fix(lab): preserve replay view and restore SLAM semantics

apps/control-station/src/components/laboratory/LaboratoryMetricEvidenceScene.tsx

@@ -41,8 +41,25 @@ export interface LaboratoryMetricCorridorVisual {
   halfWidthM: number;
 }
 
+export type LaboratoryMetricCellState =
+  | "unobserved"
+  | "ground-support"
+  | "nonground-occupied"
+  | "unknown-rejected";
+
+export interface LaboratoryMetricCellEvidence {
+  centerBodyXyM: readonly [number, number];
+  zBoundsM: readonly [number | null, number | null];
+  state: LaboratoryMetricCellState;
+}
+
 export interface LaboratoryMetricLegendEntry {
-  id: LaboratoryMetricDecision | "context" | "local-surface" | "rolling" | "low-step";
+  id: LaboratoryMetricDecision
+    | LaboratoryMetricCellState
+    | "context"
+    | "local-surface"
+    | "rolling"
+    | "low-step";
   label: string;
 }
 
@@ -197,6 +214,9 @@ LaboratoryMetricEvidenceSceneHandle,
   pointSemanticClassIds?: readonly (number | null)[];
   semanticClasses?: readonly RecordedEvidenceSemanticClass[];
   semanticPalette?: readonly RecordedEvidenceSemanticPaletteEntry[];
+  classifiedCells?: readonly LaboratoryMetricCellEvidence[];
+  classifiedCellSizeM?: number;
+  showClassifiedCells?: boolean;
 }
 >(function LaboratoryMetricEvidenceScene({
   pointCloudBodyXyzM,
@@ -214,6 +234,9 @@ LaboratoryMetricEvidenceSceneHandle,
   pointSemanticClassIds,
   semanticClasses,
   semanticPalette,
+  classifiedCells = [],
+  classifiedCellSizeM = 0.45,
+  showClassifiedCells = true,
 }, ref) {
   const hostRef = useRef<HTMLDivElement | null>(null);
   const sceneRef = useRef<THREE.Scene | null>(null);
@@ -374,6 +397,63 @@ LaboratoryMetricEvidenceSceneHandle,
       ));
     }
 
+    if (showClassifiedCells && classifiedCells.length) {
+      const cellsByState = new Map<LaboratoryMetricCellState, LaboratoryMetricCellEvidence[]>();
+      for (const cell of classifiedCells) {
+        const cells = cellsByState.get(cell.state) ?? [];
+        cells.push(cell);
+        cellsByState.set(cell.state, cells);
+      }
+      for (const [state, cells] of cellsByState) {
+        const color = state === "ground-support"
+          ? tokenColor(host, "--nodedc-success-rgb", [181, 255, 90])
+          : state === "nonground-occupied"
+            ? tokenColor(host, "--nodedc-danger-rgb", [255, 104, 112])
+            : state === "unknown-rejected"
+              ? tokenColor(host, "--nodedc-warning-rgb", [255, 197, 92])
+              : tokenColor(host, "--nodedc-text-muted", [96, 99, 106]);
+        const geometry = new THREE.BoxGeometry(
+          classifiedCellSizeM * 0.92,
+          1,
+          classifiedCellSizeM * 0.92,
+        );
+        const material = new THREE.MeshBasicMaterial({
+          color,
+          transparent: true,
+          opacity: state === "unobserved" ? 0.035 : state === "ground-support" ? 0.12 : 0.24,
+          depthWrite: false,
+        });
+        const mesh = new THREE.InstancedMesh(geometry, material, cells.length);
+        const matrix = new THREE.Matrix4();
+        const scale = new THREE.Vector3(1, 1, 1);
+        const rotation = new THREE.Quaternion();
+        cells.forEach((cell, index) => {
+          const minimum = cell.zBoundsM[0];
+          const maximum = cell.zBoundsM[1];
+          const height = minimum === null || maximum === null
+            ? 0.018
+            : Math.max(0.018, maximum - minimum);
+          const centerZ = minimum === null || maximum === null
+            ? -0.012
+            : (minimum + maximum) / 2;
+          const [sceneX, sceneY, sceneZ] = scenePoint([
+            cell.centerBodyXyM[0],
+            cell.centerBodyXyM[1],
+            centerZ,
+          ]);
+          scale.set(1, height, 1);
+          matrix.compose(
+            new THREE.Vector3(sceneX, sceneY, sceneZ),
+            rotation,
+            scale,
+          );
+          mesh.setMatrixAt(index, matrix);
+        });
+        mesh.instanceMatrix.needsUpdate = true;
+        content.add(mesh);
+      }
+    }
+
     for (const obstacle of obstacles) {
       if (
         (
@@ -451,7 +531,10 @@ LaboratoryMetricEvidenceSceneHandle,
     pointSemanticClassIds,
     semanticClasses,
     semanticPalette,
+    classifiedCells,
+    classifiedCellSizeM,
     showCurrentIncrement,
+    showClassifiedCells,
     showLocalSurface,
     showRollingMap,
     showLowStep,
@@ -563,7 +646,9 @@ LaboratoryMetricEvidenceSceneHandle,
     });
   })();
   const metricLegendEntries = laboratoryMetricLegendEntries({
-    pointCloudCount: pointCloudBodyXyzM.length,
+    pointCloudCount: pointSemanticClassIds?.every((item) => item !== null)
+      ? 0
+      : pointCloudBodyXyzM.length,
     localSurfaceCount: localSurfaceBodyXyzM.length,
     obstacles,
     showCurrentIncrement,
@@ -571,6 +656,28 @@ LaboratoryMetricEvidenceSceneHandle,
     showRollingMap,
     showLowStep,
   });
+  const classifiedLegendEntries = (() => {
+    if (!showClassifiedCells || !classifiedCells.length) return [];
+    const states = new Set(classifiedCells.map((cell) => cell.state));
+    return [
+      states.has("ground-support")
+        ? { id: "ground-support" as const, label: "Ground support" }
+        : null,
+      states.has("nonground-occupied")
+        ? { id: "nonground-occupied" as const, label: "Non-ground occupied" }
+        : null,
+      states.has("unknown-rejected")
+        ? { id: "unknown-rejected" as const, label: "Unknown / rejected" }
+        : null,
+      states.has("unobserved")
+        ? { id: "unobserved" as const, label: "Unobserved" }
+        : null,
+    ]
+      .filter((entry): entry is NonNullable<typeof entry> => entry !== null)
+      .filter((entry) => !semanticLegendEntries.some(
+        (semanticEntry) => semanticEntry.label === entry.label,
+      ));
+  })();
 
   return (
     <div className="laboratory-metric-evidence-scene">
@@ -581,6 +688,9 @@ LaboratoryMetricEvidenceSceneHandle,
         {metricLegendEntries.map((entry) => (
           <span key={entry.id} data-decision={entry.id}>{entry.label}</span>
         ))}
+        {classifiedLegendEntries.map((entry) => (
+          <span key={entry.id} data-decision={entry.id}>{entry.label}</span>
+        ))}
         {semanticLegendEntries.map((entry) => (
           <span
             key={entry.id}

apps/control-station/src/core/laboratory/advancedIndex.ts

@@ -46,6 +46,8 @@ import { fetchM48StaticOccupancyQualification } from "./m48StaticOccupancyQualif
 import { fetchM48R3StaticOccupancyShadow } from "./m48r3StaticOccupancyShadow";
 import { fetchM48SFixedClassDetectorResult } from "./m48sFixedClassDetector";
 import { fetchM48TRiskQualityResult } from "./m48tRiskQuality";
+import { fetchM49TgsFailClosedResult } from "./m49TgsFailClosed";
+import { fetchM49TgsFullShadowResult } from "./m49TgsFullShadow";
 
 export type AdvancedLaboratoryWorkId =
   | "m48-object-centric-quality"
@@ -54,6 +56,8 @@ export type AdvancedLaboratoryWorkId =
   | "m48r3-static-occupancy-shadow"
   | "m48s-fixed-class-detector"
   | "m48t-risk-quality-temporal"
+  | "m49-tgs-fail-closed-evidence"
+  | "m49-tgs-full-shadow"
   | "m47-reference-graph-shadow"
   | "m4-replay-threat"
   | "l3-pointpillars-visual-audit"
@@ -102,6 +106,8 @@ const WORK_IDS: readonly AdvancedLaboratoryWorkId[] = [
   "m48r3-static-occupancy-shadow",
   "m48s-fixed-class-detector",
   "m48t-risk-quality-temporal",
+  "m49-tgs-fail-closed-evidence",
+  "m49-tgs-full-shadow",
   "m47-reference-graph-shadow",
   "m4-replay-threat",
   "l3-pointpillars-visual-audit",
@@ -145,6 +151,8 @@ const RESULT_PREFIX: Readonly<Record<AdvancedLaboratoryWorkId, string>> = {
   "m48r3-static-occupancy-shadow": "m48r3-static-occupancy-shadow",
   "m48s-fixed-class-detector": "m48s-fixed-class-detector-lab",
   "m48t-risk-quality-temporal": "(?:m48t-risk-quality-temporal-lab|m48q-native-risk-quality-lab)",
+  "m49-tgs-fail-closed-evidence": "m49-tgs-fail-closed",
+  "m49-tgs-full-shadow": "m49-tgs-full-shadow",
   "m47-reference-graph-shadow": "m47-reference-graph-lab",
   "m4-replay-threat": "m4-threat-replay",
   "l3-pointpillars-visual-audit": "l3-pointpillars-visual-audit",
@@ -196,6 +204,8 @@ export function emptyAdvancedLaboratoryResults(): AdvancedLaboratoryResults {
     m48r3StaticOccupancy: null,
     m48s: null,
     m48t: null,
+    m49Tgs: null,
+    m49TgsFull: null,
     m4Threat: null,
     l3: null,
     l31: null,
@@ -326,6 +336,8 @@ export function advancedLaboratoryResultAvailable(
     : workId === "m48r3-static-occupancy-shadow" ? results.m48r3StaticOccupancy !== null
     : workId === "m48s-fixed-class-detector" ? results.m48s !== null
     : workId === "m48t-risk-quality-temporal" ? results.m48t !== null
+    : workId === "m49-tgs-fail-closed-evidence" ? results.m49Tgs !== null
+    : workId === "m49-tgs-full-shadow" ? results.m49TgsFull !== null
     : workId === "m47-reference-graph-shadow" ? results.m47Graph !== null
     : workId === "m4-replay-threat" ? results.m4Threat !== null
     : workId === "l3-pointpillars-visual-audit" ? results.l3 !== null
@@ -393,6 +405,12 @@ export async function fetchAdvancedLaboratoryResult(
   } else if (workId === "m48t-risk-quality-temporal") {
     if (!resultId) throw new AdvancedLaboratoryContractError("M4.8T LAB identity не выбрана.");
     results.m48t = await fetchM48TRiskQualityResult(resultId, { fetcher, signal });
+  } else if (workId === "m49-tgs-fail-closed-evidence") {
+    if (!resultId) throw new AdvancedLaboratoryContractError("M4.9 TGS LAB identity не выбрана.");
+    results.m49Tgs = await fetchM49TgsFailClosedResult(resultId, { fetcher, signal });
+  } else if (workId === "m49-tgs-full-shadow") {
+    if (!resultId) throw new AdvancedLaboratoryContractError("M4.9 full TGS shadow identity не выбрана.");
+    results.m49TgsFull = await fetchM49TgsFullShadowResult(resultId, { fetcher, signal });
   } else if (workId === "m47-reference-graph-shadow") {
     if (!resultId) {
       throw new AdvancedLaboratoryContractError("M4.7 LAB identity не выбрана.");

apps/control-station/src/core/laboratory/advancedLaboratoryResults.ts

@@ -40,6 +40,8 @@ import type { M48StaticOccupancyQualificationResult } from "./m48StaticOccupancy
 import type { M48R3StaticOccupancyShadowResult } from "./m48r3StaticOccupancyShadow";
 import type { M48SFixedClassDetectorResult } from "./m48sFixedClassDetector";
 import type { M48TRiskQualityResult } from "./m48tRiskQuality";
+import type { M49TgsFailClosedResult } from "./m49TgsFailClosed";
+import type { M49TgsFullShadowResult } from "./m49TgsFullShadow";
 
 export interface AdvancedLaboratoryResults {
   m47Graph: M47ReferenceGraphLabResult | null;
@@ -49,6 +51,8 @@ export interface AdvancedLaboratoryResults {
   m48r3StaticOccupancy: M48R3StaticOccupancyShadowResult | null;
   m48s: M48SFixedClassDetectorResult | null;
   m48t: M48TRiskQualityResult | null;
+  m49Tgs: M49TgsFailClosedResult | null;
+  m49TgsFull: M49TgsFullShadowResult | null;
   m4Threat: M4ThreatReplayResult | null;
   l3: L3PointPillarsVisualAuditResult | null;
   l31: L31PointPillarsRavnovesResult | null;

apps/control-station/src/core/laboratory/advancedResults.ts

@@ -969,7 +969,7 @@ export async function fetchAdvancedLaboratoryResults({
   return {
     m47Graph: null, m48: null, m48SmallStatic: null, m48StaticOccupancy: null,
     m48r3StaticOccupancy: null,
-    m48s: null, m48t: null, m4Threat: null,
+    m48s: null, m48t: null, m49Tgs: null, m49TgsFull: null, m4Threat: null,
     l3: null, l31: null, l32: null, l33: null,
     e31,
     e32,

apps/control-station/src/core/laboratory/m49TgsFailClosed.ts

@@ -0,0 +1,349 @@
+import type { LaboratoryFetch } from "./advancedResults";
+
+const RESULT_ID = /^m49-tgs-fail-closed-[a-f0-9]{64}$/;
+
+export type M49TgsProfile = "current_increment" | "causal_rolling_1s";
+export type M49TgsStateCode = 0 | 1 | 2 | 3;
+export type M49TgsPointStateCode = 1 | 2 | 3;
+
+export interface M49TgsAnchorSummary {
+  anchorFrameIndex: number;
+  slot: number;
+  pointCount: number;
+  groundPointCount: number;
+  nongroundPointCount: number;
+  rejectedPointCount: number;
+  groundCellCount: number;
+  nongroundCellCount: number;
+  rejectedCellCount: number;
+  unobservedCellCount: number;
+  allPointsAccounted: true;
+}
+
+export interface M49TgsFailClosedResult {
+  resultId: string;
+  createdAtUtc: string;
+  source: {
+    sourceId: "RAVNOVES00";
+    sourceSessionId: "20260720T065719Z_viewer_live";
+    sourcePackSha256: string;
+    linkedVisualResultId: string;
+    anchorFrameIndices: readonly number[];
+  };
+  configuration: {
+    profileId: string;
+    configSha256: string;
+    coordinateFrame: "map-gravity-local";
+    primaryProfile: "causal_rolling_1s";
+    cellSizeM: number;
+    radiusM: number;
+  };
+  execution: {
+    worker: "Worker 006";
+    device: "cpu";
+    gpuUsed: false;
+    wrapperElapsedSeconds: number;
+  };
+  metrics: {
+    anchorCount: number;
+    anchorProfileCount: number;
+    allEligiblePointsAccounted: true;
+    costmapCellCount: number;
+    processWallCurrentP50Ms: number;
+    processWallCurrentMaxMs: number;
+    processWallRollingP50Ms: number;
+    processWallRollingMaxMs: number;
+    processMaxRssKib: number;
+    primary: readonly M49TgsAnchorSummary[];
+  };
+  acceptance: {
+    representationComplete: true;
+    allPointsAccounted: true;
+    aosAbsent: true;
+    gpuAbsent: true;
+    visualQualityAccepted: false;
+    traversabilityAccepted: false;
+  };
+  decision: {
+    state: "visual-review-required";
+    candidateRetained: true;
+    nextAction: string;
+  };
+  limitations: readonly string[];
+}
+
+export interface M49TgsAnchorSpatial {
+  resultId: string;
+  anchorFrameIndex: number;
+  sourceSequence: number;
+  profile: M49TgsProfile;
+  coordinateFrame: "map-gravity-local";
+  pointsXyzM: readonly (readonly [number, number, number])[];
+  pointStates: readonly M49TgsPointStateCode[];
+  costmap: {
+    cellSizeM: number;
+    radiusM: number;
+    centersXyM: readonly (readonly [number, number])[];
+    states: readonly M49TgsStateCode[];
+    zBoundsM: readonly (readonly [number | null, number | null])[];
+  };
+  allPointsAccounted: true;
+  aosUsed: false;
+}
+
+export class M49TgsContractError extends Error {}
+
+function objectValue(value: unknown, label: string): Record<string, unknown> {
+  if (!value || typeof value !== "object" || Array.isArray(value)) {
+    throw new M49TgsContractError(`${label}: ожидался объект.`);
+  }
+  return value as Record<string, unknown>;
+}
+
+function text(value: unknown, label: string): string {
+  if (typeof value !== "string" || !value.trim()) {
+    throw new M49TgsContractError(`${label}: ожидалась строка.`);
+  }
+  return value;
+}
+
+function exact<T extends string | boolean>(value: unknown, expected: T, label: string): T {
+  if (value !== expected) throw new M49TgsContractError(`${label}: нарушен контракт.`);
+  return expected;
+}
+
+function numberValue(value: unknown, label: string): number {
+  if (typeof value !== "number" || !Number.isFinite(value)) {
+    throw new M49TgsContractError(`${label}: ожидалось число.`);
+  }
+  return value;
+}
+
+function integer(value: unknown, label: string): number {
+  const parsed = numberValue(value, label);
+  if (!Number.isSafeInteger(parsed) || parsed < 0) {
+    throw new M49TgsContractError(`${label}: ожидалось целое число.`);
+  }
+  return parsed;
+}
+
+function sha256(value: unknown, label: string): string {
+  const parsed = text(value, label);
+  if (!/^[a-f0-9]{64}$/.test(parsed)) {
+    throw new M49TgsContractError(`${label}: неверный SHA-256.`);
+  }
+  return parsed;
+}
+
+function resultId(value: unknown): string {
+  const parsed = text(value, "M49 result id");
+  if (!RESULT_ID.test(parsed)) throw new M49TgsContractError("M49 identity недопустима.");
+  return parsed;
+}
+
+function numbers(value: unknown, size: number, label: string): number[] {
+  if (!Array.isArray(value) || value.length !== size) {
+    throw new M49TgsContractError(`${label}: неверная размерность.`);
+  }
+  return value.map((item) => numberValue(item, label));
+}
+
+function anchor(value: unknown, label: string): M49TgsAnchorSummary {
+  const row = objectValue(value, label);
+  exact(row.profile_id, "causal_rolling_1s", `${label}.profile`);
+  exact(row.all_points_accounted, true, `${label}.accounting`);
+  const parsed: M49TgsAnchorSummary = {
+    anchorFrameIndex: integer(row.anchor_frame_index, `${label}.anchor`),
+    slot: integer(row.slot, `${label}.slot`),
+    pointCount: integer(row.point_count, `${label}.points`),
+    groundPointCount: integer(row.ground_point_count, `${label}.ground points`),
+    nongroundPointCount: integer(row.nonground_point_count, `${label}.nonground points`),
+    rejectedPointCount: integer(row.rejected_point_count, `${label}.rejected points`),
+    groundCellCount: integer(row.ground_cell_count, `${label}.ground cells`),
+    nongroundCellCount: integer(row.nonground_cell_count, `${label}.nonground cells`),
+    rejectedCellCount: integer(row.rejected_cell_count, `${label}.rejected cells`),
+    unobservedCellCount: integer(row.unobserved_cell_count, `${label}.unobserved cells`),
+    allPointsAccounted: true,
+  };
+  if (
+    parsed.groundPointCount + parsed.nongroundPointCount + parsed.rejectedPointCount
+      !== parsed.pointCount
+    || parsed.groundCellCount + parsed.nongroundCellCount
+      + parsed.rejectedCellCount + parsed.unobservedCellCount !== 2244
+  ) {
+    throw new M49TgsContractError(`${label}: fail-closed accounting нарушен.`);
+  }
+  return parsed;
+}
+
+export async function fetchM49TgsFailClosedResult(
+  id: string,
+  { fetcher = fetch, signal }: { fetcher?: LaboratoryFetch; signal?: AbortSignal } = {},
+): Promise<M49TgsFailClosedResult> {
+  if (!RESULT_ID.test(id)) throw new M49TgsContractError("M49 identity недопустима.");
+  const response = await fetcher(
+    `/api/v1/laboratory/m49/tgs-fail-closed/${encodeURIComponent(id)}`,
+    { method: "GET", headers: { Accept: "application/json" }, signal },
+  );
+  if (!response.ok) throw new M49TgsContractError(`M49 TGS недоступен: HTTP ${response.status}.`);
+  const payload = objectValue(await response.json(), "M49 TGS");
+  exact(payload.schema_version, "missioncore.m49-tgs-fail-closed-view/v1", "M49 schema");
+  exact(payload.result_id, id, "M49 result");
+  exact(payload.ground_truth, false, "M49 ground truth");
+  exact(payload.access, "read-only", "M49 access");
+  const source = objectValue(payload.source, "M49 source");
+  const configuration = objectValue(payload.configuration, "M49 configuration");
+  const execution = objectValue(payload.execution, "M49 execution");
+  const metrics = objectValue(payload.metrics, "M49 metrics");
+  const acceptance = objectValue(payload.acceptance, "M49 acceptance");
+  const decision = objectValue(payload.decision, "M49 decision");
+  if (!Array.isArray(source.anchor_frame_indices) || !Array.isArray(metrics.primary)) {
+    throw new M49TgsContractError("M49 anchors: ожидался массив.");
+  }
+  if (!Array.isArray(payload.limitations)) {
+    throw new M49TgsContractError("M49 limitations: ожидался массив.");
+  }
+  const anchorFrameIndices = source.anchor_frame_indices.map(
+    (item, index) => integer(item, `M49 anchor ${index}`),
+  );
+  const primary = metrics.primary.map((item, index) => anchor(item, `M49 primary ${index}`));
+  if (
+    anchorFrameIndices.length !== 10
+    || new Set(anchorFrameIndices).size !== 10
+    || primary.length !== 10
+    || integer(metrics.anchor_count, "M49 anchor count") !== 10
+    || integer(metrics.anchor_profile_count, "M49 profile count") !== 20
+    || integer(metrics.costmap_cell_count, "M49 costmap cells") !== 2244
+    || primary.some((item, index) => item.anchorFrameIndex !== anchorFrameIndices[index])
+  ) {
+    throw new M49TgsContractError("M49 anchor set или costmap contract нарушен.");
+  }
+  return {
+    resultId: resultId(payload.result_id),
+    createdAtUtc: text(payload.created_at_utc, "M49 created"),
+    source: {
+      sourceId: exact(source.source_id, "RAVNOVES00", "M49 source id"),
+      sourceSessionId: exact(
+        source.source_session_id,
+        "20260720T065719Z_viewer_live",
+        "M49 source session",
+      ),
+      sourcePackSha256: sha256(source.source_pack_sha256, "M49 source pack"),
+      linkedVisualResultId: text(source.linked_visual_result_id, "M49 visual result"),
+      anchorFrameIndices,
+    },
+    configuration: {
+      profileId: text(configuration.profile_id, "M49 profile"),
+      configSha256: sha256(configuration.config_sha256, "M49 config"),
+      coordinateFrame: exact(configuration.coordinate_frame, "map-gravity-local", "M49 frame"),
+      primaryProfile: exact(configuration.primary_profile, "causal_rolling_1s", "M49 primary"),
+      cellSizeM: numberValue(configuration.cell_size_m, "M49 cell size"),
+      radiusM: numberValue(configuration.radius_m, "M49 radius"),
+    },
+    execution: {
+      worker: exact(execution.worker, "Worker 006", "M49 worker"),
+      device: exact(execution.device, "cpu", "M49 device"),
+      gpuUsed: exact(execution.gpu_used, false, "M49 GPU"),
+      wrapperElapsedSeconds: numberValue(execution.wrapper_elapsed_seconds, "M49 wall"),
+    },
+    metrics: {
+      anchorCount: 10,
+      anchorProfileCount: 20,
+      allEligiblePointsAccounted: exact(metrics.all_eligible_points_accounted, true, "M49 accounting"),
+      costmapCellCount: 2244,
+      processWallCurrentP50Ms: numberValue(metrics.process_wall_current_p50_ms, "M49 current p50"),
+      processWallCurrentMaxMs: numberValue(metrics.process_wall_current_max_ms, "M49 current max"),
+      processWallRollingP50Ms: numberValue(metrics.process_wall_rolling_p50_ms, "M49 rolling p50"),
+      processWallRollingMaxMs: numberValue(metrics.process_wall_rolling_max_ms, "M49 rolling max"),
+      processMaxRssKib: integer(metrics.process_max_rss_kib, "M49 RSS"),
+      primary,
+    },
+    acceptance: {
+      representationComplete: exact(acceptance.representation_complete, true, "M49 representation"),
+      allPointsAccounted: exact(acceptance.all_points_accounted, true, "M49 points"),
+      aosAbsent: exact(acceptance.aos_absent, true, "M49 AOS"),
+      gpuAbsent: exact(acceptance.gpu_absent, true, "M49 GPU absent"),
+      visualQualityAccepted: exact(acceptance.visual_quality_accepted, false, "M49 visual quality"),
+      traversabilityAccepted: exact(acceptance.traversability_accepted, false, "M49 traversability"),
+    },
+    decision: {
+      state: exact(decision.state, "visual-review-required", "M49 decision"),
+      candidateRetained: exact(decision.candidate_retained, true, "M49 retained"),
+      nextAction: text(decision.next_action, "M49 next action"),
+    },
+    limitations: payload.limitations.map((item, index) => text(item, `M49 limitation ${index}`)),
+  };
+}
+
+export async function fetchM49TgsAnchorSpatial(
+  id: string,
+  anchorFrameIndex: number,
+  profile: M49TgsProfile = "causal_rolling_1s",
+  { fetcher = fetch, signal }: { fetcher?: LaboratoryFetch; signal?: AbortSignal } = {},
+): Promise<M49TgsAnchorSpatial> {
+  if (!RESULT_ID.test(id) || !Number.isSafeInteger(anchorFrameIndex) || anchorFrameIndex < 0) {
+    throw new M49TgsContractError("M49 anchor identity недопустима.");
+  }
+  const response = await fetcher(
+    `/api/v1/laboratory/m49/tgs-fail-closed/${encodeURIComponent(id)}/anchors/${anchorFrameIndex}/spatial?profile=${encodeURIComponent(profile)}`,
+    { method: "GET", headers: { Accept: "application/json" }, signal },
+  );
+  if (!response.ok) throw new M49TgsContractError(`M49 anchor недоступен: HTTP ${response.status}.`);
+  const payload = objectValue(await response.json(), "M49 anchor");
+  exact(payload.schema_version, "missioncore.m49-tgs-anchor-spatial/v1", "M49 anchor schema");
+  exact(payload.result_id, id, "M49 anchor result");
+  exact(payload.coordinate_frame, "map-gravity-local", "M49 anchor frame");
+  exact(payload.all_points_accounted, true, "M49 anchor accounting");
+  exact(payload.aos_used, false, "M49 anchor AOS");
+  if (!Array.isArray(payload.points_xyz_m) || !Array.isArray(payload.point_states)) {
+    throw new M49TgsContractError("M49 points: ожидался массив.");
+  }
+  const costmap = objectValue(payload.costmap, "M49 costmap");
+  if (!Array.isArray(costmap.centers_xy_m) || !Array.isArray(costmap.states) || !Array.isArray(costmap.z_bounds_m)) {
+    throw new M49TgsContractError("M49 costmap arrays: нарушен контракт.");
+  }
+  const points = payload.points_xyz_m.map((item, index) => numbers(item, 3, `M49 point ${index}`) as [number, number, number]);
+  const pointStates = payload.point_states.map((item, index) => {
+    const state = integer(item, `M49 point state ${index}`);
+    if (state !== 1 && state !== 2 && state !== 3) throw new M49TgsContractError("M49 point state неизвестен.");
+    return state;
+  });
+  if (points.length !== pointStates.length) throw new M49TgsContractError("M49 point accounting нарушен.");
+  const centers = costmap.centers_xy_m.map((item, index) => numbers(item, 2, `M49 cell ${index}`) as [number, number]);
+  const states = costmap.states.map((item, index) => {
+    const state = integer(item, `M49 cell state ${index}`);
+    if (state !== 0 && state !== 1 && state !== 2 && state !== 3) throw new M49TgsContractError("M49 cell state неизвестен.");
+    return state;
+  });
+  const zBounds = costmap.z_bounds_m.map((item, index) => {
+    if (!Array.isArray(item) || item.length !== 2) throw new M49TgsContractError(`M49 z ${index}: размерность.`);
+    return item.map((value) => value === null ? null : numberValue(value, `M49 z ${index}`)) as [number | null, number | null];
+  });
+  if (
+    centers.length !== 2244
+    || centers.length !== states.length
+    || centers.length !== zBounds.length
+    || integer(payload.anchor_frame_index, "M49 anchor frame") !== anchorFrameIndex
+    || integer(payload.source_sequence, "M49 source sequence") !== anchorFrameIndex
+  ) {
+    throw new M49TgsContractError("M49 costmap accounting нарушен.");
+  }
+  return {
+    resultId: id,
+    anchorFrameIndex,
+    sourceSequence: anchorFrameIndex,
+    profile: exact(payload.profile, profile, "M49 profile"),
+    coordinateFrame: "map-gravity-local",
+    pointsXyzM: points,
+    pointStates,
+    costmap: {
+      cellSizeM: numberValue(costmap.cell_size_m, "M49 cell size"),
+      radiusM: numberValue(costmap.radius_m, "M49 radius"),
+      centersXyM: centers,
+      states,
+      zBoundsM: zBounds,
+    },
+    allPointsAccounted: true,
+    aosUsed: false,
+  };
+}

