only ion_beam_icf
  from SCPN-ICF-BEAM-CORE and proves that its shared plan transfers no evidence
  or authority to pulsed-electron-beam ICF.
• A family-neutral, byte-canonical physical-evidence request now derives one
  configuration-specific producer boundary from an accepted diagnostic-plan
  review. The first materialised request covers direct-drive laser ICF,
  preserves event-relative, derived-cyclic, noncyclic, and numerical-only
  meanings, names thirteen missing evidence obligations, and grants no
  observation, phase, CONTROL, execution, or actuation authority.
• Registry release 1.1.0 preserves the immutable 32-configuration 1.0.0
  corpus and adds namespaced lattice-confinement and muon-catalysed-fusion
  identities, configuration-specific observability candidates, evidence-ranked
  primary sources, and version-aware review of historical producer pins. The
  refreshed 22-project portfolio refuses stale digest and ownership contracts,
  keeps both new architecture projects planless, and grants no physical,
  CONTROL, execution, or actuation authority.
• Exact diagnostic-plan envelope dispatch now preserves historical 1.1.0
  while admitting 1.2.0 only through its distinct signal, frame-
  transformation, and clock-topology shapes. The exact TOKAMAK pilot fixture
  exercises declaration-depth validation without allowing signal text,
  coordinate mappings, or clock declarations to create physical evidence,
  phase, CONTROL authority, action, or actuation.
• A byte-canonical L1 FRC-compression-MIF physical-payload request binds the
  existing SCPN-MIF-CORE merge-compression adapter as simulation-only, pins
  the semantic and observability registries, carries four unselected
  magneto-inertial candidates, and names twelve missing physical-evidence
  obligations. Its sealed request identity is linked from the producer-
  evidence priority register, while schema allocation, observation, phase,
  CONTROL, execution, and actuation remain fail closed.
• A byte-canonical L1 conventional-tokamak physical-payload request binds the
  existing SCPN-FUSION-CORE TORAX adapter as simulation-only, pins the semantic
  and observability registries, carries four unselected candidates, and names
  twelve missing physical-evidence obligations. Its sealed request identity is
  linked from the producer-evidence priority register, while schema allocation,
  observation, phase, CONTROL, execution, and actuation remain fail closed.
• A byte-canonical FAIR-MAST phase-qualification request binds the existing
  physical-source review, exact FUSION source/package identity, observability
  registry, four unselected spherical-tokamak candidates, and twelve explicit
  missing producer-evidence obligations. It remains blocked, review-only, and
  unable to select a phenomenon, admit an observation, infer phase, request
  CONTROL admission, permit execution, or authorise actuation. The companion
  priority-register correction now records the already-existing FAIR-MAST
  portable review adapter instead of routing producer work toward a duplicate.
• A digest-sealed reactor producer-evidence priority register joins all 34
  configurations to exact external context, occurrence evidence,
  observability candidates, semantic-ingress state, diagnostic-plan custody,
  all 24 upstream reactor projects plus CONTROL, producer ownership, and
  physical-evidence gaps. Five non-scalar readiness lanes distinguish one
  existing physical-source qualification, two exercised-adapter extensions,
  two missing plans, 13 accepted-plan physical intakes, and 16 configuration
  rows blocked on refused plans.
  External evidence rank never changes the lane or creates physical signal,
  phase, CONTROL, action, actuation, or machine-protection authority.
• An evidence-ranked reactor technology and diagnostic atlas maps all 34
  registered configurations across nine confinement families to 37 exact,
  configuration-specific primary sources. It distinguishes integrated fusion,
  integrated plasma, component or driver, engineering, and concept evidence
  from diagnostic capability and SCPN evidence admission. Every row remains
  review-only and fail-closed for physical observation, physical phase,
  CONTROL admission, action, actuation, and machine-protection bypass.
• A digest-sealed reactor diagnostic-plan portfolio status register records
  exact revisions, manifest and fixture hashes, configurations, plans, custody,
  and fail-closed review results for all 22 Reactor Systems device projects.
  Seven objects are accepted, thirteen are refused, two architecture projects
  have no plan, and zero current objects have byte-identical SPO custody.
  The schema and tests prohibit custody overclaim, inference, physical-evidence
  promotion, CONTROL authority, or actuation.
• An exhaustive, digest-sealed reactor configuration evidence matrix joins all
  34 registered configurations across nine confinement families to their exact
  observability candidates, semantic-ingress profiles, and occurrence-ledger
  evidence. It distinguishes five configurations with source evidence from 29
  explicit producerless rows, records only two verified simulation review
  adapters, and forbids topology-based evidence inheritance, physical-phase
  promotion, action, or actuation.
• A canonical reactor signal occurrence ledger binds 39 externally meaningful
  phase, mode, regime, event, clock, and semantic-boundary groups across exact
  SPO, SCPN-FUSION-CORE, SCPN-MIF-CORE, and SCPN-CONTROL revisions to source
  symbols and Git-blob SHA-256 values. Its strict schema and tests separate
  direct source occurrence from SPO interpretation, retire superseded atlas
  statements, preserve unresolved gaps, and forbid physical-observation,
  physical-phase, action, or actuation escalation.
• A full-fidelity FAIR-MAST magnetic physical-source review intake verifies
  exact SCPN-FUSION-CORE archive, qualification, source revision, and producer-
  wheel identities; replays all 72 arrays, 11 measurement families, 132
  channel records, and four derived clock grids; and preserves unresolved
  calibration, geometry, event, observation-operator, quality, clock,
  uncertainty, and validity boundaries without admitting a phase observation,
  semantic ingress, classification, CONTROL intent, action, or actuation.
• A reactor-family-neutral, byte-canonical and digest-sealed device diagnostic-
  plan review intake validates exact producer source, wheel, manifest,
  envelope, plan, registry, candidate, carrier, evidence-slot, frame, and clock
  identities without importing sibling source. Exact tokamak, dense-plasma-
  focus, theta-pinch, Z-pinch, MagLIF, mechanical-or-liquid-liner MIF,
  plasma-jet MIF, laser-ICF, ion/electron-beam ICF, and projectile-impact ICF
  fixtures exercise distinct event-cycle, complex-mode, bounded-feature, and
  numerical-phase declarations.
  Their typed signal and clock rows remain synthetic design declarations and
  cannot create observation, classification, semantic ingress, CONTROL intent,
  action, or actuation.