apps/control-station/src/core/laboratory/m49TgsFullShadow.ts

@@ -0,0 +1,337 @@
+import type { LaboratoryFetch } from "./advancedResults";
+
+const RESULT_ID = /^m49-tgs-full-shadow-[a-f0-9]{64}$/;
+const SEMANTIC_RESULT_ID = /^e47-semantic-slam-[a-f0-9]{64}$/;
+
+export type M49TgsFullShadowStateCode = 0 | 1 | 2 | 3;
+
+export interface M49TgsFullShadowResult {
+  resultId: string;
+  createdAtUtc: string;
+  source: {
+    sourcePackSha256: string;
+    linkedVisualResultId: string;
+    linkedSemanticResultId: string;
+  };
+  configuration: {
+    configSha256: string;
+    cellSizeM: number;
+    radiusM: number;
+    historySeconds: number;
+  };
+  execution: {
+    wrapperElapsedSeconds: number;
+  };
+  timeline: {
+    frameCount: 4489;
+    availableLidarFrameCount: 3928;
+    missingLidarFrameCount: 561;
+    durationSeconds: number;
+    effectiveFps: number;
+  };
+  pointAccounting: {
+    eligible: number;
+    ground: number;
+    nonground: number;
+    rejected: number;
+    unaccounted: 0;
+  };
+  performance: {
+    candidateTgsMs: { p50: number; p95: number; p99: number; max: number };
+    completionAgeMs: { p50: number; p95: number; p99: number; max: number };
+    capacityDropCount: number;
+  };
+  acceptance: {
+    representationComplete: true;
+    visualQualityAccepted: false;
+    integratedGraphPerformanceAccepted: false;
+    allFramesAccounted: boolean;
+    allEligiblePointsAccounted: boolean;
+  };
+  decision: {
+    state: string;
+    candidateRetained: boolean;
+    nextAction: string;
+  };
+}
+
+export interface M49TgsFullShadowSpatial {
+  resultId: string;
+  sourceSequence: number;
+  sourceFrameIndex: number;
+  sessionSeconds: number;
+  sampleAvailable: boolean;
+  costmap: {
+    cellSizeM: number;
+    radiusM: number;
+    centersXyM: readonly (readonly [number, number])[];
+    states: readonly M49TgsFullShadowStateCode[];
+    zBoundsM: readonly (readonly [number | null, number | null])[];
+  };
+  metrics: {
+    eligiblePointCount: number;
+    groundPointCount: number;
+    nongroundPointCount: number;
+    rejectedPointCount: number;
+    occupiedCellCount: number;
+  };
+}
+
+export interface M49TgsFullShadowSpatialChunk {
+  resultId: string;
+  start: number;
+  count: number;
+  frames: readonly M49TgsFullShadowSpatial[];
+}
+
+export class M49TgsFullShadowContractError extends Error {}
+
+function objectValue(value: unknown, label: string): Record<string, unknown> {
+  if (!value || typeof value !== "object" || Array.isArray(value)) {
+    throw new M49TgsFullShadowContractError(`${label}: ожидался объект.`);
+  }
+  return value as Record<string, unknown>;
+}
+
+function text(value: unknown, label: string): string {
+  if (typeof value !== "string" || !value.trim()) {
+    throw new M49TgsFullShadowContractError(`${label}: ожидалась строка.`);
+  }
+  return value;
+}
+
+function numberValue(value: unknown, label: string): number {
+  if (typeof value !== "number" || !Number.isFinite(value)) {
+    throw new M49TgsFullShadowContractError(`${label}: ожидалось число.`);
+  }
+  return value;
+}
+
+function integer(value: unknown, label: string): number {
+  const parsed = numberValue(value, label);
+  if (!Number.isSafeInteger(parsed) || parsed < 0) {
+    throw new M49TgsFullShadowContractError(`${label}: ожидалось целое число.`);
+  }
+  return parsed;
+}
+
+function booleanValue(value: unknown, label: string): boolean {
+  if (typeof value !== "boolean") {
+    throw new M49TgsFullShadowContractError(`${label}: ожидался boolean.`);
+  }
+  return value;
+}
+
+function exact<T extends string | boolean | number>(value: unknown, expected: T, label: string): T {
+  if (value !== expected) throw new M49TgsFullShadowContractError(`${label}: нарушен контракт.`);
+  return expected;
+}
+
+function timing(value: unknown, label: string) {
+  const row = objectValue(value, label);
+  return {
+    p50: numberValue(row.p50, `${label}.p50`),
+    p95: numberValue(row.p95, `${label}.p95`),
+    p99: numberValue(row.p99, `${label}.p99`),
+    max: numberValue(row.max, `${label}.max`),
+  };
+}
+
+export async function fetchM49TgsFullShadowResult(
+  id: string,
+  { fetcher = fetch, signal }: { fetcher?: LaboratoryFetch; signal?: AbortSignal } = {},
+): Promise<M49TgsFullShadowResult> {
+  if (!RESULT_ID.test(id)) throw new M49TgsFullShadowContractError("M49 full-shadow identity недопустима.");
+  const response = await fetcher(
+    `/api/v1/laboratory/m49/tgs-full-shadow/${encodeURIComponent(id)}`,
+    { method: "GET", headers: { Accept: "application/json" }, signal },
+  );
+  if (!response.ok) throw new M49TgsFullShadowContractError(`M49 full shadow недоступен: HTTP ${response.status}.`);
+  const payload = objectValue(await response.json(), "M49 full shadow");
+  exact(payload.schema_version, "missioncore.m49-tgs-full-shadow-view/v1", "M49 full schema");
+  exact(payload.result_id, id, "M49 full result");
+  exact(payload.ground_truth, false, "M49 full ground truth");
+  exact(payload.access, "read-only", "M49 full access");
+  const source = objectValue(payload.source, "M49 full source");
+  const configuration = objectValue(payload.configuration, "M49 full configuration");
+  const execution = objectValue(payload.execution, "M49 full execution");
+  const timeline = objectValue(payload.timeline, "M49 full timeline");
+  const accounting = objectValue(payload.point_accounting, "M49 full accounting");
+  const performance = objectValue(payload.performance, "M49 full performance");
+  const acceptance = objectValue(payload.acceptance, "M49 full acceptance");
+  const decision = objectValue(payload.decision, "M49 full decision");
+  const linkedSemanticResultId = text(
+    source.linked_semantic_result_id,
+    "M49 full semantic result",
+  );
+  if (!SEMANTIC_RESULT_ID.test(linkedSemanticResultId)) {
+    throw new M49TgsFullShadowContractError("M49 full semantic result: нарушена идентичность.");
+  }
+  return {
+    resultId: text(payload.result_id, "M49 full result"),
+    createdAtUtc: text(payload.created_at_utc, "M49 full created"),
+    source: {
+      sourcePackSha256: text(source.source_pack_sha256, "M49 full source pack"),
+      linkedVisualResultId: text(source.linked_visual_result_id, "M49 full visual result"),
+      linkedSemanticResultId,
+    },
+    configuration: {
+      configSha256: text(configuration.config_sha256, "M49 full config"),
+      cellSizeM: numberValue(configuration.cell_size_m, "M49 full cell size"),
+      radiusM: numberValue(configuration.radius_m, "M49 full radius"),
+      historySeconds: numberValue(configuration.history_seconds, "M49 full history"),
+    },
+    execution: { wrapperElapsedSeconds: numberValue(execution.wrapper_elapsed_seconds, "M49 full wall") },
+    timeline: {
+      frameCount: exact(integer(timeline.frame_count, "M49 full frames"), 4489, "M49 full frames"),
+      availableLidarFrameCount: exact(integer(timeline.available_lidar_frame_count, "M49 full available"), 3928, "M49 full available"),
+      missingLidarFrameCount: exact(integer(timeline.missing_lidar_frame_count, "M49 full missing"), 561, "M49 full missing"),
+      durationSeconds: numberValue(timeline.duration_seconds, "M49 full duration"),
+      effectiveFps: numberValue(timeline.effective_fps, "M49 full fps"),
+    },
+    pointAccounting: {
+      eligible: integer(accounting.eligible, "M49 full eligible"),
+      ground: integer(accounting.ground, "M49 full ground"),
+      nonground: integer(accounting.nonground, "M49 full nonground"),
+      rejected: integer(accounting.rejected, "M49 full rejected"),
+      unaccounted: exact(integer(accounting.unaccounted, "M49 full unaccounted"), 0, "M49 full unaccounted"),
+    },
+    performance: {
+      candidateTgsMs: timing(performance.candidate_tgs_ms, "M49 full TGS"),
+      completionAgeMs: timing(performance.completion_age_ms, "M49 full completion"),
+      capacityDropCount: integer(performance.capacity_drop_count, "M49 full drops"),
+    },
+    acceptance: {
+      representationComplete: exact(acceptance.representation_complete, true, "M49 full representation"),
+      visualQualityAccepted: exact(acceptance.visual_quality_accepted, false, "M49 full visual"),
+      integratedGraphPerformanceAccepted: exact(acceptance.integrated_graph_performance_accepted, false, "M49 full integrated"),
+      allFramesAccounted: booleanValue(acceptance.all_frames_accounted, "M49 full frame accounting"),
+      allEligiblePointsAccounted: booleanValue(acceptance.all_eligible_points_accounted, "M49 full point accounting"),
+    },
+    decision: {
+      state: text(decision.state, "M49 full decision"),
+      candidateRetained: booleanValue(decision.candidate_retained, "M49 full retained"),
+      nextAction: text(decision.next_action, "M49 full next action"),
+    },
+  };
+}
+
+function pair(value: unknown, label: string): readonly [number, number] {
+  if (!Array.isArray(value) || value.length !== 2) {
+    throw new M49TgsFullShadowContractError(`${label}: неверная размерность.`);
+  }
+  return [numberValue(value[0], label), numberValue(value[1], label)];
+}
+
+function parseSpatial(
+  id: string,
+  sourceSequence: number,
+  payload: Record<string, unknown>,
+  sharedCostmap?: Record<string, unknown>,
+): M49TgsFullShadowSpatial {
+  exact(payload.source_sequence, sourceSequence, "M49 full spatial sequence");
+  const costmap = sharedCostmap ?? objectValue(payload.costmap, "M49 full costmap");
+  const metrics = objectValue(payload.metrics, "M49 full frame metrics");
+  const centers = costmap.centers_xy_m;
+  const statesRaw = sharedCostmap ? payload.states : costmap.states;
+  const zBoundsRaw = sharedCostmap ? payload.z_bounds_m : costmap.z_bounds_m;
+  if (!Array.isArray(centers) || !Array.isArray(statesRaw) || !Array.isArray(zBoundsRaw)
+      || centers.length !== 2244 || statesRaw.length !== 2244 || zBoundsRaw.length !== 2244) {
+    throw new M49TgsFullShadowContractError("M49 full costmap: неверная размерность.");
+  }
+  const states = statesRaw.map((value, index) => {
+    const parsed = integer(value, `M49 full state ${index}`);
+    if (parsed > 3) throw new M49TgsFullShadowContractError("M49 full state недопустим.");
+    return parsed as M49TgsFullShadowStateCode;
+  });
+  const zBounds = zBoundsRaw.map((value, index) => {
+    if (!Array.isArray(value) || value.length !== 2) {
+      throw new M49TgsFullShadowContractError(`M49 full z ${index}: неверная размерность.`);
+    }
+    return value.map((item) => item === null
+      ? null
+      : numberValue(item, `M49 full z ${index}`)) as [number | null, number | null];
+  });
+  const sampleAvailable = booleanValue(payload.sample_available, "M49 full sample available");
+  if (!sampleAvailable && states.some((state) => state !== 0)) {
+    throw new M49TgsFullShadowContractError("M49 missing LiDAR frame не остался UNOBSERVED.");
+  }
+  return {
+    resultId: id,
+    sourceSequence,
+    sourceFrameIndex: integer(payload.source_frame_index, "M49 full source frame"),
+    sessionSeconds: numberValue(payload.session_seconds, "M49 full session seconds"),
+    sampleAvailable,
+    costmap: {
+      cellSizeM: numberValue(costmap.cell_size_m, "M49 full cell size"),
+      radiusM: numberValue(costmap.radius_m, "M49 full radius"),
+      centersXyM: centers.map((value, index) => pair(value, `M49 full center ${index}`)),
+      states,
+      zBoundsM: zBounds,
+    },
+    metrics: {
+      eligiblePointCount: integer(metrics.eligible_point_count, "M49 full eligible"),
+      groundPointCount: integer(metrics.ground_point_count, "M49 full ground"),
+      nongroundPointCount: integer(metrics.nonground_point_count, "M49 full nonground"),
+      rejectedPointCount: integer(metrics.rejected_point_count, "M49 full rejected"),
+      occupiedCellCount: integer(metrics.occupied_cell_count, "M49 full occupied cells"),
+    },
+  };
+}
+
+export async function fetchM49TgsFullShadowSpatial(
+  id: string,
+  sourceSequence: number,
+  { fetcher = fetch, signal }: { fetcher?: LaboratoryFetch; signal?: AbortSignal } = {},
+): Promise<M49TgsFullShadowSpatial> {
+  if (!RESULT_ID.test(id) || !Number.isSafeInteger(sourceSequence) || sourceSequence < 0 || sourceSequence >= 4489) {
+    throw new M49TgsFullShadowContractError("M49 full-shadow frame identity недопустима.");
+  }
+  const response = await fetcher(
+    `/api/v1/laboratory/m49/tgs-full-shadow/${encodeURIComponent(id)}/frames/${sourceSequence}/spatial`,
+    { method: "GET", headers: { Accept: "application/json" }, signal },
+  );
+  if (!response.ok) throw new M49TgsFullShadowContractError(`M49 full-shadow frame недоступен: HTTP ${response.status}.`);
+  const payload = objectValue(await response.json(), "M49 full spatial");
+  exact(payload.schema_version, "missioncore.m49-tgs-full-shadow-spatial/v1", "M49 full spatial schema");
+  exact(payload.result_id, id, "M49 full spatial result");
+  return parseSpatial(id, sourceSequence, payload);
+}
+
+export async function fetchM49TgsFullShadowSpatialChunk(
+  id: string,
+  start: number,
+  count = 24,
+  { fetcher = fetch, signal }: { fetcher?: LaboratoryFetch; signal?: AbortSignal } = {},
+): Promise<M49TgsFullShadowSpatialChunk> {
+  if (!RESULT_ID.test(id) || !Number.isSafeInteger(start) || start < 0 || start >= 4489
+      || !Number.isSafeInteger(count) || count < 1 || count > 24) {
+    throw new M49TgsFullShadowContractError("M49 full-shadow chunk identity недопустима.");
+  }
+  const response = await fetcher(
+    `/api/v1/laboratory/m49/tgs-full-shadow/${encodeURIComponent(id)}/spatial/chunk?start=${start}&count=${count}`,
+    { method: "GET", headers: { Accept: "application/json" }, signal },
+  );
+  if (!response.ok) throw new M49TgsFullShadowContractError(`M49 full-shadow chunk недоступен: HTTP ${response.status}.`);
+  const payload = objectValue(await response.json(), "M49 full spatial chunk");
+  exact(payload.schema_version, "missioncore.m49-tgs-full-shadow-spatial-chunk/v1", "M49 full chunk schema");
+  exact(payload.result_id, id, "M49 full chunk result");
+  exact(payload.start, start, "M49 full chunk start");
+  const returnedCount = integer(payload.count, "M49 full chunk count");
+  if (!Array.isArray(payload.frames) || payload.frames.length !== returnedCount
+      || returnedCount !== Math.min(count, 4489 - start)) {
+    throw new M49TgsFullShadowContractError("M49 full chunk: неверная размерность.");
+  }
+  const costmap = objectValue(payload.costmap, "M49 full chunk costmap");
+  return {
+    resultId: id,
+    start,
+    count: returnedCount,
+    frames: payload.frames.map((value, index) => parseSpatial(
+      id,
+      start + index,
+      objectValue(value, `M49 full chunk frame ${index}`),
+      costmap,
+    )),
+  };
+}

apps/control-station/src/styles/m4-replay-threat.css

@@ -91,6 +91,11 @@
   justify-content: space-between;
 }
 
+.m4-replay-threat-evidence-viewer[data-mode-controls="content"]
+  .m4-replay-threat-visual__pane-toolbar[data-pane-toolbar="spatial"] {
+  right: 3.8rem;
+}
+
 .m4-replay-threat-visual__pane-toolbar > *,
 .m4-replay-threat-visual__spatial-toolbar-end > * {
   pointer-events: auto;
@@ -350,3 +355,20 @@
   background: rgb(var(--nodedc-accent-rgb));
   opacity: 0.72;
 }
+
+.laboratory-metric-evidence-scene__legend span[data-decision="ground-support"]::before {
+  background: rgb(var(--nodedc-success-rgb));
+}
+
+.laboratory-metric-evidence-scene__legend span[data-decision="nonground-occupied"]::before {
+  background: rgb(var(--nodedc-danger-rgb));
+}
+
+.laboratory-metric-evidence-scene__legend span[data-decision="unknown-rejected"]::before {
+  background: rgb(var(--nodedc-warning-rgb));
+}
+
+.laboratory-metric-evidence-scene__legend span[data-decision="unobserved"]::before {
+  background: var(--nodedc-text-muted);
+  opacity: 0.42;
+}

apps/control-station/src/workspaces/laboratory/AdvancedLaboratoryResult.tsx

@@ -48,6 +48,8 @@ import { M48StaticOccupancyQualificationResultView } from "./M48StaticOccupancyQ
 import { M48R3StaticOccupancyShadowResultView } from "./M48R3StaticOccupancyShadowResult";
 import { M48SFixedClassDetectorResultView } from "./M48SFixedClassDetectorResult";
 import { M48TRiskQualityResultView } from "./M48TRiskQualityResult";
+import { M49TgsFailClosedResultView } from "./M49TgsFailClosedResult";
+import { M49TgsFullShadowResultView } from "./M49TgsFullShadowResult";
 
 export { isAdvancedLaboratoryWorkId };
 export type { AdvancedLaboratoryWorkId };
@@ -108,6 +110,12 @@ export function AdvancedLaboratoryResult({
   if (workId === "m48t-risk-quality-temporal" && results.m48t) {
     return <M48TRiskQualityResultView rigLabel={rigLabel} result={results.m48t} />;
   }
+  if (workId === "m49-tgs-fail-closed-evidence" && results.m49Tgs) {
+    return <M49TgsFailClosedResultView rigLabel={rigLabel} result={results.m49Tgs} />;
+  }
+  if (workId === "m49-tgs-full-shadow" && results.m49TgsFull) {
+    return <M49TgsFullShadowResultView rigLabel={rigLabel} result={results.m49TgsFull} />;
+  }
   if (workId === "m47-reference-graph-shadow" && results.m47Graph) {
     return <M47ReferenceGraphResultView rigLabel={rigLabel} result={results.m47Graph} />;
   }

apps/control-station/src/workspaces/laboratory/M49TgsFailClosedEvidence.tsx

@@ -0,0 +1,152 @@
+import { useCallback, useEffect, useMemo, useRef, useState } from "react";
+
+import type {
+  RecordedEvidenceSemanticClass,
+  RecordedEvidenceSemanticPaletteEntry,
+} from "../../components/laboratory/RecordedEvidenceSemanticMaskOverlay";
+import {
+  fetchM49TgsAnchorSpatial,
+  type M49TgsAnchorSpatial,
+  type M49TgsFailClosedResult,
+  type M49TgsStateCode,
+} from "../../core/laboratory/m49TgsFailClosed";
+import {
+  M4ReplayThreatVisual,
+  type M4ReplayClassifiedSpatialFrame,
+  type M4ReplayThreatReviewAnchor,
+} from "./M4ReplayThreatVisual";
+
+const CLASSES: readonly RecordedEvidenceSemanticClass[] = [
+  { id: 1, label: "Ground support" },
+  { id: 2, label: "Non-ground occupied" },
+  { id: 3, label: "Unknown / rejected" },
+];
+
+const PALETTE: readonly RecordedEvidenceSemanticPaletteEntry[] = [
+  { classId: 1, color: { kind: "token", token: "--nodedc-success-rgb" } },
+  { classId: 2, color: { kind: "token", token: "--nodedc-danger-rgb" } },
+  { classId: 3, color: { kind: "token", token: "--nodedc-warning-rgb" } },
+];
+
+function cellState(code: M49TgsStateCode): M4ReplayClassifiedSpatialFrame["cellsMapGravityLocal"][number]["state"] {
+  if (code === 1) return "ground-support";
+  if (code === 2) return "nonground-occupied";
+  if (code === 3) return "unknown-rejected";
+  return "unobserved";
+}
+
+function message(error: unknown): string {
+  return error instanceof Error && error.message.trim()
+    ? error.message
+    : "M49 TGS spatial evidence недоступно.";
+}
+
+export function M49TgsFailClosedEvidence({
+  result,
+}: {
+  result: M49TgsFailClosedResult;
+}) {
+  const [activeSequence, setActiveSequence] = useState<number | null>(null);
+  const [spatial, setSpatial] = useState<M49TgsAnchorSpatial | null>(null);
+  const [loading, setLoading] = useState(false);
+  const [error, setError] = useState<string | null>(null);
+  const spatialCacheRef = useRef(new Map<string, M49TgsAnchorSpatial>());
+  const anchorSequences = useMemo(
+    () => new Set(result.metrics.primary.map((item) => item.anchorFrameIndex)),
+    [result.metrics.primary],
+  );
+  const expectedAtSequence = activeSequence !== null && anchorSequences.has(activeSequence);
+
+  useEffect(() => {
+    if (activeSequence === null || !anchorSequences.has(activeSequence)) {
+      setSpatial(null);
+      setLoading(false);
+      setError(null);
+      return;
+    }
+    const cacheKey = `${result.resultId}:${activeSequence}`;
+    const cached = spatialCacheRef.current.get(cacheKey);
+    if (cached) {
+      setSpatial(cached);
+      setLoading(false);
+      setError(null);
+      return;
+    }
+    const controller = new AbortController();
+    setSpatial(null);
+    setLoading(true);
+    setError(null);
+    void fetchM49TgsAnchorSpatial(
+      result.resultId,
+      activeSequence,
+      "causal_rolling_1s",
+      { signal: controller.signal },
+    )
+      .then((next) => {
+        if (!controller.signal.aborted) {
+          spatialCacheRef.current.set(cacheKey, next);
+          setSpatial(next);
+        }
+      })
+      .catch((caught: unknown) => {
+        if (!controller.signal.aborted) setError(message(caught));
+      })
+      .finally(() => {
+        if (!controller.signal.aborted) setLoading(false);
+      });
+    return () => controller.abort();
+  }, [activeSequence, anchorSequences, result.resultId]);
+
+  const reviewAnchors = useMemo<readonly M4ReplayThreatReviewAnchor[]>(
+    () => result.metrics.primary.map((item) => ({
+      id: `m49-tgs-${item.anchorFrameIndex}`,
+      sourceSequence: item.anchorFrameIndex,
+      extentXyxyNormalized: [0, 0, 0, 0],
+      matchedAtThreshold: false,
+      statusLabel: "визуальная проверка",
+    })),
+    [result.metrics.primary],
+  );
+
+  const classifiedFrame = useMemo<M4ReplayClassifiedSpatialFrame | null>(() => {
+    if (!spatial) return null;
+    return {
+      sourceSequence: spatial.sourceSequence,
+      pointsMapGravityLocalXyzM: spatial.pointsXyzM,
+      pointClassIds: spatial.pointStates,
+      cellsMapGravityLocal: spatial.costmap.centersXyM.map((center, index) => ({
+        centerXyM: center,
+        zBoundsM: spatial.costmap.zBoundsM[index]!,
+        state: cellState(spatial.costmap.states[index]!),
+      })),
+      cellSizeM: spatial.costmap.cellSizeM,
+      classes: CLASSES,
+      palette: PALETTE,
+    };
+  }, [spatial]);
+
+  const handleSequenceChange = useCallback((sequence: number | null) => {
+    setActiveSequence(sequence);
+  }, []);
+
+  return (
+    <M4ReplayThreatVisual
+      resultId={result.source.linkedVisualResultId}
+      reviewAnchors={reviewAnchors}
+      showReviewAnchorBoxes={false}
+      reviewLabel="10 gravity-aligned TGS anchors"
+      evidenceLabel="M49 · TGS fail-closed"
+      initialSpatialMode="3d"
+      onActiveSequenceChange={handleSequenceChange}
+      classifiedSpatialLayer={{
+        label: "TGS fail-closed · causal rolling 1 s",
+        pointLayerLabel: "TGS POINTS",
+        cellLayerLabel: "COSTMAP",
+        expectedAtSequence,
+        frame: classifiedFrame,
+        loading,
+        error,
+      }}
+    />
+  );
+}

apps/control-station/src/workspaces/laboratory/M49TgsFailClosedResult.tsx

@@ -0,0 +1,92 @@
+import {
+  LaboratoryEvidence,
+  LaboratoryResultSummary,
+  LaboratorySummary,
+  LaboratoryWorkTemplate,
+} from "../../components/laboratory/LaboratoryPresentation";
+import type { M49TgsFailClosedResult } from "../../core/laboratory/m49TgsFailClosed";
+import { M49TgsFailClosedEvidence } from "./M49TgsFailClosedEvidence";
+
+function number(value: number, digits = 1): string {
+  return value.toLocaleString("ru-RU", { maximumFractionDigits: digits });
+}
+
+export function M49TgsFailClosedResultView({
+  rigLabel,
+  result,
+}: {
+  rigLabel: string;
+  result: M49TgsFailClosedResult;
+}) {
+  const worst = [...result.metrics.primary].sort(
+    (left, right) => (
+      right.nongroundPointCount / Math.max(right.pointCount, 1)
+      - left.nongroundPointCount / Math.max(left.pointCount, 1)
+    ),
+  )[0]!;
+  const status = "Representation complete; визуальное качество ещё не принято";
+  return (
+    <LaboratoryWorkTemplate
+      summary={(
+        <LaboratorySummary
+          title="M4.9T4 · TRAVEL TGS fail-closed evidence"
+          description="TRAVEL GroundSeg запущен без AOS на десяти immutable RAVNOVES00 anchors. Вход сохранён в gravity-aligned map frame; каждый eligible point получил состояние, а каждая costmap-ячейка остаётся ground, occupied, rejected или unobserved."
+          status={status}
+          statusTone="warning"
+          facts={[
+            { label: "Конфигурация", value: `${rigLabel} RIGHT · Camera + gravity-aligned LiDAR · 10 anchors` },
+            { label: "Метод", value: "TRAVEL TGS only · AOS OFF · causal rolling 1 s" },
+            { label: "Evidence", value: `${result.metrics.anchorCount} anchors · ${result.metrics.costmapCellCount.toLocaleString("ru-RU")} cells/anchor · all points accounted` },
+            { label: "Нагрузка", value: `Worker 006 CPU-only · GPU 0 · wrapper ${number(result.execution.wrapperElapsedSeconds, 2)} с` },
+            { label: "Authority", value: "REPLAY-SIMULATED · visual/traversability/navigation/actuation OFF" },
+          ]}
+          brief={{
+            question: "Отделяет ли готовый TRAVEL TGS опорную поверхность от неизвестной занятой геометрии достаточно чисто, чтобы заменить самодельный static-obstacle threshold pipeline?",
+            approach: "На десяти сложных кадрах проверяется полный gravity-aligned point set и fail-closed costmap. Зелёное — опора, красное — non-ground occupied, жёлтое — rejected/unknown, тёмное — unobserved; камера остаётся синхронным первичным контекстом.",
+            principalResult: `Контракт представления закрыт: ${result.metrics.anchorProfileCount}/20 профилей, ни одной потерянной eligible point, AOS и GPU отсутствуют. Process wall rolling p50/max: ${number(result.metrics.processWallRollingP50Ms, 0)}/${number(result.metrics.processWallRollingMaxMs, 0)} мс.`,
+            limitation: `Качество не принято: особенно проверить кадр ${worst.anchorFrameIndex + 1}, где ${number(worst.nongroundPointCount / Math.max(worst.pointCount, 1) * 100)}% rolling points помечены non-ground. Это может быть реальная боковая геометрия либо ложная блокировка поверхности.`,
+          }}
+          method={{
+            completeness: "complete",
+            executionClass: "deterministic",
+            pipelineId: "travel-tgs-gravity-aligned-fail-closed/v1",
+            components: [
+              { kind: "source", name: "RAVNOVES00", version: "10 immutable anchors", role: "camera + registered map increments", identitySha256: result.source.sourcePackSha256 },
+              { kind: "algorithm", name: "TRAVEL GroundSeg", version: "95dc2fbd66a343efd9060c45a5711b6307a950a4", role: "ground/nonground separation; AOS excluded", identitySha256: result.configuration.configSha256 },
+              { kind: "algorithm", name: "fail-closed complement adapter", version: "v1", role: "explicit rejected points and unobserved cells", identitySha256: result.resultId.split("-").at(-1) ?? null },
+            ],
+          }}
+        />
+      )}
+      evidence={(
+        <LaboratoryEvidence
+          eyebrow="M4.9T4 VISUAL EVIDENCE · CAMERA + GRAVITY-ALIGNED TGS"
+          title="10 anchors: полный TGS point set и четырёхсостояний costmap на том же recorded timeline"
+          kind="recorded-replay"
+          resizable
+        >
+          <M49TgsFailClosedEvidence result={result} />
+        </LaboratoryEvidence>
+      )}
+      result={(
+        <LaboratoryResultSummary
+          title="Что уже доказано и что проверяем глазами"
+          status={status}
+          statusTone="warning"
+          metrics={[
+            { label: "Point accounting", value: "100%", hint: "ground + non-ground + rejected = exact eligible input" },
+            { label: "Anchors", value: `${result.metrics.anchorCount}/10`, hint: "current + causal rolling 1 s" },
+            { label: "Costmap", value: `${result.metrics.costmapCellCount.toLocaleString("ru-RU")} cells`, hint: `${number(result.configuration.cellSizeM, 2)} м · radius ${number(result.configuration.radiusM, 0)} м` },
+            { label: "Process wall rolling", value: `${number(result.metrics.processWallRollingP50Ms, 0)} / ${number(result.metrics.processWallRollingMaxMs, 0)} мс`, hint: "p50 / max · CPU process envelope, не realtime integration" },
+            { label: "GPU / AOS", value: "0 / OFF", hint: "Worker 006; Frigate budget не затронут" },
+          ]}
+          conclusion={{
+            proved: "Готовый TGS можно встроить fail-closed: исходные точки не теряются, unknown не становится free, AOS не нужен, а вычисление укладывается в лёгкий CPU-контур на этих anchors.",
+            notProved: "Не доказано, что красный non-ground слой не режет дорогу, траву или допустимые просветы. Нет независимой terrain truth, полного replay, realtime graph integration и модели корпуса.",
+            decision: "Открыть десять anchors по очереди. Если красное остаётся на реальных препятствиях и не перекрывает видимую опорную поверхность, TGS идёт в полный shadow; иначе кандидат отклоняется без ручной подгонки порогов под эти кадры.",
+          }}
+        />
+      )}
+    />
+  );
+}

apps/control-station/src/workspaces/laboratory/M49TgsFullShadowEvidence.tsx

@@ -0,0 +1,202 @@
+import { useCallback, useEffect, useMemo, useRef, useState } from "react";
+
+import type {
+  RecordedEvidenceSemanticClass,
+  RecordedEvidenceSemanticPaletteEntry,
+} from "../../components/laboratory/RecordedEvidenceSemanticMaskOverlay";
+import {
+  fetchE47SemanticSlamResult,
+  type E47SemanticSlamResult,
+} from "../../core/laboratory/e47SemanticSlam";
+import {
+  fetchM49TgsFullShadowSpatialChunk,
+  type M49TgsFullShadowResult,
+  type M49TgsFullShadowSpatial,
+  type M49TgsFullShadowSpatialChunk,
+  type M49TgsFullShadowStateCode,
+} from "../../core/laboratory/m49TgsFullShadow";
+import {
+  M4ReplayThreatVisual,
+  type M4ReplayClassifiedSpatialFrame,
+} from "./M4ReplayThreatVisual";
+
+const CLASSES: readonly RecordedEvidenceSemanticClass[] = [
+  { id: 1, label: "Ground support" },
+  { id: 2, label: "Non-ground occupied" },
+  { id: 3, label: "Unknown / rejected" },
+];
+
+const PALETTE: readonly RecordedEvidenceSemanticPaletteEntry[] = [
+  { classId: 1, color: { kind: "token", token: "--nodedc-success-rgb" } },
+  { classId: 2, color: { kind: "token", token: "--nodedc-danger-rgb" } },
+  { classId: 3, color: { kind: "token", token: "--nodedc-warning-rgb" } },
+];
+
+const CHUNK_FRAMES = 24;
+const RETAINED_CHUNKS = 4;
+
+function cellState(code: M49TgsFullShadowStateCode): M4ReplayClassifiedSpatialFrame["cellsMapGravityLocal"][number]["state"] {
+  if (code === 1) return "ground-support";
+  if (code === 2) return "nonground-occupied";
+  if (code === 3) return "unknown-rejected";
+  return "unobserved";
+}
+
+function message(error: unknown): string {
+  return error instanceof Error && error.message.trim()
+    ? error.message
+    : "Полный TGS spatial frame недоступен.";
+}
+
+export function M49TgsFullShadowEvidence({ result }: { result: M49TgsFullShadowResult }) {
+  const [activeSequence, setActiveSequence] = useState<number | null>(null);
+  const [semantic, setSemantic] = useState<E47SemanticSlamResult | null>(null);
+  const [semanticError, setSemanticError] = useState<string | null>(null);
+  const [chunks, setChunks] = useState<ReadonlyMap<number, M49TgsFullShadowSpatialChunk>>(
+    () => new Map(),
+  );
+  const [error, setError] = useState<string | null>(null);
+  const inFlightRef = useRef(new Map<number, AbortController>());
+  const activeChunkStart = activeSequence === null
+    ? null
+    : Math.floor(activeSequence / CHUNK_FRAMES) * CHUNK_FRAMES;
+  const activeChunkStartRef = useRef(activeChunkStart);
+  activeChunkStartRef.current = activeChunkStart;
+
+  useEffect(() => {
+    const controller = new AbortController();
+    setSemantic(null);
+    setSemanticError(null);
+    void fetchE47SemanticSlamResult({
+      resultId: result.source.linkedSemanticResultId,
+      signal: controller.signal,
+    })
+      .then((next) => {
+        if (controller.signal.aborted) return;
+        if (!next || next.baseM4ResultId !== result.source.linkedVisualResultId) {
+          throw new Error("Semantic archive не совпал с исходным M4 timeline.");
+        }
+        setSemantic(next);
+      })
+      .catch((caught: unknown) => {
+        if (!controller.signal.aborted) setSemanticError(message(caught));
+      });
+    return () => controller.abort();
+  }, [result.source.linkedSemanticResultId, result.source.linkedVisualResultId]);
+
+  useEffect(() => {
+    for (const controller of inFlightRef.current.values()) controller.abort();
+    inFlightRef.current.clear();
+    setChunks(new Map());
+    setError(null);
+    return () => {
+      for (const controller of inFlightRef.current.values()) controller.abort();
+      inFlightRef.current.clear();
+    };
+  }, [result.resultId]);
+
+  useEffect(() => {
+    if (activeChunkStart === null) return;
+    const desiredStarts = [activeChunkStart, activeChunkStart + CHUNK_FRAMES]
+      .filter((start) => start < result.timeline.frameCount);
+    const desired = new Set(desiredStarts);
+    for (const [start, controller] of inFlightRef.current) {
+      if (desired.has(start)) continue;
+      controller.abort();
+      inFlightRef.current.delete(start);
+    }
+    for (const start of desiredStarts) {
+      if (chunks.has(start) || inFlightRef.current.has(start)) continue;
+      const controller = new AbortController();
+      inFlightRef.current.set(start, controller);
+      void fetchM49TgsFullShadowSpatialChunk(result.resultId, start, CHUNK_FRAMES, {
+        signal: controller.signal,
+      })
+        .then((chunk) => {
+          if (controller.signal.aborted) return;
+          setChunks((current) => {
+            const next = new Map(current);
+            next.set(start, chunk);
+            const center = activeChunkStartRef.current ?? start;
+            const retained = [...next.keys()]
+              .sort((left, right) => Math.abs(left - center) - Math.abs(right - center))
+              .slice(0, RETAINED_CHUNKS);
+            return new Map(retained.map((key) => [key, next.get(key)!]));
+          });
+          if (start === activeChunkStartRef.current) setError(null);
+        })
+        .catch((caught: unknown) => {
+          if (!controller.signal.aborted && start === activeChunkStartRef.current) {
+            setError(message(caught));
+          }
+        })
+        .finally(() => {
+          if (inFlightRef.current.get(start) === controller) inFlightRef.current.delete(start);
+        });
+      break;
+    }
+  }, [activeChunkStart, chunks, result.resultId, result.timeline.frameCount]);
+
+  const spatial = useMemo<M49TgsFullShadowSpatial | null>(() => {
+    if (activeSequence === null || activeChunkStart === null) return null;
+    return chunks.get(activeChunkStart)?.frames.find(
+      (frame) => frame.sourceSequence === activeSequence,
+    ) ?? null;
+  }, [activeChunkStart, activeSequence, chunks]);
+  const loading = activeSequence !== null && !spatial && !error;
+
+  const classifiedFrame = useMemo<M4ReplayClassifiedSpatialFrame | null>(() => {
+    if (!spatial) return null;
+    return {
+      sourceSequence: spatial.sourceSequence,
+      sampleAvailable: spatial.sampleAvailable,
+      sourcePointCount: spatial.metrics.eligiblePointCount,
+      pointsMapGravityLocalXyzM: [],
+      pointClassIds: [],
+      cellsMapGravityLocal: spatial.costmap.centersXyM.map((center, index) => ({
+        centerXyM: center,
+        zBoundsM: spatial.costmap.zBoundsM[index]!,
+        state: cellState(spatial.costmap.states[index]!),
+      })),
+      cellSizeM: spatial.costmap.cellSizeM,
+      classes: CLASSES,
+      palette: PALETTE,
+    };
+  }, [spatial]);
+
+  const handleSequenceChange = useCallback((sequence: number | null) => {
+    setActiveSequence(sequence);
+  }, []);
+
+  return (
+    <>
+      <M4ReplayThreatVisual
+        resultId={result.source.linkedVisualResultId}
+        semantic={semantic ? {
+          resultId: semantic.resultId,
+          taxonomy: semantic.taxonomy,
+        } : undefined}
+        showReviewAnchorBoxes={false}
+        reviewLabel="4 489 source-paced TGS frames"
+        evidenceLabel="M49 · full TGS shadow"
+        initialSpatialMode="3d"
+        onActiveSequenceChange={handleSequenceChange}
+        classifiedSpatialLayer={{
+          label: "TGS full shadow · causal rolling 1 s",
+          pointLayerLabel: "SOURCE POINTS",
+          cellLayerLabel: "TGS COSTMAP",
+          expectedAtSequence: true,
+          frame: classifiedFrame,
+          loading,
+          error,
+          replacePointCloud: false,
+        }}
+      />
+      {semanticError ? (
+        <div className="m4-replay-threat-visual__pane-status" role="alert">
+          Semantic overlay недоступен: {semanticError}
+        </div>
+      ) : null}
+    </>
+  );
+}