Security

• Raise the Tornado development/notebook floor to 6.5.8 and the Material for
  MkDocs floor to 9.7.7, closing the current request-parsing denial-of-service,
  cookie-validation, and documentation-search DOM XSS advisories across every
  canonical interpreter and platform lock.

Fixed

• Restore the exact strict-mypy audit-envelope boundary across both legacy and
  current protobuf stub surfaces, and document every reactor-semantics helper
  so the source-wide 100 percent documentation gate remains fail closed.
• Re-seal six diagnostic-plan review fixtures after the producer-owned manifest
  digest corrections, and expose all new reactor-semantic modules through the
  governed API documentation surface with complete NumPy-style contracts.
• Device diagnostic-plan review now accepts catalogue-declared direct cyclic
  channels on facility-monotonic clocks. A byte-exact BEAM-TARGET producer
  fixture guards the public intake while preserving synthetic, review-only,
  non-observational, non-phase, non-CONTROL, and non-actuating authority.
• Single-node algebraic connectivity now returns its exact zero boundary
  before invoking optional eigensolver backends, avoiding an invalid Mojo
  LAPACK call. Generated correlation-dimension radii now remain below the
  estimated attractor diameter, preserving strict-threshold parity across
  NumPy and Mojo round-off. Obsolete protobuf typing suppressions that failed
  the current strict mypy gate were also removed, and the polyglot
  documentation guard now follows the reusable integration workflow where the
  native documentation jobs are defined. Mojo phase entropy now uses its
  documented 1e-9 boundary tolerance and clamps only accepted round-off to
  the exact Shannon-entropy range. Mojo torus output also canonicalises a
  post-wrap 2π round-off value to zero while retaining the strict [0, 2π)
  output contract.