apps/control-station/src/workspaces/laboratory/M49TgsFullShadowResult.tsx

@@ -0,0 +1,90 @@
+import {
+  LaboratoryEvidence,
+  LaboratoryResultSummary,
+  LaboratorySummary,
+  LaboratoryWorkTemplate,
+} from "../../components/laboratory/LaboratoryPresentation";
+import type { M49TgsFullShadowResult } from "../../core/laboratory/m49TgsFullShadow";
+import { M49TgsFullShadowEvidence } from "./M49TgsFullShadowEvidence";
+
+function number(value: number, digits = 1): string {
+  return value.toLocaleString("ru-RU", { maximumFractionDigits: digits });
+}
+
+export function M49TgsFullShadowResultView({
+  rigLabel,
+  result,
+}: {
+  rigLabel: string;
+  result: M49TgsFullShadowResult;
+}) {
+  const accepted = result.decision.candidateRetained;
+  const status = accepted
+    ? "Source-paced CPU shadow принят; визуальное и integrated-graph качество ещё проверяются"
+    : "Source-paced CPU shadow не прошёл performance gate";
+  return (
+    <LaboratoryWorkTemplate
+      summary={(
+        <LaboratorySummary
+          title="M4.9T5 · полный source-paced TRAVEL TGS shadow"
+          description="Один CPU-only процесс прошёл весь recorded timeline RAVNOVES00 в исходном темпе. Камера остаётся владельцем времени; dense source cloud сохраняется, поверх него показывается четырёхсостояний TGS costmap."
+          status={status}
+          statusTone={accepted ? "success" : "danger"}
+          facts={[
+            { label: "Конфигурация", value: `${rigLabel} RIGHT · gravity-aligned LiDAR · causal ${number(result.configuration.historySeconds)} с` },
+            { label: "Timeline", value: `${result.timeline.frameCount.toLocaleString("ru-RU")} кадров · ${result.timeline.availableLidarFrameCount.toLocaleString("ru-RU")} LiDAR · ${result.timeline.missingLidarFrameCount} UNOBSERVED` },
+            { label: "Нагрузка", value: `Worker 006 CPU-only · GPU 0 · ${number(result.timeline.effectiveFps, 3)} source FPS` },
+            { label: "Authority", value: "REPLAY-SIMULATED · navigation/actuation OFF · integrated graph отдельно" },
+          ]}
+          brief={{
+            question: "Удерживает ли готовый TRAVEL TGS полный десятигерцовый replay без очереди и потери кадров?",
+            approach: "Все 4 489 camera frames планируются по исходным timestamps. Для 3 928 доступных LiDAR frames выполняется causal rolling 1 s; 561 пропуск остаётся полностью UNOBSERVED.",
+            principalResult: `${result.timeline.frameCount}/4 489 frames и ${result.pointAccounting.eligible.toLocaleString("ru-RU")} eligible points учтены; TGS p95/p99 ${number(result.performance.candidateTgsMs.p95, 2)}/${number(result.performance.candidateTgsMs.p99, 2)} мс, completion age p99 ${number(result.performance.completionAgeMs.p99, 2)} мс, capacity drops ${result.performance.capacityDropCount}.`,
+            limitation: "Это isolated CPU shadow. Он ещё не доказывает качество красных occupied-ячеек, проходимость для конкретного корпуса или регрессию FPS полного world-state graph.",
+          }}
+          method={{
+            completeness: "complete",
+            executionClass: "deterministic",
+            pipelineId: "travel-tgs-full-source-paced-shadow/v1",
+            components: [
+              { kind: "source", name: "RAVNOVES00", version: "4 489-frame recorded timeline", role: "camera-owned source clock + registered LiDAR", identitySha256: result.source.sourcePackSha256 },
+              { kind: "algorithm", name: "TRAVEL GroundSeg", version: "95dc2fbd66a343efd9060c45a5711b6307a950a4", role: "gravity-aligned ground/non-ground separation; AOS OFF", identitySha256: result.configuration.configSha256 },
+              { kind: "algorithm", name: "fail-closed costmap adapter", version: "v1", role: "occupied > rejected > ground > unobserved; no free inference", identitySha256: result.resultId.split("-").at(-1) ?? null },
+            ],
+          }}
+        />
+      )}
+      evidence={(
+        <LaboratoryEvidence
+          eyebrow="M4.9T5 VISUAL EVIDENCE · FULL CAMERA TIMELINE + SOURCE CLOUD + TGS COSTMAP"
+          title="Полный timeline: dense исходные точки сохранены, TGS-ячейки синхронны каждому кадру"
+          kind="recorded-replay"
+          resizable
+        >
+          <M49TgsFullShadowEvidence result={result} />
+        </LaboratoryEvidence>
+      )}
+      result={(
+        <LaboratoryResultSummary
+          title="Что доказал полный прогон"
+          status={status}
+          statusTone={accepted ? "success" : "danger"}
+          metrics={[
+            { label: "Timeline", value: `${result.timeline.frameCount}/4 489`, hint: `${result.timeline.availableLidarFrameCount} LiDAR + ${result.timeline.missingLidarFrameCount} explicit UNOBSERVED` },
+            { label: "TGS p95 / p99", value: `${number(result.performance.candidateTgsMs.p95, 2)} / ${number(result.performance.candidateTgsMs.p99, 2)} мс`, hint: "чистый candidate stage на CPU" },
+            { label: "Completion age p99", value: `${number(result.performance.completionAgeMs.p99, 2)} мс`, hint: "от source timestamp до готового frame result" },
+            { label: "Capacity drops", value: String(result.performance.capacityDropCount), hint: "кадры не отбрасывались ради темпа" },
+            { label: "Point accounting", value: "100%", hint: `${result.pointAccounting.eligible.toLocaleString("ru-RU")} eligible points` },
+          ]}
+          conclusion={{
+            proved: "Полный CPU-only TGS shadow воспроизводимо проходит recorded source clock, сохраняет fail-closed представление и не использует AOS/GPU.",
+            notProved: "Не приняты visual traversability, модель корпуса, камера-проекция TGS и нагрузка после встраивания в полный realtime world-state graph.",
+            decision: accepted
+              ? "Кандидат остаётся. Просмотреть полный timeline, затем подключить shadow к realtime graph и измерить общий FPS/latency regression."
+              : "Кандидат не встраивать; сначала локализовать performance gate, который не прошёл полный replay.",
+          }}
+        />
+      )}
+    />
+  );
+}

apps/control-station/src/workspaces/laboratory/M4ReplayThreatVisual.tsx

@@ -1,4 +1,4 @@
-import { useEffect, useMemo, useRef, useState, type CSSProperties } from "react";
+import { useCallback, useEffect, useMemo, useRef, useState, type CSSProperties } from "react";
 import {
   Button,
   Icon,
@@ -12,6 +12,7 @@ import {
 import { ObservationTimeline } from "../../components/ObservationTimeline";
 import {
   LaboratoryMetricEvidenceScene,
+  type LaboratoryMetricCellEvidence,
   type LaboratoryMetricEvidenceSceneHandle,
   type LaboratoryMetricSceneMode,
 } from "../../components/laboratory/LaboratoryMetricEvidenceScene";
@@ -103,6 +104,34 @@ export interface M4ReplayThreatReviewAnchor {
   sourceSequence: number;
   extentXyxyNormalized: readonly [number, number, number, number];
   matchedAtThreshold: boolean;
+  statusLabel?: string;
+}
+
+export interface M4ReplayClassifiedSpatialFrame {
+  sourceSequence: number;
+  sampleAvailable?: boolean;
+  sourcePointCount?: number;
+  pointsMapGravityLocalXyzM: readonly (readonly [number, number, number])[];
+  pointClassIds: readonly (number | null)[];
+  cellsMapGravityLocal: readonly {
+    centerXyM: readonly [number, number];
+    zBoundsM: readonly [number | null, number | null];
+    state: LaboratoryMetricCellEvidence["state"];
+  }[];
+  cellSizeM: number;
+  classes: readonly RecordedEvidenceSemanticClass[];
+  palette: readonly RecordedEvidenceSemanticPaletteEntry[];
+}
+
+export interface M4ReplayClassifiedSpatialLayer {
+  label: string;
+  pointLayerLabel: string;
+  cellLayerLabel: string;
+  expectedAtSequence: boolean;
+  frame: M4ReplayClassifiedSpatialFrame | null;
+  loading: boolean;
+  error: string | null;
+  replacePointCloud?: boolean;
 }
 
 const EMPTY_REVIEW_ANCHORS: readonly M4ReplayThreatReviewAnchor[] = [];
@@ -115,6 +144,9 @@ export function M4ReplayThreatVisual({
   reviewLabel = "Контрольные примеры M4.8R1",
   timelineEndpointRoot,
   evidenceLabel = "M4.6",
+  initialSpatialMode = null,
+  classifiedSpatialLayer,
+  onActiveSequenceChange,
 }: {
   resultId: string;
   semantic?: M4ReplayThreatSemanticLayer;
@@ -123,9 +155,14 @@ export function M4ReplayThreatVisual({
   reviewLabel?: string;
   timelineEndpointRoot?: string;
   evidenceLabel?: string;
+  initialSpatialMode?: LaboratoryMetricSceneMode | null;
+  classifiedSpatialLayer?: M4ReplayClassifiedSpatialLayer;
+  onActiveSequenceChange?: (sequence: number | null) => void;
 }) {
   const [mediaMode, setMediaMode] = useState<M4ThreatMediaMode | null>("video");
-  const [spatialMode, setSpatialMode] = useState<LaboratoryMetricSceneMode | null>(null);
+  const [spatialMode, setSpatialMode] = useState<LaboratoryMetricSceneMode | null>(
+    initialSpatialMode,
+  );
   const [showCurrentIncrement, setShowCurrentIncrement] = useState(true);
   const [showLocalSurface, setShowLocalSurface] = useState(true);
   const [showRollingMap, setShowRollingMap] = useState(true);
@@ -220,6 +257,9 @@ export function M4ReplayThreatVisual({
   }, [resultId]);
   if (timelineFrame.activeFrame) lastFrameRef.current = timelineFrame.activeFrame;
   const frame = timelineFrame.activeFrame ?? lastFrameRef.current;
+  useEffect(() => {
+    onActiveSequenceChange?.(timelineFrame.activeSequence);
+  }, [onActiveSequenceChange, timelineFrame.activeSequence]);
   const lastSpatialFrameRef = useRef<{
     resultId: string;
     frame: M4ThreatTimelineFrame;
@@ -303,12 +343,12 @@ export function M4ReplayThreatVisual({
     );
   }, [frame, metadata.timeline, showStaticObstacles]);
   const activeBoxes = useMemo(
-    () => [
+    () => classifiedSpatialLayer ? [] : [
       ...boxes(frame?.cameraProposals ?? []),
       ...staticObstacleBoxes,
       ...reviewAnchorBoxes,
     ],
-    [frame, reviewAnchorBoxes, staticObstacleBoxes],
+    [classifiedSpatialLayer, frame, reviewAnchorBoxes, staticObstacleBoxes],
   );
   const semanticClasses = useMemo<readonly RecordedEvidenceSemanticClass[]>(
     () => semantic?.taxonomy.map((item) => ({
@@ -363,6 +403,81 @@ export function M4ReplayThreatVisual({
       return status === 2 || status === 3 ? classId : null;
     });
   }, [semantic, semanticIntegrityError, showSpatialSemantic, spatialFrame, spatialSemanticFrame]);
+  const activeSpatialFrame = spatialFrame?.sequence === timelineFrame.activeSequence
+    ? spatialFrame
+    : null;
+  const classifiedSpatialFrame = classifiedSpatialLayer?.frame?.sourceSequence === timelineFrame.activeSequence
+    ? classifiedSpatialLayer?.frame ?? null
+    : null;
+  const lastClassifiedSpatialFrameRef = useRef<{
+    resultId: string;
+    frame: M4ReplayClassifiedSpatialFrame;
+  } | null>(null);
+  if (classifiedSpatialLayer?.frame) {
+    lastClassifiedSpatialFrameRef.current = { resultId, frame: classifiedSpatialLayer.frame };
+  }
+  const displayedClassifiedSpatialFrame = classifiedSpatialFrame
+    ?? (lastClassifiedSpatialFrameRef.current?.resultId === resultId
+      ? lastClassifiedSpatialFrameRef.current.frame
+      : null);
+  const replaceClassifiedPointCloud = classifiedSpatialLayer?.replacePointCloud ?? true;
+  const nominalSensorHeightM = metadata.timeline?.rig.nominalSensorHeightM ?? 0;
+  const mapGravityLocalSensorToBodyGround = useCallback((
+    point: readonly [number, number, number],
+  ): readonly [number, number, number] => {
+    const basis = activeSpatialFrame?.bodyFrame?.basisMapFromBody;
+    const rotated: readonly [number, number, number] = basis ? [
+      basis[0][0] * point[0] + basis[1][0] * point[1] + basis[2][0] * point[2],
+      basis[0][1] * point[0] + basis[1][1] * point[1] + basis[2][1] * point[2],
+      basis[0][2] * point[0] + basis[1][2] * point[1] + basis[2][2] * point[2],
+    ] : point;
+    // TGS evidence is translation-only map-gravity-local with the current LiDAR
+    // as its origin. The metric scene uses the body ground projection as z=0.
+    return [rotated[0], rotated[1], rotated[2] + nominalSensorHeightM];
+  }, [activeSpatialFrame?.bodyFrame?.basisMapFromBody, nominalSensorHeightM]);
+  const classifiedPointsBody = useMemo(
+    () => displayedClassifiedSpatialFrame?.pointsMapGravityLocalXyzM.map(
+      mapGravityLocalSensorToBodyGround,
+    ) ?? [],
+    [displayedClassifiedSpatialFrame, mapGravityLocalSensorToBodyGround],
+  );
+  const classifiedCellsBody = useMemo<readonly LaboratoryMetricCellEvidence[]>(
+    () => displayedClassifiedSpatialFrame?.cellsMapGravityLocal.map((cell) => {
+      const body = mapGravityLocalSensorToBodyGround([
+        cell.centerXyM[0],
+        cell.centerXyM[1],
+        0,
+      ]);
+      const [minimumSensorRelativeZ, maximumSensorRelativeZ] = cell.zBoundsM;
+      return {
+        centerBodyXyM: [body[0], body[1]],
+        zBoundsM: [
+          minimumSensorRelativeZ === null
+            ? null
+            : minimumSensorRelativeZ + nominalSensorHeightM,
+          maximumSensorRelativeZ === null
+            ? null
+            : maximumSensorRelativeZ + nominalSensorHeightM,
+        ],
+        state: cell.state,
+      };
+    }) ?? [],
+    [displayedClassifiedSpatialFrame, mapGravityLocalSensorToBodyGround, nominalSensorHeightM],
+  );
+  const classifiedCellCounts = useMemo(() => ({
+    ground: classifiedSpatialFrame?.cellsMapGravityLocal.filter(
+      (cell) => cell.state === "ground-support",
+    ).length ?? 0,
+    occupied: classifiedSpatialFrame?.cellsMapGravityLocal.filter(
+      (cell) => cell.state === "nonground-occupied",
+    ).length ?? 0,
+    rejected: classifiedSpatialFrame?.cellsMapGravityLocal.filter(
+      (cell) => cell.state === "unknown-rejected",
+    ).length ?? 0,
+    unobserved: classifiedSpatialFrame?.cellsMapGravityLocal.filter(
+      (cell) => cell.state === "unobserved",
+    ).length ?? 0,
+  }), [classifiedSpatialFrame]);
   const sceneObstacles = useMemo(() => spatialFrame?.metricObstacles.map((obstacle) => ({
     id: obstacle.componentId,
     decision: obstacle.assessment.decision,
@@ -522,7 +637,53 @@ export function M4ReplayThreatVisual({
     </div>
   ) : null;
 
-  const spatialLayerControls = (
+  const spatialLayerControls = classifiedSpatialLayer ? (
+    <div
+      className="m4-replay-threat-visual__pane-layer-controls"
+      role="group"
+      aria-label={`Слои ${classifiedSpatialLayer.label}`}
+    >
+      <Button
+        size="compact"
+        shape="pill"
+        variant={showCurrentIncrement ? "primary" : "secondary"}
+        aria-pressed={showCurrentIncrement}
+        onClick={() => setShowCurrentIncrement((visible) => !visible)}
+      >
+        {classifiedSpatialLayer.pointLayerLabel}
+      </Button>
+      <Button
+        size="compact"
+        shape="pill"
+        variant={showLocalSurface ? "primary" : "secondary"}
+        aria-pressed={showLocalSurface}
+        title="Bounded local SLAM surface · visual-derived"
+        onClick={() => setShowLocalSurface((visible) => !visible)}
+      >
+        LOCAL SLAM
+      </Button>
+      <Button
+        size="compact"
+        shape="pill"
+        variant={showRollingMap ? "primary" : "secondary"}
+        aria-pressed={showRollingMap}
+        onClick={() => setShowRollingMap((visible) => !visible)}
+      >
+        {classifiedSpatialLayer.cellLayerLabel}
+      </Button>
+      {semantic ? (
+        <Button
+          size="compact"
+          shape="pill"
+          variant={showSpatialSemantic ? "primary" : "secondary"}
+          aria-pressed={showSpatialSemantic}
+          onClick={() => setShowSpatialSemantic((visible) => !visible)}
+        >
+          SEMANTICS
+        </Button>
+      ) : null}
+    </div>
+  ) : (
     <div
       className="m4-replay-threat-visual__pane-layer-controls"
       role="group"
@@ -626,7 +787,7 @@ export function M4ReplayThreatVisual({
         value={String(selectedReviewAnchorIndex)}
         options={reviewAnchors.map((anchor, index) => ({
           value: String(index),
-          label: `${index + 1}/${reviewAnchors.length} · кадр ${anchor.sourceSequence + 1} · ${anchor.matchedAtThreshold ? "покрыт" : "пропуск"}`,
+          label: `${index + 1}/${reviewAnchors.length} · кадр ${anchor.sourceSequence + 1} · ${anchor.statusLabel ?? (anchor.matchedAtThreshold ? "покрыт" : "пропуск")}`,
         }))}
         variant="split"
         menuWidth="anchor"
@@ -652,7 +813,11 @@ export function M4ReplayThreatVisual({
     ? splitPrimarySize
     : 100;
 
-  const overlay = metadata.timeline && frame ? (
+  const overlaySequence = timelineFrame.activeSequence ?? frame?.sequence ?? null;
+  const overlaySessionSeconds = overlaySequence === null
+    ? null
+    : (metadata.timeline?.frameTimesNs[overlaySequence] ?? 0) / 1_000_000_000;
+  const overlay = metadata.timeline && overlaySequence !== null ? (
     <div
       className="l3-visual-audit__overlay m4-replay-threat-visual__overlay"
       style={{
@@ -661,9 +826,9 @@ export function M4ReplayThreatVisual({
     >
       <div>
         <span>RAVNOVES00 · recorded realtime</span>
-        <strong>frame {frame.sequence + 1}/{metadata.timeline.frameCount}</strong>
+        <strong>frame {overlaySequence + 1}/{metadata.timeline.frameCount}</strong>
         <small>
-          +{(frame.sessionSeconds - metadata.timeline.timelineStartSeconds).toFixed(3)} с
+          +{((overlaySessionSeconds ?? metadata.timeline.timelineStartSeconds) - metadata.timeline.timelineStartSeconds).toFixed(3)} с
           · {displayingBufferedFrame
             ? "держим последний кадр, следующий в буфере"
             : playbackController.playback.playing ? "воспроизведение" : "пауза / seek"}
@@ -671,39 +836,56 @@ export function M4ReplayThreatVisual({
       </div>
       <div>
         <span>Spatial evidence</span>
-        <strong>
-          {currentIncrementObstacles.length} current · {rollingMapObstacles.length} rolling
-          {metadata.timeline.occupancyProvenanceDelivery
-            ? ` · ${lowStepObstacles.length} low-step`
-            : ""}
-        </strong>
-        <small>
-          {spatialFrame
-            ? `${spatialFrame.pointCloudSampleCount}/${spatialFrame.pointCloudSourceCount} exact · ${localSurface.pointsBodyXyzM.length} local SLAM / ${localSurface.sourceFrameCount} frames`
-            : "квалифицированный spatial frame ещё не получен"}
-          {frame.worldStateAvailable
-            ? " · world-state delivered"
-            : ` · world-state gap (${frame.terminalOutcome})`}
-          {accumulatedCameraPoints
-            ? ` · camera points ${accumulatedCameraPoints.sampleCount}/${accumulatedCameraPoints.projectedPointCount} · causal ${accumulatedCameraPoints.windowSeconds.toFixed(1)} с / ${accumulatedCameraPoints.sourceFrameCount} frames`
-            : pointCloudOverlay
-              ? ` · camera points ${frame.cameraProjectedSampleCount}/${frame.cameraProjectedPointCount} exact-current · накопление загружается`
-              : showMediaPoints && cameraPointOverlay.error
-                ? " · накопленное camera cloud недоступно"
-                : ""}
-          {semantic && spatialSemanticFrame
-            ? ` · semantic L ${spatialSemanticFrame.counts.labeled} · A ${spatialSemanticFrame.counts.ambiguous} · U ${spatialSemanticFrame.counts.unprojected} · Ø ${spatialSemanticFrame.counts.absent}`
-            : semantic ? " · semantic buffer" : ""}
-        </small>
+        <strong>{classifiedSpatialLayer
+          ? classifiedSpatialFrame
+            ? replaceClassifiedPointCloud
+              ? `${classifiedSpatialFrame.pointsMapGravityLocalXyzM.length.toLocaleString("ru-RU")} TGS points · ${classifiedSpatialFrame.cellsMapGravityLocal.length.toLocaleString("ru-RU")} cells`
+              : `${(activeSpatialFrame?.pointCloudSourceCount ?? classifiedSpatialFrame.sourcePointCount ?? 0).toLocaleString("ru-RU")} source points · ${classifiedSpatialFrame.cellsMapGravityLocal.length.toLocaleString("ru-RU")} TGS cells`
+            : "TGS spatial buffer"
+          : `${currentIncrementObstacles.length} current · ${rollingMapObstacles.length} rolling${metadata.timeline.occupancyProvenanceDelivery ? ` · ${lowStepObstacles.length} low-step` : ""}`}</strong>
+        <small>{classifiedSpatialLayer
+          ? classifiedSpatialFrame
+            ? classifiedSpatialFrame.sampleAvailable === false
+              ? "LiDAR отсутствует · все ячейки принудительно UNOBSERVED · causal rolling 1 s"
+              : activeSpatialFrame
+                ? "map-gravity-local · all eligible points accounted · causal rolling 1 s"
+                : "TGS рассчитан · linked source cloud недоступен для этого кадра"
+            : classifiedSpatialLayer.error ?? `Открываем ${classifiedSpatialLayer.label}`
+          : (
+              <>
+                {spatialFrame
+                  ? `${spatialFrame.pointCloudSampleCount}/${spatialFrame.pointCloudSourceCount} exact · ${localSurface.pointsBodyXyzM.length} local SLAM / ${localSurface.sourceFrameCount} frames`
+                  : "квалифицированный spatial frame ещё не получен"}
+                {frame
+                  ? frame.worldStateAvailable
+                    ? " · world-state delivered"
+                    : ` · world-state gap (${frame.terminalOutcome})`
+                  : " · world-state frame unavailable"}
+                {accumulatedCameraPoints
+                  ? ` · camera points ${accumulatedCameraPoints.sampleCount}/${accumulatedCameraPoints.projectedPointCount} · causal ${accumulatedCameraPoints.windowSeconds.toFixed(1)} с / ${accumulatedCameraPoints.sourceFrameCount} frames`
+                  : pointCloudOverlay && frame
+                    ? ` · camera points ${frame.cameraProjectedSampleCount}/${frame.cameraProjectedPointCount} exact-current · накопление загружается`
+                    : showMediaPoints && cameraPointOverlay.error
+                      ? " · накопленное camera cloud недоступно"
+                      : ""}
+                {semantic && spatialSemanticFrame
+                  ? ` · semantic L ${spatialSemanticFrame.counts.labeled} · A ${spatialSemanticFrame.counts.ambiguous} · U ${spatialSemanticFrame.counts.unprojected} · Ø ${spatialSemanticFrame.counts.absent}`
+                  : semantic ? " · semantic buffer" : ""}
+              </>
+            )}</small>
       </div>
       <div>
-        <span>Virtual corridor</span>
-        <strong>
-          {spatialFrame?.decisionCounts.threat ?? 0} threat · nearest {nearest === null ? "—" : `${nearest.toFixed(2)} м`}
-        </strong>
-        <small>
-          {metadata.timeline.corridor.forwardLengthM} м · body {metadata.timeline.rig.lengthM}×{metadata.timeline.rig.widthM} м · REPLAY-SIMULATED
-        </small>
+        <span>{classifiedSpatialLayer ? "TGS fail-closed" : "Virtual corridor"}</span>
+        <strong>{classifiedSpatialLayer
+          ? classifiedSpatialFrame
+            ? `${classifiedCellCounts.occupied} occupied · ${classifiedCellCounts.rejected} rejected · ${classifiedCellCounts.unobserved} unobserved`
+            : classifiedSpatialLayer.loading || displayingBufferedFrame ? "loading" : "unavailable"
+          : `${spatialFrame?.decisionCounts.threat ?? 0} threat · nearest ${nearest === null ? "—" : `${nearest.toFixed(2)} м`}`}</strong>
+        <small>{classifiedSpatialLayer
+          ? classifiedSpatialFrame
+            ? `${classifiedCellCounts.ground} ground-support · visual review only · navigation authority OFF`
+            : "visual review only · navigation authority OFF"
+          : `${metadata.timeline.corridor.forwardLengthM} м · body ${metadata.timeline.rig.lengthM}×${metadata.timeline.rig.widthM} м · REPLAY-SIMULATED`}</small>
       </div>
     </div>
   ) : undefined;
@@ -802,27 +984,59 @@ export function M4ReplayThreatVisual({
             </div>
           </div>
         ) : null}
-        {spatialFrame ? (
+        {(!classifiedSpatialLayer ? spatialFrame : displayedClassifiedSpatialFrame) ? (
           <LaboratoryMetricEvidenceScene
             ref={metricSceneRef}
-            pointCloudBodyXyzM={spatialFrame.pointCloudBodyXyzM}
+            pointCloudBodyXyzM={displayedClassifiedSpatialFrame && replaceClassifiedPointCloud
+              ? classifiedPointsBody
+              : activeSpatialFrame?.pointCloudBodyXyzM ?? []}
             localSurfaceBodyXyzM={localSurface.pointsBodyXyzM}
-            obstacles={sceneObstacles}
+            obstacles={displayedClassifiedSpatialFrame ? [] : sceneObstacles}
             rig={timeline.rig}
             corridor={timeline.corridor}
-            occupiedVoxelSizeM={timeline.occupiedVoxelSizeM}
+            occupiedVoxelSizeM={displayedClassifiedSpatialFrame?.cellSizeM ?? timeline.occupiedVoxelSizeM}
             mode={spatialMode}
             label={`${evidenceLabel} exact current increment, bounded local SLAM surface and rolling occupancy`}
             showCurrentIncrement={showCurrentIncrement}
             showLocalSurface={showLocalSurface}
             showRollingMap={showRollingMap}
-            showLowStep={showLowStep}
-            pointSemanticClassIds={alignedSemanticPointIds}
-            semanticClasses={semanticClasses}
-            semanticPalette={semanticPalette}
+            showLowStep={displayedClassifiedSpatialFrame ? false : showLowStep}
+            pointSemanticClassIds={displayedClassifiedSpatialFrame && replaceClassifiedPointCloud
+              ? displayedClassifiedSpatialFrame.pointClassIds
+              : alignedSemanticPointIds}
+            semanticClasses={displayedClassifiedSpatialFrame && replaceClassifiedPointCloud
+              ? displayedClassifiedSpatialFrame.classes
+              : semanticClasses}
+            semanticPalette={displayedClassifiedSpatialFrame && replaceClassifiedPointCloud
+              ? displayedClassifiedSpatialFrame.palette
+              : semanticPalette}
+            classifiedCells={classifiedCellsBody}
+            classifiedCellSizeM={displayedClassifiedSpatialFrame?.cellSizeM}
+            showClassifiedCells={showRollingMap}
           />
         ) : null}
-        {frame && !frame.spatialAvailable ? (
+        {classifiedSpatialLayer && !displayedClassifiedSpatialFrame ? (
+          <div className="l3-visual-audit__state" role={classifiedSpatialLayer.error ? "alert" : "status"}>
+            {classifiedSpatialLayer.loading || displayingBufferedFrame
+              ? <span className="busy-indicator" aria-hidden="true" />
+              : <Icon name="alert" size={18} />}
+            <span>{classifiedSpatialLayer.error
+              ?? (classifiedSpatialLayer.loading || displayingBufferedFrame
+                ? `Открываем ${classifiedSpatialLayer.label}`
+                : classifiedSpatialLayer.expectedAtSequence
+                ? `Открываем ${classifiedSpatialLayer.label}`
+                : `${classifiedSpatialLayer.label} рассчитан только на 10 контрольных кадров.`)}</span>
+          </div>
+        ) : null}
+        {classifiedSpatialFrame?.sampleAvailable === false ? (
+          <div className="m4-replay-threat-visual__pane-status" role="status">
+            Кадр {classifiedSpatialFrame.sourceSequence + 1}: LiDAR отсутствует; все 2 244 TGS-ячейки явно UNOBSERVED.
+          </div>
+        ) : classifiedSpatialFrame && !activeSpatialFrame ? (
+          <div className="m4-replay-threat-visual__pane-status" role="status">
+            Кадр {classifiedSpatialFrame.sourceSequence + 1}: TGS costmap показан cell-only; linked source cloud для отрисовки отсутствует.
+          </div>
+        ) : frame && !frame.spatialAvailable ? (
           <div className="m4-replay-threat-visual__pane-status" role="status">
             {spatialFrame
               ? `На кадре ${frame.sequence + 1} нет body frame; держим spatial evidence кадра ${spatialFrame.sequence + 1}.`

apps/control-station/src/workspaces/laboratory/laboratoryArchiveProfiles.ts

@@ -105,6 +105,20 @@ const KNOWN_WORKS: Readonly<Record<Exclude<LaboratoryWorkId, `session:${string}`
     experimentName: "RF-DETR native risk review and temporal identity",
     variantName: "M4.8Q · native raw KB4 review · quality not adjudicated",
   },
+  "m49-tgs-fail-closed-evidence": {
+    profileId: "rig-dual-evidence-virtual-corridor-v1",
+    profileName: (rigLabel) => `${rig(rigLabel)} RIGHT · Camera + gravity-aligned LiDAR`,
+    experimentId: "m49-tgs-fail-closed-evidence",
+    experimentName: "TRAVEL TGS fail-closed traversability evidence",
+    variantName: "M4.9T4 · 10 anchors · causal rolling 1 s · AOS OFF",
+  },
+  "m49-tgs-full-shadow": {
+    profileId: "rig-dual-evidence-virtual-corridor-v1",
+    profileName: (rigLabel) => `${rig(rigLabel)} RIGHT · Camera + gravity-aligned LiDAR`,
+    experimentId: "m49-tgs-full-shadow",
+    experimentName: "TRAVEL TGS complete source-paced shadow",
+    variantName: "M4.9T5 · 4 489 frames · causal rolling 1 s · CPU-only",
+  },
   "m47-reference-graph-shadow": {
     profileId: "rig-dual-evidence-virtual-corridor-v1",
     profileName: (rigLabel) => `${rig(rigLabel)} RIGHT · Camera + LiDAR dual evidence`,

apps/control-station/src/workspaces/laboratory/useAdvancedLaboratoryCatalog.ts

@@ -25,6 +25,8 @@ function mergeResults(
     m48r3StaticOccupancy: next.m48r3StaticOccupancy ?? current.m48r3StaticOccupancy,
     m48s: next.m48s ?? current.m48s,
     m48t: next.m48t ?? current.m48t,
+    m49Tgs: next.m49Tgs ?? current.m49Tgs,
+    m49TgsFull: next.m49TgsFull ?? current.m49TgsFull,
     m4Threat: next.m4Threat ?? current.m4Threat,
     l3: next.l3 ?? current.l3,
     l31: next.l31 ?? current.l31,
@@ -124,6 +126,8 @@ export function useAdvancedLaboratoryCatalog({
         "m48-small-static-passage-regression",
         "m48-static-occupancy-qualification",
         "m48r3-static-occupancy-shadow",
+        "m49-tgs-fail-closed-evidence",
+        "m49-tgs-full-shadow",
       ].includes(selectedWorkId)
       && !indexedResultId
     ) return;

apps/control-station/test/advancedLaboratoryResults.test.mjs

@@ -7,6 +7,7 @@ let server;
 let fetchAdvancedLaboratoryResults;
 let fetchAdvancedLaboratoryIndex;
 let fetchAdvancedLaboratoryResult;
+let fetchM49TgsAnchorSpatial;
 let AdvancedLaboratoryContractError;
 let buildLaboratoryCatalog;
 let buildLaboratoryProfiles;
@@ -836,6 +837,77 @@ function e40() {
   };
 }
 
+function m49View() {
+  const anchors = [171, 306, 368, 402, 450, 509, 525, 744, 1122, 1856];
+  return {
+    schema_version: "missioncore.m49-tgs-fail-closed-view/v1",
+    result_id: `m49-tgs-fail-closed-${"2".repeat(64)}`,
+    created_at_utc: "2026-08-26T17:09:43Z",
+    source: {
+      source_id: "RAVNOVES00",
+      source_session_id: "20260720T065719Z_viewer_live",
+      source_pack_sha256: "3".repeat(64),
+      linked_visual_result_id: `m4-threat-replay-${"4".repeat(64)}`,
+      anchor_frame_indices: anchors,
+    },
+    configuration: {
+      profile_id: "m49-ravnoves00-tgs-fail-closed-evidence/v1",
+      config_sha256: "5".repeat(64),
+      coordinate_frame: "map-gravity-local",
+      primary_profile: "causal_rolling_1s",
+      cell_size_m: 0.45,
+      radius_m: 12,
+    },
+    execution: {
+      worker: "Worker 006",
+      device: "cpu",
+      gpu_used: false,
+      wrapper_elapsed_seconds: 12.1,
+    },
+    metrics: {
+      anchor_count: 10,
+      anchor_profile_count: 20,
+      all_eligible_points_accounted: true,
+      costmap_cell_count: 2244,
+      process_wall_current_p50_ms: 20,
+      process_wall_current_max_ms: 30,
+      process_wall_rolling_p50_ms: 30,
+      process_wall_rolling_max_ms: 40,
+      process_max_rss_kib: 9292,
+      primary: anchors.map((anchor, slot) => ({
+        anchor_frame_index: anchor,
+        slot,
+        profile_id: "causal_rolling_1s",
+        point_count: 3,
+        ground_point_count: 1,
+        nonground_point_count: 1,
+        rejected_point_count: 1,
+        ground_cell_count: 1,
+        nonground_cell_count: 1,
+        rejected_cell_count: 1,
+        unobserved_cell_count: 2241,
+        all_points_accounted: true,
+      })),
+    },
+    acceptance: {
+      representation_complete: true,
+      all_points_accounted: true,
+      aos_absent: true,
+      gpu_absent: true,
+      visual_quality_accepted: false,
+      traversability_accepted: false,
+    },
+    decision: {
+      state: "visual-review-required",
+      candidate_retained: true,
+      next_action: "Review the ten anchors.",
+    },
+    limitations: ["bounded diagnostic evidence"],
+    ground_truth: false,
+    access: "read-only",
+  };
+}
+
 before(async () => {
   server = await createServer({
     appType: "custom",
@@ -850,6 +922,9 @@ before(async () => {
     fetchAdvancedLaboratoryIndex,
     fetchAdvancedLaboratoryResult,
   } = await server.ssrLoadModule("/src/core/laboratory/advancedIndex.ts"));
+  ({ fetchM49TgsAnchorSpatial } = await server.ssrLoadModule(
+    "/src/core/laboratory/m49TgsFailClosed.ts",
+  ));
   ({
     buildLaboratoryCatalog,
     buildLaboratoryProfiles,
@@ -1596,6 +1671,62 @@ test("selected advanced LAB fetches only its own strict catalog", async () => {
   ]);
 });
 
+test("selected M49 LAB preserves the sealed fail-closed contract", async () => {
+  const payload = m49View();
+  const requests = [];
+  const decoded = await fetchAdvancedLaboratoryResult(
+    "m49-tgs-fail-closed-evidence",
+    {
+      resultId: payload.result_id,
+      fetcher: async (input, init) => {
+        requests.push({ input: String(input), method: init?.method });
+        return new Response(JSON.stringify(payload), { status: 200 });
+      },
+    },
+  );
+
+  assert.equal(decoded.m49Tgs.metrics.anchorCount, 10);
+  assert.equal(decoded.m49Tgs.metrics.costmapCellCount, 2244);
+  assert.equal(decoded.m49Tgs.acceptance.visualQualityAccepted, false);
+  assert.deepEqual(requests, [{
+    input: `/api/v1/laboratory/m49/tgs-fail-closed/${payload.result_id}`,
+    method: "GET",
+  }]);
+});
+
+test("M49 anchor fetch keeps every point and all four costmap states", async () => {
+  const resultId = m49View().result_id;
+  const centers = Array.from({ length: 2244 }, (_, index) => [index * 0.45, 0]);
+  const states = Array.from({ length: 2244 }, (_, index) => index % 4);
+  const zBounds = states.map((state) => state === 0 ? [null, null] : [0, 0.2]);
+  const decoded = await fetchM49TgsAnchorSpatial(resultId, 171, "causal_rolling_1s", {
+    fetcher: async () => new Response(JSON.stringify({
+      schema_version: "missioncore.m49-tgs-anchor-spatial/v1",
+      result_id: resultId,
+      anchor_frame_index: 171,
+      source_sequence: 171,
+      profile: "causal_rolling_1s",
+      coordinate_frame: "map-gravity-local",
+      points_xyz_m: [[1, 2, 3], [4, 5, 6], [7, 8, 9]],
+      point_states: [1, 2, 3],
+      costmap: {
+        cell_size_m: 0.45,
+        radius_m: 12,
+        centers_xy_m: centers,
+        states,
+        z_bounds_m: zBounds,
+      },
+      all_points_accounted: true,
+      aos_used: false,
+      access: "read-only",
+    }), { status: 200 }),
+  });
+
+  assert.deepEqual(decoded.pointStates, [1, 2, 3]);
+  assert.equal(decoded.costmap.states.length, 2244);
+  assert.deepEqual(new Set(decoded.costmap.states), new Set([0, 1, 2, 3]));
+});
+
 test("keeps valid LAB catalogs available when one transport endpoint fails", async () => {
   const decoded = await fetchAdvancedLaboratoryResults({
     fetcher: async (input) => {

apps/control-station/test/m49TgsFullShadow.test.mjs

@@ -0,0 +1,95 @@
+import assert from "node:assert/strict";
+import { readFile } from "node:fs/promises";
+import { after, before, test } from "node:test";
+
+import { createServer } from "vite";
+
+let server;
+let fetchM49TgsFullShadowSpatialChunk;
+
+const resultId = `m49-tgs-full-shadow-${"a".repeat(64)}`;
+
+before(async () => {
+  server = await createServer({
+    appType: "custom",
+    logLevel: "silent",
+    server: { middlewareMode: true },
+  });
+  ({ fetchM49TgsFullShadowSpatialChunk } = await server.ssrLoadModule(
+    "/src/core/laboratory/m49TgsFullShadow.ts",
+  ));
+});
+
+after(async () => {
+  await server?.close();
+});
+
+test("M4.9T5 chunk keeps every missing-LiDAR cell explicitly UNOBSERVED", async () => {
+  let requestedUrl = "";
+  const centers = Array.from({ length: 2244 }, (_, index) => [index * 0.45, 0]);
+  const unobserved = Array.from({ length: 2244 }, () => 0);
+  const zBounds = Array.from({ length: 2244 }, () => [null, null]);
+  const metrics = {
+    eligible_point_count: 0,
+    ground_point_count: 0,
+    nonground_point_count: 0,
+    rejected_point_count: 0,
+    occupied_cell_count: 0,
+  };
+  const chunk = await fetchM49TgsFullShadowSpatialChunk(resultId, 7, 1, {
+    fetcher: async (url) => {
+      requestedUrl = String(url);
+      return new Response(JSON.stringify({
+        schema_version: "missioncore.m49-tgs-full-shadow-spatial-chunk/v1",
+        result_id: resultId,
+        start: 7,
+        count: 1,
+        coordinate_frame: "map-gravity-local",
+        costmap: {
+          cell_size_m: 0.45,
+          radius_m: 12,
+          centers_xy_m: centers,
+        },
+        frames: [{
+          source_sequence: 7,
+          source_frame_index: 7,
+          session_seconds: 36.119857292,
+          sample_available: false,
+          states: unobserved,
+          z_bounds_m: zBounds,
+          metrics,
+        }],
+      }), { status: 200, headers: { "Content-Type": "application/json" } });
+    },
+  });
+
+  assert.equal(
+    requestedUrl,
+    `/api/v1/laboratory/m49/tgs-full-shadow/${resultId}/spatial/chunk?start=7&count=1`,
+  );
+  assert.equal(chunk.count, 1);
+  assert.equal(chunk.frames[0].sampleAvailable, false);
+  assert.equal(chunk.frames[0].costmap.states.length, 2244);
+  assert.deepEqual(new Set(chunk.frames[0].costmap.states), new Set([0]));
+});
+
+test("M4.9T5 viewer prefetches 24-frame immutable chunks", async () => {
+  const [source, contract] = await Promise.all([
+    readFile(
+      new URL("../src/workspaces/laboratory/M49TgsFullShadowEvidence.tsx", import.meta.url),
+      "utf8",
+    ),
+    readFile(
+      new URL("../src/core/laboratory/m49TgsFullShadow.ts", import.meta.url),
+      "utf8",
+    ),
+  ]);
+  assert.match(source, /const CHUNK_FRAMES = 24/);
+  assert.match(source, /activeChunkStart \+ CHUNK_FRAMES/);
+  assert.match(source, /fetchM49TgsFullShadowSpatialChunk/);
+  assert.match(source, /sampleAvailable: spatial\.sampleAvailable/);
+  assert.match(source, /fetchE47SemanticSlamResult/);
+  assert.match(source, /next\.baseM4ResultId !== result\.source\.linkedVisualResultId/);
+  assert.match(source, /semantic=\{semantic \? \{/);
+  assert.match(contract, /linked_semantic_result_id/);
+});

apps/control-station/test/m4ReplayThreat.test.mjs

@@ -799,7 +799,27 @@ test("M4.6 viewer keeps media and spatial panes on one playback clock", async ()
   assert.match(metricScene, /OrbitControls/);
   assert.match(visual, /LOCAL SLAM/);
   assert.match(visual, /showLocalSurface/);
-  assert.match(visual, /pointCloudBodyXyzM=\{spatialFrame\.pointCloudBodyXyzM\}/);
+  assert.match(
+    visual,
+    /pointCloudBodyXyzM=\{displayedClassifiedSpatialFrame && replaceClassifiedPointCloud[\s\S]*\? classifiedPointsBody[\s\S]*: activeSpatialFrame\?\.pointCloudBodyXyzM \?\? \[\]\}/,
+  );
+  assert.match(
+    visual,
+    /const classifiedSpatialFrame = classifiedSpatialLayer\?\.frame\?\.sourceSequence === timelineFrame\.activeSequence[\s\S]*const displayedClassifiedSpatialFrame = classifiedSpatialFrame[\s\S]*lastClassifiedSpatialFrameRef/,
+  );
+  assert.match(visual, /localSurfaceBodyXyzM=\{localSurface\.pointsBodyXyzM\}/);
+  assert.match(visual, /showLocalSurface=\{showLocalSurface\}/);
+  assert.match(visual, /\{semantic \? \([\s\S]*>\s*SEMANTICS\s*<\/Button>/);
+  assert.match(visual, /все 2 244 TGS-ячейки явно UNOBSERVED/);
+  assert.match(
+    visual,
+    /mapGravityLocalSensorToBodyGround[\s\S]*rotated\[2\] \+ nominalSensorHeightM/,
+  );
+  assert.match(
+    visual,
+    /zBoundsM: \[[\s\S]*minimumSensorRelativeZ \+ nominalSensorHeightM[\s\S]*maximumSensorRelativeZ \+ nominalSensorHeightM/,
+  );
+  assert.match(visual, /classifiedCells=\{classifiedCellsBody\}/);
   assert.match(metricScene, /Локальная SLAM-поверхность/);
   assert.match(visual, /showJumpToEnd=\{false\}/);
   assert.doesNotMatch(visual, /Назад на 5 секунд/);

config/laboratories/m49-tgs-fail-closed-evidence.json

@@ -0,0 +1,10 @@
+{
+  "schema_version": "missioncore.laboratory-evidence-definition/v1",
+  "work_id": "m49-tgs-fail-closed-evidence",
+  "evidence": {
+    "runtime_relative_root": "m49/tgs-fail-closed-results",
+    "result_id_prefix": "m49-tgs-fail-closed",
+    "document_name": "manifest.json",
+    "schema_version": "missioncore.m49-tgs-fail-closed-result/v1"
+  }
+}

config/laboratories/m49-tgs-full-shadow.json

@@ -0,0 +1,10 @@
+{
+  "schema_version": "missioncore.laboratory-evidence-definition/v1",
+  "work_id": "m49-tgs-full-shadow",
+  "evidence": {
+    "runtime_relative_root": "m49/tgs-full-shadow-results",
+    "result_id_prefix": "m49-tgs-full-shadow",
+    "document_name": "manifest.json",
+    "schema_version": "missioncore.m49-tgs-full-shadow-lab/v1"
+  }
+}

config/laboratory-execution.json

@@ -162,6 +162,34 @@
         "run": "missioncore.laboratory-run/v1",
         "evidence": "missioncore.m48t-risk-quality-temporal-lab/v1"
       }
+    },
+    {
+      "work_id": "m49-tgs-fail-closed-evidence",
+      "lifecycle": "experimental",
+      "isolation": "bounded-adapter",
+      "adapter_id": "experimental.m49-tgs-fail-closed-evidence/v1",
+      "input_roles": ["repository_root"],
+      "contracts": {
+        "source": "missioncore.m49-tgs-worker-evidence/v1",
+        "provider": "missioncore.travel-tgs-ground-segmentation/v1",
+        "graph": "missioncore.m49-fail-closed-costmap-evidence/v1",
+        "run": "missioncore.laboratory-run/v1",
+        "evidence": "missioncore.m49-tgs-fail-closed-result/v1"
+      }
+    },
+    {
+      "work_id": "m49-tgs-full-shadow",
+      "lifecycle": "experimental",
+      "isolation": "bounded-adapter",
+      "adapter_id": "experimental.m49-tgs-full-shadow/v1",
+      "input_roles": ["repository_root"],
+      "contracts": {
+        "source": "missioncore.m49-tgs-full-shadow-result/v1",
+        "provider": "missioncore.travel-tgs-ground-segmentation/v1",
+        "graph": "missioncore.m49-tgs-full-source-paced-shadow/v1",
+        "run": "missioncore.laboratory-run/v1",
+        "evidence": "missioncore.m49-tgs-full-shadow-lab/v1"
+      }
     }
   ],
   "legacy_work_ids": [

config/laboratory-value-review.json

@@ -267,6 +267,20 @@
       "signal": "progress",
       "lifecycle": "current",
       "visual_evidence": "available"
+    },
+    {
+      "catalog_id": "m49-tgs-fail-closed-evidence",
+      "evidence_id": "m49-tgs-fail-closed-9d5cb089bb5cc23f829acb47eda52eaa886db3b60f8fa7d57e02e27f642e837b",
+      "signal": "progress",
+      "lifecycle": "current",
+      "visual_evidence": "available"
+    },
+    {
+      "catalog_id": "m49-tgs-full-shadow",
+      "evidence_id": "m49-tgs-full-shadow-0faeaaf3aba8dccae974eab51ff9cccf264a13ec285abb1920c1e5a09a7e87bc",
+      "signal": "progress",
+      "lifecycle": "current",
+      "visual_evidence": "available"
     }
   ]
 }

config/perception/m49-tgs-full-shadow-v1.json

@@ -0,0 +1,77 @@
+{
+  "schema_version": "missioncore.m49-tgs-full-shadow-profile/v1",
+  "profile_id": "m49-ravnoves00-tgs-full-shadow/v1",
+  "source": {
+    "source_id": "RAVNOVES00",
+    "session_id": "20260720T065719Z_viewer_live",
+    "source_pack_id": "e10-lidar-pack-576c994a6c814e2592dd6240ace3902a5db94843312c759a73ba0c9166157d2b",
+    "source_pack_sha256": "0685d24219d8236caf8b7f1685e93f6d6b59e7fd015a768d88a92bbe8b154944",
+    "travel_revision": "95dc2fbd66a343efd9060c45a5711b6307a950a4",
+    "input_coordinate_frame": "map-gravity-local-translation-only",
+    "expected_timeline_frames": 4489,
+    "expected_available_lidar_frames": 3928
+  },
+  "tgs": {
+    "max_range_m": 80.0,
+    "min_range_m": 1.0,
+    "resolution_m": 8.0,
+    "num_iterations": 3,
+    "num_lowest_representative_points": 5,
+    "minimum_points": 10,
+    "seed_threshold_m": 0.5,
+    "distance_threshold_m": 0.125,
+    "outlier_threshold_m": 0.3,
+    "normal_threshold": 0.94,
+    "weight_threshold": 200.0,
+    "lcc_normal_similarity": 0.03,
+    "lcc_planar_distance_m": 0.1,
+    "obstacle_height_m": 1.0,
+    "refine_mode": true
+  },
+  "profile": {
+    "id": "causal_rolling_1s",
+    "history_seconds": 1.0,
+    "local_radius_m": 12.0,
+    "missing_lidar_policy": "all-cells-unobserved"
+  },
+  "costmap": {
+    "coordinate_frame": "map-gravity-local",
+    "cell_size_m": 0.45,
+    "radius_m": 12.0,
+    "state_priority": [
+      "NONGROUND_OCCUPIED",
+      "UNKNOWN_REJECTED",
+      "GROUND_SUPPORT",
+      "UNOBSERVED"
+    ]
+  },
+  "state_codes": {
+    "UNOBSERVED": 0,
+    "GROUND_SUPPORT": 1,
+    "NONGROUND_OCCUPIED": 2,
+    "UNKNOWN_REJECTED": 3
+  },
+  "acceptance": {
+    "recorded_source_rate_hz": 10.0,
+    "minimum_effective_timeline_fps": 10.0,
+    "candidate_stage_p95_ms_max": 25.0,
+    "candidate_stage_p99_ms_max": 50.0,
+    "completion_age_p99_ms_max": 100.0,
+    "capacity_drop_count_max": 0,
+    "unaccounted_available_frame_count_max": 0,
+    "unaccounted_eligible_point_count_max": 0
+  },
+  "invariants": {
+    "all_eligible_input_points_accounted": true,
+    "aos_allowed": false,
+    "lidar_orientation_applied_to_tgs_input": false,
+    "map_gravity_axis_preserved": true,
+    "missing_support_means_free": false,
+    "missing_lidar_means_unobserved": true,
+    "unobserved_cells_are_emitted": true,
+    "future_frames_used": false,
+    "camera_projection_is_authoritative": false,
+    "gpu_allowed": false,
+    "navigation_or_actuation_allowed": false
+  }
+}

docs/23_TRAVERSABILITY_AND_STATIC_OBSTACLE_STACK_DECISION.md

@@ -326,7 +326,23 @@ invocation. It emits a deterministic `0.45 m`, `12 m` local evidence grid while
 preserving unobserved cells. See
 [`experiments/perception/M49_TGS_FAIL_CLOSED_EVIDENCE_2026-08-26.md`](../experiments/perception/M49_TGS_FAIL_CLOSED_EVIDENCE_2026-08-26.md).
 Representation/accounting is accepted for the visual gate; obstacle and
-traversability quality are not yet accepted.
+traversability quality are not yet accepted. The sealed result is now published
+in the canonical LAB as
+`m49-tgs-fail-closed-9d5cb089bb5cc23f829acb47eda52eaa886db3b60f8fa7d57e02e27f642e837b`,
+linked to the current M4 v3 recorded timeline. Publication adds no navigation
+or actuation authority.
+
+The subsequent complete source-paced TGS-only shadow passed on `2026-08-26`.
+It accounted for all `4,489` timeline frames and all `63,646,163` eligible
+points, emitted the `561` missing-LiDAR frames as explicit all-cell
+`UNOBSERVED`, sustained `10.003945 FPS`, measured TGS `p95/p99` at
+`1.694/2.08973 ms`, completion-age `p99` at `41.42845548 ms`, and recorded zero
+capacity drops. The immutable result and canonical LAB acceptance are recorded
+in
+[`experiments/perception/M49_TGS_FULL_SHADOW_ACCEPTANCE_2026-08-26.md`](../experiments/perception/M49_TGS_FULL_SHADOW_ACCEPTANCE_2026-08-26.md).
+This accepts the CPU source-paced shadow and retains the candidate. It does not
+accept visual traversability quality, integrated graph performance, navigation
+or actuation.
 
 ### T4 — Candidate C occupancy/ESDF probe
 
@@ -345,12 +361,15 @@ traversability quality are not yet accepted.
 
 ## Immediate next action
 
-Import the sealed gravity-aligned TGS evidence pack into one laboratory review
-surface and review the ten anchors in metric 3D/costmap space. Preserve
+Review the published full gravity-aligned TGS timeline in camera, metric 3D and
+costmap space. Preserve
 `GROUND_SUPPORT`, `NONGROUND_OCCUPIED`, `UNKNOWN_REJECTED` and `UNOBSERVED` as
 separate products; do not use AOS or infer free cells from absent
-republication. Add diagnostic camera projection only after the metric evidence
-is accepted. Candidate A T2 replay remains blocked by the failed unmodified T1
-gate. No LOW-STEP tuning, new object model, camera resize, fisheye
+republication. Attach the unchanged candidate to the realtime world-state graph
+in non-authoritative shadow mode and measure complete-graph FPS, latency,
+resource use and queue drops. If occupied evidence carpets the route, soft
+traversable vegetation or usable gaps, reject TGS without tuning it against the
+review evidence. Candidate A T2 replay remains blocked by the failed unmodified
+T1 gate. No LOW-STEP tuning, new object model, camera resize, fisheye
 rectification, manual dataset or parallel heavy Worker job is authorized by
 this decision.

experiments/perception/M49_TGS_FAIL_CLOSED_EVIDENCE_2026-08-26.md

@@ -1,7 +1,8 @@
 # M4.9 gravity-aligned TGS fail-closed evidence — 2026-08-26
 
-Status: **evidence adapter completed**; visual obstacle/traversability quality,
-full-source realtime, navigation and actuation remain disabled
+Status: **evidence adapter and canonical LAB publication completed**; visual
+obstacle/traversability quality, full-source realtime, navigation and actuation
+remain disabled
 
 ## Decision
 
@@ -13,7 +14,9 @@ accounts for every point inside TRAVEL's declared `1–80 m` processing range.
 This closes the representation and accounting gate. It does not accept visual
 quality. Several anchors still contain a high non-ground share, so the next
 gate is a 3D/costmap review of the sealed output rather than threshold tuning or
-camera boxes.
+camera boxes. The metric output is now available in the canonical LAB on port
+`8000` as an operator-review surface; publication does not upgrade it to an
+accepted terrain provider.
 
 ## Frozen identity
 
@@ -125,15 +128,73 @@ Worker root:
 | `inputs/input-manifest.json` | `13,357` | `7a22e61606e1fa9ab61d47417f941b590a68b29e4529d4c290992ce7eb89a642` |
 | `tgs-timing.tsv` | `638` | `a8f439996205286335ac392b37f51b54056d04e9cb07e03648c52c1e8f2f6ae8` |
 
+## Canonical LAB publication
+
+- LAB result:
+  `m49-tgs-fail-closed-9d5cb089bb5cc23f829acb47eda52eaa886db3b60f8fa7d57e02e27f642e837b`;
+- linked current visual timeline:
+  `m4-threat-replay-2a953c5f27f2a5b1dddc5c658c1de2c323d7796084a099c024987a1da03aa324`;
+- source LiDAR pack SHA-256 remains
+  `0685d24219d8236caf8b7f1685e93f6d6b59e7fd015a768d88a92bbe8b154944`;
+- the earlier immutable seal beginning `200b57cc` is superseded only because it
+  linked the retired M4 v2 visual profile, whose timeline is incompatible with
+  the current v3 API. No Worker evidence byte or TGS result was changed.
+
+The LAB reuses the canonical recorded-realtime camera and spatial instrument.
+It exposes the ten anchors, the full classified TGS point set, the four-state
+costmap, `VIDEO/CAMERA`, `3D/PLAN`, independent point/costmap visibility and
+exact per-anchor counts. Camera proposal boxes, AOS clusters, future frames and
+navigation authority are absent. Browser acceptance confirmed that every
+operator control is independent and that a loading anchor cannot temporarily
+display unrelated M4 obstacle statistics.
+
+The published TGS coordinates are intentionally sensor-origin
+`map-gravity-local`: the ground is therefore approximately `-1.25 m` below the
+current LiDAR. The canonical LAB presentation converts these immutable values
+to the scene's body-ground origin by adding the rig's sealed
+`nominal_sensor_height_m = 1.25` after the gravity-to-body rotation. The same
+offset is applied to point heights and costmap z-bounds; x/y evidence and the
+sealed Worker result remain unchanged. A regression check across all ten
+anchors puts the median displayed ground points in `-0.299…+0.135 m` and the
+median ground-cell centres in `-0.308…+0.237 m`. On anchor `1856`, where the
+presentation defect was reported, those medians are `+0.004 m` and `-0.063 m`.
+The residual variation is measured terrain/pose variation, not the previous
+systematic `1.25 m` lift.
+
+## Preliminary corridor triage
+
+The published evidence was also projected into the frozen virtual body
+corridor (`-0.5…8.0 m` longitudinal, `±0.5 m` lateral) without changing the
+sealed result. The table counts costmap cell centres inside that corridor:
+
+| Anchor | Current occupied | Rolling occupied | Rolling unobserved | Preliminary reading |
+| ---: | ---: | ---: | ---: | --- |
+| `171` | `0` | `8` | `6` | unresolved rolling band at `5.63…7.90 m` |
+| `306` | `0` | `0` | `12` | no corridor obstruction |
+| `368` | `1` | `1` | `16` | compact occupied support at `6.34 m` |
+| `402` | `0` | `1` | `10` | compact occupied support at `1.89 m` |
+| `450` | `0` | `0` | `8` | no corridor obstruction |
+| `509` | `0` | `0` | `7` | no corridor obstruction |
+| `525` | `0` | `0` | `9` | no corridor obstruction |
+| `744` | `0` | `3` | `25` | weakly observed; requires explicit review |
+| `1122` | `0` | `2` | `24` | rolling recovers compact obstacle support |
+| `1856` | `1` | `1` | `13` | compact occupied support at `4.14 m` |
+
+This is a diagnostic, not ground truth. It shows why the causal window cannot
+simply be removed: anchor `1122` is entirely unobserved in the current-only
+corridor but gains ground and occupied evidence in rolling. It also identifies
+the opposing risk: at anchor `171`, rolling adds eight occupied cells where the
+camera/PLAN comparison appears to show an open route apart from side geometry.
+Anchor `744` remains too weakly observed for acceptance. These two anchors keep
+the visual gate open even though the remaining frames do not show a continuous
+occupied carpet.
+
 ## Next gate
 
-Import the sealed `evidence.npz` into one laboratory review surface with:
-
-1. 3D points colored by the four evidence states;
-2. top-down `0.45 m` costmap cells in the same local coordinate frame;
-3. anchor selection and exact counts from `result.json`;
-4. no AOS clusters, camera boxes, future frames or physical authority.
-
-The operator review must focus on the mandatory obstacle/gap cases and on the
-large non-ground populations listed above. Camera projection follows only if
-the metric evidence is accepted.
+Review the published ten anchors in both metric `3D` and `PLAN`. The operator
+must focus on the mandatory obstacle/gap cases and the large non-ground
+populations listed above. Accept TGS for a full recorded-source shadow only if
+red non-ground evidence remains on real geometry without carpeting the visible
+route, soft traversable vegetation or usable gaps. Otherwise reject this
+candidate without tuning thresholds against these ten anchors. Camera
+projection follows only if the metric evidence is accepted.

experiments/perception/M49_TGS_FULL_SHADOW_ACCEPTANCE_2026-08-26.md

@@ -0,0 +1,105 @@
+# M4.9T5 full source-paced TRAVEL TGS shadow — 2026-08-26
+
+Status: **recorded source-paced CPU shadow accepted; candidate retained**.
+Visual traversability quality, full-graph performance, navigation and actuation
+remain unaccepted.
+
+## Decision
+
+The gravity-aligned TGS-only candidate completed the entire `RAVNOVES00`
+recorded timeline on Worker 006. Every source timeline frame and every eligible
+LiDAR point is represented. No frame was dropped for capacity, AOS was never
+invoked, no GPU was requested and the canonical Triton container remained
+healthy with the same identity.
+
+This accepts TGS as a bounded CPU shadow candidate. It does not yet make TGS a
+navigation authority or prove that its occupied/ground interpretation is
+correct for vegetation, terrain, gaps or the future vehicle envelope.
+
+## Frozen identity
+
+| Item | Identity |
+| --- | --- |
+| LAB result | `m49-tgs-full-shadow-ef98de7db7596d48e8c8c0549ce68e6704ee03e87c8c4bcf1e3e748b7ccb032e` |
+| Worker run | `Worker 006 / ravnoves00-full-001` |
+| Mission Core revision used by Worker | `40c850b167dda366d8aa45d828520168affaf9fd` |
+| Deterministic Worker artifact | `5e0ea16c7a5cc760463836718b0cd8b0006ffc4b202e5f706a20a86ef2f912ab` |
+| Source pack SHA-256 | `0685d24219d8236caf8b7f1685e93f6d6b59e7fd015a768d88a92bbe8b154944` |
+| TGS config SHA-256 | `c2e07010aaee78259d36c057962d6bfb885349251ff7356d867e5813e632881c` |
+| Linked visual result | `m4-threat-replay-2a953c5f27f2a5b1dddc5c658c1de2c323d7796084a099c024987a1da03aa324` |
+| Linked semantic result | `e47-semantic-slam-f8c53ba12e9719856195890c7dcd2ba6c114434674159f948064368012e6b3bc` |
+
+The profile is a `0.45 m`, `12 m` radius, `1 s` causal rolling
+`map-gravity-local` costmap. The four states remain separate:
+`GROUND_SUPPORT`, `NONGROUND_OCCUPIED`, `UNKNOWN_REJECTED` and `UNOBSERVED`.
+
+## Full-run result
+
+| Measurement | Result |
+| --- | ---: |
+| Timeline | `4,489 / 4,489` frames |
+| LiDAR available / missing | `3,928 / 561` |
+| Recorded duration / effective rate | `448.623 s / 10.003945 FPS` |
+| Eligible points | `63,646,163` |
+| Ground / non-ground / rejected | `16,579,467 / 47,046,815 / 19,881` |
+| Unaccounted points | `0` |
+| TGS p50 / p95 / p99 / max | `1.190 / 1.694 / 2.08973 / 9.106 ms` |
+| Completion age p50 / p95 / p99 / max | `17.920792 / 29.140775 / 41.428455 / 659.980295 ms` |
+| Capacity drops | `0` |
+| Worker wrapper wall time | `569.401084 s` |
+
+All formal source-paced, point-accounting and capacity gates passed. The
+completion-age maximum is retained as an outlier; the accepted gate is the
+measured `p99 = 41.43 ms`, not the maximum.
+
+## Fail-closed behavior
+
+The `561` timeline frames without a LiDAR sample are not removed, interpolated
+or copied from the preceding frame. Each is emitted as a complete `2,244`-cell
+costmap with all states explicitly `UNOBSERVED`, zero occupied cells and zero
+source points. This preserves chronology without inventing free space.
+
+## Canonical LAB publication and UI acceptance
+
+The immutable result is published on the canonical Mission Core service at
+port `8000`. The LAB reuses the recorded camera timeline and exposes independent
+`SOURCE POINTS`, `LOCAL SLAM`, `TGS COSTMAP`, `SEMANTICS`, `3D` and `PLAN`
+controls. The semantic mask is the exact E47 full-route archive bound to the
+same M4 camera result; it remains diagnostic and does not change TGS states.
+
+The first implementation fetched one large JSON frame at a time. Browser QA
+showed that this was only intermittently exact at `1×`. The published viewer
+therefore uses immutable chunks of `24` spatial frames and prefetches the next
+chunk. A final `1×` browser acceptance sampled twenty consecutive positions
+across chunk boundaries: all twenty displayed the exact TGS frame and none
+showed the loading placeholder. A real missing-LiDAR frame was separately
+accepted with `0 source points`, `2,244 unobserved`, `0 occupied` and no retained
+previous cloud.
+
+A follow-up playback acceptance fixed a UI lifecycle defect that remounted the
+Three.js scene while the next TGS frame crossed the React buffer boundary. The
+scene now retains the last sealed classified frame until the exact next frame
+arrives, so orbit controls and the operator-selected view survive continuous
+playback. Twelve consecutive `250 ms` playback samples kept one mounted 3D
+canvas, a visible bounded `LOCAL SLAM` surface and the synchronized E47 video
+mask. Pausing is no longer required to rotate, pan or zoom the 3D view.
+
+## Sealed evidence
+
+| File | Bytes | SHA-256 |
+| --- | ---: | --- |
+| `costmap-states.npy` | `10,073,444` | `4173f8b6b743755d6d6cbd863c433fe44853974bdd4898002e08574a65783a1a` |
+| `costmap-z-bounds-m.npy` | `80,586,656` | `f8d68b5ca09860ef857a22e15e114bbeabde32492d5c821ee59dfa62ae40373a` |
+| `frames.ndjson` | `1,178,262` | `878e20806d55cdc1fa16692397e11a0e56af760a83c173c8fc73319a39b0e779` |
+| `worker-summary.json` | `1,076` | `487d4be6f3d724bca647aa2e48ba89e7ffd0c78f1bfc5e7e5a01e1db82379dc7` |
+
+## Next gate
+
+Attach this unchanged TGS stage to the realtime world-state graph in shadow
+mode alongside the frozen camera semantic-risk provider. Measure complete-graph
+FPS, end-to-end latency, CPU/GPU/VRAM and queue drops against the accepted
+baseline. Keep navigation authority off. In parallel, review the full camera +
+source cloud + TGS timeline for false occupied carpets, missed compact
+obstacles, vegetation and usable gaps. Only the combination of acceptable
+visual behavior and acceptable integrated-graph regression can promote the
+candidate beyond shadow.

experiments/perception/seal_m49_tgs_fail_closed.py

@@ -0,0 +1,36 @@
+#!/usr/bin/env python3
+"""Publish the verified Worker TGS pack as immutable Mission Core LAB evidence."""
+
+from __future__ import annotations
+
+import argparse
+from pathlib import Path
+
+from k1link.laboratory.m49_tgs_fail_closed import (
+    M49TgsFailClosedError,
+    seal_m49_tgs_fail_closed,
+)
+
+
+def main() -> int:
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--source-root", type=Path, required=True)
+    parser.add_argument("--destination-root", type=Path, required=True)
+    parser.add_argument("--profile", type=Path, required=True)
+    parser.add_argument("--linked-visual-result-id", required=True)
+    arguments = parser.parse_args()
+    try:
+        result = seal_m49_tgs_fail_closed(
+            source_root=arguments.source_root,
+            destination_root=arguments.destination_root,
+            profile_path=arguments.profile,
+            linked_visual_result_id=arguments.linked_visual_result_id,
+        )
+    except (M49TgsFailClosedError, OSError, ValueError) as exc:
+        parser.error(str(exc))
+    print(result.result_id)
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

experiments/perception/worker/Invoke-M49TgsFullShadow.ps1

@@ -0,0 +1,187 @@
+[CmdletBinding()]
+param(
+    [Parameter(Mandatory = $true)]
+    [string]$ReleaseRoot,
+    [Parameter(Mandatory = $true)]
+    [ValidatePattern("^[A-Za-z0-9._-]{1,96}$")]
+    [string]$RunId,
+    [string]$SourcePackPath = "D:\NDC_MISSIONCORE\runtime\derived\e10-lidar-pack-576c994a6c814e2592dd6240ace3902a5db94843312c759a73ba0c9166157d2b\lidar-pack.npz",
+    [string]$OutputRoot = "D:\NDC_MISSIONCORE\runtime\results\m49-tgs-full-shadow"
+)
+
+$ErrorActionPreference = "Stop"
+$ProgressPreference = "SilentlyContinue"
+$TravelImageTag = "ndc/mission-core-m49-t3-travel:20260826"
+$TravelImageId = "sha256:7b412020f4d8392d1d1ed1b33beadc44140f0ea8f781e62dd69796042334300f"
+$ParityImageTag = "ndc-mission-core-m48t-upstream-parity:1.9.4-cu130"
+$ParityImageId = "sha256:ceb13548617e4bd3f619766bfdff00af3fa5160946b367828da6d2233dcdcba0"
+
+function Assert-LastExitCode([string]$Operation) {
+    if ($LASTEXITCODE -ne 0) { throw "$Operation failed with exit code $LASTEXITCODE" }
+}
+
+function Resolve-DDirectory([string]$Path, [string]$Label, [bool]$Create) {
+    if ($Create -and -not (Test-Path -LiteralPath $Path)) {
+        $null = New-Item -ItemType Directory -Path $Path
+    }
+    $item = Get-Item -LiteralPath (Resolve-Path -LiteralPath $Path).Path -Force
+    if (
+        -not $item.PSIsContainer -or
+        ($item.Attributes -band [IO.FileAttributes]::ReparsePoint) -or
+        [IO.Path]::GetPathRoot($item.FullName).TrimEnd("\") -ine "D:"
+    ) { throw "$Label must be a real D: directory" }
+    return $item.FullName
+}
+
+function Resolve-DFile([string]$Path, [string]$Label) {
+    $item = Get-Item -LiteralPath (Resolve-Path -LiteralPath $Path).Path -Force
+    if (
+        $item.PSIsContainer -or
+        ($item.Attributes -band [IO.FileAttributes]::ReparsePoint) -or
+        [IO.Path]::GetPathRoot($item.FullName).TrimEnd("\") -ine "D:"
+    ) { throw "$Label must be a real D: file" }
+    return $item.FullName
+}
+
+function Convert-ToDockerPath([string]$Path) { return ($Path -replace "\\", "/") }
+
+function Get-Container([string]$Name) {
+    $rows = @(((& docker inspect $Name) | ConvertFrom-Json))
+    Assert-LastExitCode "Docker inspection for $Name"
+    if ($rows.Count -ne 1) { throw "Container identity for $Name is not unique" }
+    return $rows[0]
+}
+
+function Assert-Image([string]$Tag, [string]$ExpectedId) {
+    $rows = @(((& docker image inspect $Tag) | ConvertFrom-Json))
+    Assert-LastExitCode "Docker image inspection for $Tag"
+    if ($rows.Count -ne 1 -or [string]$rows[0].Id -cne $ExpectedId) {
+        throw "Pinned image identity changed for $Tag"
+    }
+}
+
+function Remove-ExactContainer([string]$Name) {
+    if (& docker ps -a --format "{{.Names}}" --filter "name=^/$Name$") {
+        & docker rm --force $Name *> $null
+    }
+}
+
+if ($env:COMPUTERNAME -cne "DESKTOP-OPJ8J04") { throw "M49 TGS full shadow is pinned to Worker 006" }
+$release = Resolve-DDirectory $ReleaseRoot "M49 TGS full-shadow release" $false
+$payload = Resolve-DDirectory (Join-Path $release "payload") "M49 TGS full-shadow payload" $false
+$sourcePack = Resolve-DFile $SourcePackPath "RAVNOVES00 source pack"
+$output = Resolve-DDirectory $OutputRoot "M49 TGS full-shadow output root" $true
+$runCandidate = Join-Path $output $RunId
+if (Test-Path -LiteralPath $runCandidate) { throw "M49 TGS full-shadow output already exists" }
+$null = New-Item -ItemType Directory -Path $runCandidate
+$runOutput = Resolve-DDirectory $runCandidate "M49 TGS full-shadow run output" $false
+
+$releaseDocument = Get-Content -LiteralPath (Join-Path $payload "release.json") -Raw | ConvertFrom-Json
+if (
+    $releaseDocument.schema_version -cne "missioncore.m49-tgs-full-shadow-worker-release/v1" -or
+    $releaseDocument.worker_id -cne "worker-006" -or
+    $releaseDocument.candidate_id -cne "travel-tgs-full-shadow"
+) { throw "M49 TGS full-shadow release contract changed" }
+foreach ($property in $releaseDocument.files.PSObject.Properties) {
+    $path = Join-Path $payload $property.Name
+    $actual = (Get-FileHash -Algorithm SHA256 -LiteralPath $path).Hash.ToLowerInvariant()
+    if ($actual -cne [string]$property.Value.sha256) {
+        throw "M49 TGS full-shadow payload digest changed: $($property.Name)"
+    }
+}
+$sourcePackSha = (Get-FileHash -Algorithm SHA256 -LiteralPath $sourcePack).Hash.ToLowerInvariant()
+if ($sourcePackSha -cne "0685d24219d8236caf8b7f1685e93f6d6b59e7fd015a768d88a92bbe8b154944") {
+    throw "RAVNOVES00 source pack digest changed"
+}
+$os = Get-CimInstance Win32_OperatingSystem
+$freeMemoryGiB = [double]$os.FreePhysicalMemory / 1MB
+if ($freeMemoryGiB -lt 24.0) {
+    throw ("M49 TGS full shadow requires 24 GiB free memory; observed {0:N2} GiB" -f $freeMemoryGiB)
+}
+$tritonBefore = Get-Container "ndc-mission-core-triton"
+if (-not $tritonBefore.State.Running -or $tritonBefore.State.Health.Status -cne "healthy") {
+    throw "Canonical Mission Core Triton must remain healthy during M49 TGS full shadow"
+}
+Assert-Image $TravelImageTag $TravelImageId
+Assert-Image $ParityImageTag $ParityImageId
+
+$prepareName = "ndc-mission-core-m49-tgs-full-prepare-$RunId"
+$runName = "ndc-mission-core-m49-tgs-full-run-$RunId"
+$analyzeName = "ndc-mission-core-m49-tgs-full-analyze-$RunId"
+foreach ($name in @($prepareName, $runName, $analyzeName)) {
+    if (& docker ps -a --format "{{.Names}}" --filter "name=^/$name$") {
+        throw "M49 TGS full-shadow container name already exists: $name"
+    }
+}
+
+$started = [DateTimeOffset]::UtcNow
+try {
+    & docker run --rm --name $prepareName --network none --cpus 8 --memory 16g `
+        --entrypoint python3 `
+        --volume ((Convert-ToDockerPath $sourcePack) + ":/source/lidar-pack.npz:ro") `
+        --volume ((Convert-ToDockerPath $payload) + ":/release:ro") `
+        --volume ((Convert-ToDockerPath $runOutput) + ":/tgs") `
+        $ParityImageTag /release/prepare_tgs_full_shadow_inputs.py `
+        --source-pack /source/lidar-pack.npz `
+        --config /release/m49-tgs-full-shadow-v1.json `
+        --output-root /tgs/inputs
+    Assert-LastExitCode "M49 TGS full-shadow input preparation"
+
+    & docker run --rm --name $runName --network none --cpus 16 --memory 24g `
+        --entrypoint /bin/bash `
+        --volume ((Convert-ToDockerPath $payload) + ":/release:ro") `
+        --volume ((Convert-ToDockerPath $runOutput) + ":/tgs") `
+        $TravelImageTag /release/run_tgs_full_shadow.sh
+    Assert-LastExitCode "M49 source-paced TGS full-shadow run"
+
+    & docker run --rm --name $analyzeName --network none --cpus 8 --memory 16g `
+        --entrypoint python3 `
+        --volume ((Convert-ToDockerPath $payload) + ":/release:ro") `
+        --volume ((Convert-ToDockerPath $runOutput) + ":/tgs") `
+        $ParityImageTag /release/build_tgs_full_shadow_evidence.py `
+        --run-root /tgs `
+        --config /release/m49-tgs-full-shadow-v1.json `
+        --output-root /tgs/evidence
+    Assert-LastExitCode "M49 TGS full-shadow evidence analysis"
+} finally {
+    foreach ($name in @($prepareName, $runName, $analyzeName)) { Remove-ExactContainer $name }
+}
+$completed = [DateTimeOffset]::UtcNow
+$resultPath = Join-Path $runOutput "evidence\result.json"
+if (-not (Test-Path -LiteralPath $resultPath -PathType Leaf)) { throw "M49 TGS full-shadow result is missing" }
+$result = Get-Content -LiteralPath $resultPath -Raw | ConvertFrom-Json
+if (
+    $result.timeline.frame_count -ne 4489 -or
+    $result.timeline.available_lidar_frame_count -ne 3928 -or
+    $result.timeline.missing_lidar_frame_count -ne 561 -or
+    $result.point_accounting.unaccounted -ne 0
+) { throw "M49 TGS full-shadow structural acceptance failed" }
+$tritonAfter = Get-Container "ndc-mission-core-triton"
+if (
+    -not $tritonAfter.State.Running -or
+    $tritonAfter.State.Health.Status -cne "healthy" -or
+    [string]$tritonAfter.Id -cne [string]$tritonBefore.Id
+) { throw "Canonical Mission Core Triton changed during M49 TGS full shadow" }
+$summary = [ordered]@{
+    schema_version = "missioncore.m49-tgs-full-shadow-worker-summary/v1"
+    worker_id = "worker-006"
+    run_id = $RunId
+    code_revision = [string]$releaseDocument.code_revision
+    source_pack_sha256 = $sourcePackSha
+    travel_image_id = $TravelImageId
+    parity_image_id = $ParityImageId
+    started_utc = $started.ToString("o")
+    wall_seconds = [math]::Round(($completed - $started).TotalSeconds, 6)
+    free_memory_gib_before = [math]::Round($freeMemoryGiB, 6)
+    canonical_triton_id = [string]$tritonAfter.Id
+    canonical_triton_health = [string]$tritonAfter.State.Health.Status
+    result_status = [string]$result.status
+    all_timeline_frames_accounted = $true
+    all_eligible_points_accounted = $true
+    aos_used = $false
+    gpu_requested = $false
+    integrated_graph_performance_accepted = $false
+    navigation_or_actuation_allowed = $false
+}
+$summary | ConvertTo-Json -Depth 3 | Set-Content -LiteralPath (Join-Path $runOutput "worker-summary.json") -Encoding utf8
+$summary | ConvertTo-Json -Depth 3

experiments/perception/worker/Invoke-M49TgsFullShadowAsInteractiveUser.ps1

@@ -0,0 +1,45 @@
+[CmdletBinding()]
+param(
+    [Parameter(Mandatory = $true)]
+    [string]$ReleaseRoot,
+    [Parameter(Mandatory = $true)]
+    [ValidatePattern("^[A-Za-z0-9._-]{1,96}$")]
+    [string]$RunId
+)
+
+$ErrorActionPreference = "Stop"
+$taskName = "MissionCore-M49TgsFullShadow"
+$release = (Resolve-Path -LiteralPath $ReleaseRoot).Path
+$runner = Join-Path $release "payload\Invoke-M49TgsFullShadow.ps1"
+if (-not (Test-Path -LiteralPath $runner -PathType Leaf)) { throw "M49 TGS full-shadow runner is missing" }
+$existing = Get-ScheduledTask -TaskName $taskName -ErrorAction SilentlyContinue
+if ($existing -and $existing.State -eq "Running") { throw "$taskName is already running" }
+$powerShell = "$env:SystemRoot\System32\WindowsPowerShell\v1.0\powershell.exe"
+$arguments = @(
+    "-NoLogo", "-NoProfile", "-NonInteractive", "-ExecutionPolicy", "Bypass",
+    "-File", "`"$runner`"", "-ReleaseRoot", "`"$release`"", "-RunId", "`"$RunId`""
+) -join " "
+$userId = [System.Security.Principal.WindowsIdentity]::GetCurrent().Name
+$action = New-ScheduledTaskAction -Execute $powerShell -Argument $arguments -WorkingDirectory $release
+$principal = New-ScheduledTaskPrincipal -UserId $userId -LogonType Interactive -RunLevel Limited
+$trigger = New-ScheduledTaskTrigger -Once -At ((Get-Date).AddMinutes(30))
+$settings = New-ScheduledTaskSettingsSet `
+    -AllowStartIfOnBatteries `
+    -DontStopIfGoingOnBatteries `
+    -StartWhenAvailable `
+    -ExecutionTimeLimit ([TimeSpan]::FromHours(2))
+Register-ScheduledTask `
+    -TaskName $taskName `
+    -Action $action `
+    -Principal $principal `
+    -Trigger $trigger `
+    -Settings $settings `
+    -Description "One-shot CPU-only source-paced TGS full shadow." `
+    -Force | Out-Null
+Start-ScheduledTask -TaskName $taskName
+[pscustomobject]@{
+    task_name = $taskName
+    run_id = $RunId
+    release_root = $release
+    state = (Get-ScheduledTask -TaskName $taskName).State.ToString()
+} | ConvertTo-Json -Compress

experiments/perception/worker/m49_t3_travel/build_tgs_full_shadow_evidence.py

@@ -0,0 +1,332 @@
+#!/usr/bin/env python3
+"""Build mmap-friendly evidence for the complete source-paced TGS shadow."""
+
+from __future__ import annotations
+
+import argparse
+import csv
+import hashlib
+import json
+import math
+from pathlib import Path
+
+import numpy as np
+
+from build_tgs_fail_closed_evidence import costmap_grid
+
+
+class FullShadowError(RuntimeError):
+    """The complete TGS shadow or its fail-closed contract is invalid."""
+
+
+def sha256_file(path: Path) -> str:
+    digest = hashlib.sha256()
+    with path.open("rb") as stream:
+        for chunk in iter(lambda: stream.read(1024 * 1024), b""):
+            digest.update(chunk)
+    return digest.hexdigest()
+
+
+def load_xyzi(path: Path) -> np.ndarray:
+    if path.is_symlink() or not path.is_file():
+        raise FullShadowError(f"sealed input is unavailable: {path.name}")
+    values = np.fromfile(path, dtype=np.float32)
+    if values.size % 4:
+        raise FullShadowError(f"sealed XYZI shape changed: {path.name}")
+    result = values.reshape(-1, 4)
+    if not np.isfinite(result).all():
+        raise FullShadowError(f"sealed XYZI is non-finite: {path.name}")
+    return result
+
+
+def classify_exact_input(
+    native: np.ndarray, ground: np.ndarray, nonground: np.ndarray
+) -> tuple[np.ndarray, np.ndarray]:
+    ranges = np.linalg.norm(native[:, :2].astype(np.float64), axis=1)
+    points = np.ascontiguousarray(native[(ranges > 1.0) & (ranges < 80.0), :3])
+    output = np.ascontiguousarray(np.concatenate((ground[:, :3], nonground[:, :3]), axis=0))
+    output_states = np.concatenate(
+        (np.ones(ground.shape[0], dtype=np.uint8), np.full(nonground.shape[0], 2, dtype=np.uint8))
+    )
+    key_dtype = np.dtype((np.void, 12))
+    input_keys = points.view(key_dtype).reshape(-1)
+    output_keys = output.view(key_dtype).reshape(-1)
+    input_order = np.argsort(input_keys, kind="stable")
+    output_order = np.argsort(output_keys, kind="stable")
+    sorted_input = input_keys[input_order]
+    sorted_output = output_keys[output_order]
+    positions = np.searchsorted(sorted_input, sorted_output, side="left")
+    if sorted_output.size:
+        group_starts = np.r_[0, np.flatnonzero(sorted_output[1:] != sorted_output[:-1]) + 1]
+        group_lengths = np.diff(np.r_[group_starts, sorted_output.size])
+        occurrence = np.arange(sorted_output.size) - np.repeat(group_starts, group_lengths)
+        targets = positions + occurrence
+        if (
+            np.any(targets >= sorted_input.size)
+            or np.any(sorted_input[targets] != sorted_output)
+            or np.unique(targets).size != targets.size
+        ):
+            raise FullShadowError("TGS output is not a multiset subset of its exact input")
+    else:
+        targets = np.empty(0, dtype=np.int64)
+    sorted_states = np.full(points.shape[0], 3, dtype=np.uint8)
+    sorted_states[targets] = output_states[output_order]
+    states = np.empty_like(sorted_states)
+    states[input_order] = sorted_states
+    return points, states
+
+
+def rasterize(
+    points: np.ndarray,
+    states: np.ndarray,
+    grid: np.ndarray,
+    cell_size_m: float,
+) -> tuple[np.ndarray, np.ndarray]:
+    minimum_ix = int(np.min(grid[:, 0]))
+    maximum_ix = int(np.max(grid[:, 0]))
+    minimum_iy = int(np.min(grid[:, 1]))
+    maximum_iy = int(np.max(grid[:, 1]))
+    lookup = np.full(
+        (maximum_ix - minimum_ix + 1, maximum_iy - minimum_iy + 1), -1, dtype=np.int32
+    )
+    lookup[
+        grid[:, 0].astype(np.int32) - minimum_ix,
+        grid[:, 1].astype(np.int32) - minimum_iy,
+    ] = np.arange(grid.shape[0], dtype=np.int32)
+    cell_xy = np.floor(points[:, :2] / cell_size_m).astype(np.int32)
+    inside = (
+        (cell_xy[:, 0] >= minimum_ix)
+        & (cell_xy[:, 0] <= maximum_ix)
+        & (cell_xy[:, 1] >= minimum_iy)
+        & (cell_xy[:, 1] <= maximum_iy)
+    )
+    point_indices = np.flatnonzero(inside)
+    cell_indices = lookup[
+        cell_xy[inside, 0] - minimum_ix, cell_xy[inside, 1] - minimum_iy
+    ]
+    valid = cell_indices >= 0
+    point_indices = point_indices[valid]
+    cell_indices = cell_indices[valid]
+    cell_states = np.zeros(grid.shape[0], dtype=np.uint8)
+    selected_states = states[point_indices]
+    ground_cells = np.zeros(grid.shape[0], dtype=np.uint8)
+    rejected_cells = np.zeros(grid.shape[0], dtype=np.uint8)
+    nonground_cells = np.zeros(grid.shape[0], dtype=np.uint8)
+    np.maximum.at(ground_cells, cell_indices, (selected_states == 1).astype(np.uint8))
+    np.maximum.at(rejected_cells, cell_indices, (selected_states == 3).astype(np.uint8))
+    np.maximum.at(nonground_cells, cell_indices, (selected_states == 2).astype(np.uint8))
+    cell_states[ground_cells > 0] = 1
+    cell_states[rejected_cells > 0] = 3
+    cell_states[nonground_cells > 0] = 2
+    minimum_z = np.full(grid.shape[0], np.inf, dtype=np.float32)
+    maximum_z = np.full(grid.shape[0], -np.inf, dtype=np.float32)
+    np.minimum.at(minimum_z, cell_indices, points[point_indices, 2])
+    np.maximum.at(maximum_z, cell_indices, points[point_indices, 2])
+    z_bounds = np.column_stack((minimum_z, maximum_z)).astype(np.float32, copy=False)
+    z_bounds[~np.isfinite(z_bounds)] = np.nan
+    return cell_states, z_bounds
+
+
+def percentile(values: np.ndarray, value: float) -> float:
+    return float(np.percentile(values.astype(np.float64), value)) if values.size else 0.0
+
+
+def build(run_root: Path, config_path: Path, output_root: Path) -> dict[str, object]:
+    if output_root.exists():
+        raise FullShadowError("full-shadow evidence output already exists")
+    config = json.loads(config_path.read_text(encoding="utf-8"))
+    if (
+        config.get("schema_version") != "missioncore.m49-tgs-full-shadow-profile/v1"
+        or config.get("invariants", {}).get("aos_allowed") is not False
+        or config.get("invariants", {}).get("gpu_allowed") is not False
+        or config.get("invariants", {}).get("missing_lidar_means_unobserved") is not True
+    ):
+        raise FullShadowError("full-shadow profile changed")
+    manifest_path = run_root / "inputs" / "input-manifest.json"
+    manifest = json.loads(manifest_path.read_text(encoding="utf-8"))
+    records = manifest.get("records", [])
+    if (
+        manifest.get("schema_version") != "missioncore.m49-tgs-full-shadow-input/v1"
+        or manifest.get("source_pack_sha256") != config["source"]["source_pack_sha256"]
+        or manifest.get("config_sha256") != sha256_file(config_path)
+        or manifest.get("future_frames_used") is not False
+        or len(records) != 4489
+        or sum(bool(row["sample_available"]) for row in records) != 3928
+    ):
+        raise FullShadowError("full-shadow input manifest changed")
+    with (run_root / "tgs-full-timing.tsv").open("r", encoding="utf-8", newline="") as stream:
+        timing_rows = list(csv.DictReader(stream, delimiter="\t"))
+    if len(timing_rows) != 4489:
+        raise FullShadowError("full-shadow timing frame accounting changed")
+
+    cell_size = float(config["costmap"]["cell_size_m"])
+    radius = float(config["costmap"]["radius_m"])
+    grid = costmap_grid(radius, cell_size)
+    output_root.mkdir(parents=True)
+    np.save(output_root / "costmap-cell-indices-xy.npy", grid[:, :2].astype(np.int32))
+    np.save(output_root / "costmap-cell-centers-xy-m.npy", grid[:, 2:].astype(np.float32))
+    states_out = np.lib.format.open_memmap(
+        output_root / "costmap-states.npy", mode="w+", dtype=np.uint8, shape=(4489, grid.shape[0])
+    )
+    z_out = np.lib.format.open_memmap(
+        output_root / "costmap-z-bounds-m.npy",
+        mode="w+",
+        dtype=np.float32,
+        shape=(4489, grid.shape[0], 2),
+    )
+    z_out[:] = np.nan
+    summaries: list[dict[str, object]] = []
+    eligible_total = ground_total = nonground_total = rejected_total = 0
+    available_seen = 0
+    for frame_index, (record, timing) in enumerate(zip(records, timing_rows, strict=True)):
+        if int(timing["timeline_frame_index"]) != frame_index:
+            raise FullShadowError("full-shadow timing order changed")
+        available = bool(record["sample_available"])
+        if not available:
+            if int(timing["sample_available"]) != 0:
+                raise FullShadowError("missing LiDAR frame was processed")
+            states_out[frame_index] = 0
+            summaries.append(
+                {
+                    "timeline_frame_index": frame_index,
+                    "source_frame_index": int(record["source_frame_index"]),
+                    "session_seconds": float(record["session_seconds"]),
+                    "sample_available": False,
+                    "eligible_point_count": 0,
+                    "ground_point_count": 0,
+                    "nonground_point_count": 0,
+                    "rejected_point_count": 0,
+                    "occupied_cell_count": 0,
+                }
+            )
+            continue
+        available_seen += 1
+        native_path = run_root / "inputs" / str(record["relative_path"])
+        if sha256_file(native_path) != record["sha256"]:
+            raise FullShadowError("sealed gravity-aligned full-shadow input changed")
+        output = run_root / "outputs" / "causal_rolling_1s"
+        ground = load_xyzi(output / f"{frame_index}_ground.bin")
+        nonground = load_xyzi(output / f"{frame_index}_nonground.bin")
+        points, point_states = classify_exact_input(load_xyzi(native_path), ground, nonground)
+        cell_states, z_bounds = rasterize(points, point_states, grid, cell_size)
+        states_out[frame_index] = cell_states
+        z_out[frame_index] = z_bounds
+        ground_count = int(np.count_nonzero(point_states == 1))
+        nonground_count = int(np.count_nonzero(point_states == 2))
+        rejected_count = int(np.count_nonzero(point_states == 3))
+        eligible_total += int(points.shape[0])
+        ground_total += ground_count
+        nonground_total += nonground_count
+        rejected_total += rejected_count
+        summaries.append(
+            {
+                "timeline_frame_index": frame_index,
+                "source_frame_index": int(record["source_frame_index"]),
+                "session_seconds": float(record["session_seconds"]),
+                "sample_available": True,
+                "eligible_point_count": int(points.shape[0]),
+                "ground_point_count": ground_count,
+                "nonground_point_count": nonground_count,
+                "rejected_point_count": rejected_count,
+                "occupied_cell_count": int(np.count_nonzero(cell_states == 2)),
+            }
+        )
+    states_out.flush()
+    z_out.flush()
+    if available_seen != 3928 or eligible_total != ground_total + nonground_total + rejected_total:
+        raise FullShadowError("full-shadow eligible point accounting failed")
+    frames_path = output_root / "frames.ndjson"
+    frames_path.write_text(
+        "".join(json.dumps(row, sort_keys=True) + "\n" for row in summaries), encoding="utf-8"
+    )
+
+    available_timings = [row for row in timing_rows if int(row["sample_available"]) == 1]
+    tgs_ms = np.asarray([float(row["tgs_ms"]) for row in available_timings])
+    completion_ms = np.asarray([float(row["completion_age_ms"]) for row in timing_rows])
+    capacity_drops = sum(int(row["capacity_drop"]) for row in timing_rows)
+    duration = float(records[-1]["session_seconds"]) - float(records[0]["session_seconds"])
+    effective_fps = (len(records) - 1) / duration
+    thresholds = config["acceptance"]
+    acceptance = {
+        "minimum_effective_timeline_fps": effective_fps >= float(thresholds["minimum_effective_timeline_fps"]),
+        "candidate_stage_p95_ms": percentile(tgs_ms, 95) <= float(thresholds["candidate_stage_p95_ms_max"]),
+        "candidate_stage_p99_ms": percentile(tgs_ms, 99) <= float(thresholds["candidate_stage_p99_ms_max"]),
+        "completion_age_p99_ms": percentile(completion_ms, 99) <= float(thresholds["completion_age_p99_ms_max"]),
+        "capacity_drop_count": capacity_drops <= int(thresholds["capacity_drop_count_max"]),
+        "all_frames_accounted": len(summaries) == 4489 and available_seen == 3928,
+        "all_eligible_points_accounted": eligible_total == ground_total + nonground_total + rejected_total,
+    }
+    files = {}
+    for path in sorted(output_root.iterdir()):
+        if path.is_file() and path.name != "result.json":
+            files[path.name] = {"bytes": path.stat().st_size, "sha256": sha256_file(path)}
+    result = {
+        "schema_version": "missioncore.m49-tgs-full-shadow-result/v1",
+        "status": "passed" if all(acceptance.values()) else "failed",
+        "config_sha256": sha256_file(config_path),
+        "source_pack_sha256": manifest["source_pack_sha256"],
+        "input_manifest_sha256": sha256_file(manifest_path),
+        "timeline": {
+            "frame_count": 4489,
+            "available_lidar_frame_count": 3928,
+            "missing_lidar_frame_count": 561,
+            "duration_seconds": duration,
+            "effective_fps": effective_fps,
+        },
+        "costmap": {
+            "coordinate_frame": "map-gravity-local",
+            "cell_size_m": cell_size,
+            "radius_m": radius,
+            "cell_count": int(grid.shape[0]),
+        },
+        "point_accounting": {
+            "eligible": eligible_total,
+            "ground": ground_total,
+            "nonground": nonground_total,
+            "rejected": rejected_total,
+            "unaccounted": eligible_total - ground_total - nonground_total - rejected_total,
+        },
+        "performance": {
+            "candidate_tgs_ms": {
+                "p50": percentile(tgs_ms, 50),
+                "p95": percentile(tgs_ms, 95),
+                "p99": percentile(tgs_ms, 99),
+                "max": float(np.max(tgs_ms)),
+            },
+            "completion_age_ms": {
+                "p50": percentile(completion_ms, 50),
+                "p95": percentile(completion_ms, 95),
+                "p99": percentile(completion_ms, 99),
+                "max": float(np.max(completion_ms)),
+            },
+            "capacity_drop_count": capacity_drops,
+        },
+        "acceptance": acceptance,
+        "files": files,
+        "authority": {
+            "visual_quality_accepted": False,
+            "traversability_accepted": False,
+            "realtime_accepted": bool(all(acceptance.values())),
+            "integrated_graph_performance_accepted": False,
+            "navigation_or_actuation_allowed": False,
+        },
+    }
+    (output_root / "result.json").write_text(
+        json.dumps(result, indent=2, sort_keys=True) + "\n", encoding="utf-8"
+    )
+    return result
+
+
+def main() -> int:
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--run-root", type=Path, required=True)
+    parser.add_argument("--config", type=Path, required=True)
+    parser.add_argument("--output-root", type=Path, required=True)
+    arguments = parser.parse_args()
+    result = build(arguments.run_root, arguments.config, arguments.output_root)
+    print(json.dumps({"status": result["status"], **result["performance"]}, sort_keys=True))
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

experiments/perception/worker/m49_t3_travel/prepare_tgs_full_shadow_inputs.py

@@ -0,0 +1,193 @@
+#!/usr/bin/env python3
+"""Prepare every available RAVNOVES00 frame for the source-paced TGS shadow."""
+
+from __future__ import annotations
+
+import argparse
+import hashlib
+import json
+from pathlib import Path
+
+import numpy as np
+
+from prepare_tgs_fail_closed_inputs import (
+    EXPECTED_ARRAYS,
+    SOURCE_PACK_SHA256,
+    TgsInputError,
+    _frame_points,
+    _validate_source,
+    gravity_local_xyzi,
+    sha256_file,
+)
+
+FULL_SCHEMA = "missioncore.m49-tgs-full-shadow-profile/v1"
+INPUT_SCHEMA = "missioncore.m49-tgs-full-shadow-input/v1"
+TIMELINE_FRAME_COUNT = 4_489
+AVAILABLE_LIDAR_FRAME_COUNT = 3_928
+
+
+def _bytes_sha256(content: bytes) -> str:
+    return hashlib.sha256(content).hexdigest()
+
+
+def prepare(source_pack: Path, config_path: Path, output_root: Path) -> dict[str, object]:
+    if output_root.exists():
+        raise TgsInputError("TGS full-shadow input output root already exists")
+    if sha256_file(source_pack) != SOURCE_PACK_SHA256:
+        raise TgsInputError("RAVNOVES00 lidar-pack digest changed")
+    config = json.loads(config_path.read_text(encoding="utf-8"))
+    source = config.get("source", {})
+    profile = config.get("profile", {})
+    invariants = config.get("invariants", {})
+    if (
+        config.get("schema_version") != FULL_SCHEMA
+        or source.get("source_pack_sha256") != SOURCE_PACK_SHA256
+        or source.get("expected_timeline_frames") != TIMELINE_FRAME_COUNT
+        or source.get("expected_available_lidar_frames") != AVAILABLE_LIDAR_FRAME_COUNT
+        or source.get("input_coordinate_frame")
+        != "map-gravity-local-translation-only"
+        or profile.get("id") != "causal_rolling_1s"
+        or profile.get("missing_lidar_policy") != "all-cells-unobserved"
+        or invariants.get("lidar_orientation_applied_to_tgs_input") is not False
+        or invariants.get("future_frames_used") is not False
+        or invariants.get("missing_lidar_means_unobserved") is not True
+    ):
+        raise TgsInputError("TGS full-shadow profile changed")
+    required = EXPECTED_ARRAYS | {"source_frame_indices", "pose_quaternions_map_from_lidar"}
+    with np.load(source_pack, allow_pickle=False) as archive:
+        if not required.issubset(archive.files):
+            raise TgsInputError("RAVNOVES00 lidar-pack members changed")
+        arrays = {name: archive[name] for name in required}
+    _validate_source(arrays)
+    if (
+        arrays["source_frame_indices"].shape != (TIMELINE_FRAME_COUNT,)
+        or arrays["source_frame_indices"].dtype != np.int64
+        or arrays["pose_quaternions_map_from_lidar"].shape != (TIMELINE_FRAME_COUNT, 4)
+        or arrays["pose_quaternions_map_from_lidar"].dtype != np.float64
+        or int(np.count_nonzero(arrays["sample_available"])) != AVAILABLE_LIDAR_FRAME_COUNT
+    ):
+        raise TgsInputError("RAVNOVES00 full timeline contract changed")
+
+    sequence_root = output_root / "profiles" / "causal_rolling_1s" / "velodyne"
+    sequence_root.mkdir(parents=True)
+    seconds = arrays["session_seconds"]
+    availability = arrays["sample_available"]
+    positions = arrays["pose_positions_map"]
+    history_seconds = float(profile["history_seconds"])
+    local_radius_m = float(profile["local_radius_m"])
+    records: list[dict[str, object]] = []
+    schedule_rows = [
+        "timeline_frame_index\tsource_frame_index\tsession_seconds\tavailable_slot\tpoint_count"
+    ]
+    available_slot = 0
+    for frame_index in range(TIMELINE_FRAME_COUNT):
+        base = {
+            "timeline_frame_index": frame_index,
+            "frame_index": int(arrays["frame_indices"][frame_index]),
+            "source_frame_index": int(arrays["source_frame_indices"][frame_index]),
+            "session_seconds": float(seconds[frame_index]),
+            "position_map_m": [float(value) for value in positions[frame_index]],
+            "sample_available": bool(availability[frame_index]),
+        }
+        if not bool(availability[frame_index]):
+            records.append(
+                {
+                    **base,
+                    "available_slot": None,
+                    "point_count": 0,
+                    "contributing_frame_indices": [],
+                    "relative_path": None,
+                    "bytes": 0,
+                    "sha256": None,
+                }
+            )
+            schedule_rows.append(
+                f"{frame_index}\t{base['source_frame_index']}\t{seconds[frame_index]:.9f}\t-1\t0"
+            )
+            continue
+
+        start = int(np.searchsorted(seconds, seconds[frame_index] - history_seconds, side="left"))
+        contributors = tuple(
+            index for index in range(start, frame_index + 1) if bool(availability[index])
+        )
+        if not contributors or contributors[-1] != frame_index:
+            raise TgsInputError("causal full-shadow profile does not contain its current frame")
+        points_map = np.concatenate(
+            [_frame_points(arrays, index) for index in contributors], axis=0
+        )
+        relative_xy = points_map[:, :2].astype(np.float64) - positions[frame_index, :2]
+        points_map = points_map[np.linalg.norm(relative_xy, axis=1) <= local_radius_m]
+        native = gravity_local_xyzi(points_map, positions[frame_index])
+        if native.shape[0] == 0:
+            raise TgsInputError("available full-shadow frame produced an empty cloud")
+        content = np.ascontiguousarray(native).tobytes()
+        target = sequence_root / f"{available_slot:06d}.bin"
+        target.write_bytes(content)
+        records.append(
+            {
+                **base,
+                "available_slot": available_slot,
+                "point_count": int(native.shape[0]),
+                "contributing_frame_indices": list(contributors),
+                "relative_path": target.relative_to(output_root).as_posix(),
+                "bytes": len(content),
+                "sha256": _bytes_sha256(content),
+            }
+        )
+        schedule_rows.append(
+            f"{frame_index}\t{base['source_frame_index']}\t{seconds[frame_index]:.9f}"
+            f"\t{available_slot}\t{native.shape[0]}"
+        )
+        available_slot += 1
+
+    if available_slot != AVAILABLE_LIDAR_FRAME_COUNT or len(records) != TIMELINE_FRAME_COUNT:
+        raise TgsInputError("TGS full-shadow frame accounting changed")
+    schedule_path = output_root / "schedule.tsv"
+    schedule_path.write_text("\n".join(schedule_rows) + "\n", encoding="utf-8")
+    manifest = {
+        "schema_version": INPUT_SCHEMA,
+        "source_pack_sha256": SOURCE_PACK_SHA256,
+        "config_sha256": sha256_file(config_path),
+        "coordinate_frame": "map-gravity-local",
+        "transform": "translation-only-preserve-map-gravity-axis",
+        "intensity_policy": "zero-filled-algorithm-compatibility-only",
+        "future_frames_used": False,
+        "timeline_frame_count": TIMELINE_FRAME_COUNT,
+        "available_lidar_frame_count": AVAILABLE_LIDAR_FRAME_COUNT,
+        "missing_lidar_frame_count": TIMELINE_FRAME_COUNT - AVAILABLE_LIDAR_FRAME_COUNT,
+        "schedule": {
+            "path": "schedule.tsv",
+            "bytes": schedule_path.stat().st_size,
+            "sha256": sha256_file(schedule_path),
+        },
+        "records": records,
+    }
+    manifest_path = output_root / "input-manifest.json"
+    manifest_path.write_text(
+        json.dumps(manifest, indent=2, sort_keys=True) + "\n", encoding="utf-8"
+    )
+    return manifest
+
+
+def main() -> int:
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--source-pack", type=Path, required=True)
+    parser.add_argument("--config", type=Path, required=True)
+    parser.add_argument("--output-root", type=Path, required=True)
+    arguments = parser.parse_args()
+    manifest = prepare(arguments.source_pack, arguments.config, arguments.output_root)
+    print(
+        json.dumps(
+            {
+                "ok": True,
+                "timeline_frames": manifest["timeline_frame_count"],
+                "available_lidar_frames": manifest["available_lidar_frame_count"],
+            },
+            sort_keys=True,
+        )
+    )
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

experiments/perception/worker/m49_t3_travel/run_tgs_full_shadow.cpp

@@ -0,0 +1,181 @@
+#include <chrono>
+#include <filesystem>
+#include <fstream>
+#include <iomanip>
+#include <iostream>
+#include <memory>
+#include <sstream>
+#include <stdexcept>
+#include <string>
+#include <thread>
+#include <vector>
+
+#include "travel/kitti_loader.hpp"
+#include "travel/point_types.hpp"
+#include "travel/tgs.hpp"
+
+namespace {
+
+using Clock = std::chrono::steady_clock;
+
+struct ScheduleRow {
+    std::size_t timeline_frame_index;
+    long long source_frame_index;
+    double session_seconds;
+    long long available_slot;
+    std::size_t point_count;
+};
+
+std::vector<ScheduleRow> readSchedule(const std::string& path) {
+    std::ifstream input(path);
+    if (!input) {
+        throw std::runtime_error("cannot open full-shadow schedule");
+    }
+    std::string line;
+    std::getline(input, line);
+    if (line != "timeline_frame_index\tsource_frame_index\tsession_seconds\tavailable_slot\tpoint_count") {
+        throw std::runtime_error("full-shadow schedule header changed");
+    }
+    std::vector<ScheduleRow> rows;
+    while (std::getline(input, line)) {
+        if (line.empty()) {
+            continue;
+        }
+        std::istringstream stream(line);
+        ScheduleRow row{};
+        if (!(stream >> row.timeline_frame_index >> row.source_frame_index >> row.session_seconds
+              >> row.available_slot >> row.point_count)) {
+            throw std::runtime_error("invalid full-shadow schedule row");
+        }
+        if (row.timeline_frame_index != rows.size()) {
+            throw std::runtime_error("full-shadow schedule is not contiguous");
+        }
+        rows.push_back(row);
+    }
+    if (rows.size() != 4489) {
+        throw std::runtime_error("full-shadow timeline frame count changed");
+    }
+    return rows;
+}
+
+void writeXYZI(const std::string& path, const travel::PointCloud<PointXYZILID>& cloud) {
+    std::ofstream output(path, std::ios::binary);
+    if (!output) {
+        throw std::runtime_error("cannot open full-shadow TGS output");
+    }
+    for (const auto& point : cloud.points) {
+        const float row[4] = {point.x, point.y, point.z, point.intensity};
+        output.write(reinterpret_cast<const char*>(row), sizeof(row));
+    }
+    if (!output) {
+        throw std::runtime_error("cannot write full-shadow TGS output");
+    }
+}
+
+double milliseconds(Clock::duration duration) {
+    return std::chrono::duration<double, std::milli>(duration).count();
+}
+
+}  // namespace
+
+int main(int argc, char** argv) {
+    if (argc != 5) {
+        std::cerr << "Usage: run_tgs_full_shadow <sequence_dir> <schedule.tsv> <output_dir> <timing.tsv>\n";
+        return 1;
+    }
+    try {
+        const std::string sequence_dir = argv[1];
+        const std::string schedule_path = argv[2];
+        const std::string output_dir = argv[3];
+        const std::string timing_path = argv[4];
+        const auto schedule = readSchedule(schedule_path);
+        KittiLoader loader(sequence_dir);
+        if (loader.size() != 3928) {
+            throw std::runtime_error("full-shadow available LiDAR frame count changed");
+        }
+        std::filesystem::create_directories(output_dir);
+        std::ofstream timing(timing_path);
+        if (!timing) {
+            throw std::runtime_error("cannot open full-shadow timing output");
+        }
+        timing << "timeline_frame_index\tsource_frame_index\tsession_seconds\tsample_available"
+               << "\tavailable_slot\tinput_points\tground_points\tnonground_points"
+               << "\ttgs_ms\tstage_wall_ms\tqueue_delay_ms\tcompletion_age_ms\tcapacity_drop\n";
+        timing << std::fixed << std::setprecision(6);
+
+        const double first_source_seconds = schedule.front().session_seconds;
+        const auto run_started = Clock::now();
+        std::size_t expected_slot = 0;
+        for (const auto& row : schedule) {
+            const auto target = run_started + std::chrono::duration_cast<Clock::duration>(
+                std::chrono::duration<double>(row.session_seconds - first_source_seconds));
+            const auto before_wait = Clock::now();
+            if (before_wait < target) {
+                std::this_thread::sleep_until(target);
+            }
+            const auto stage_started = Clock::now();
+            const double queue_delay_ms = std::max(0.0, milliseconds(stage_started - target));
+            std::size_t input_points = 0;
+            std::size_t ground_points = 0;
+            std::size_t nonground_points = 0;
+            double tgs_seconds = 0.0;
+
+            if (row.available_slot >= 0) {
+                if (static_cast<std::size_t>(row.available_slot) != expected_slot) {
+                    throw std::runtime_error("full-shadow available slot order changed");
+                }
+                auto input_xyzi = loader.cloud(expected_slot);
+                if (!input_xyzi || input_xyzi->size() != row.point_count) {
+                    throw std::runtime_error("full-shadow input point count changed");
+                }
+                auto input = std::make_shared<travel::PointCloud<PointXYZILID>>();
+                input->reserve(input_xyzi->size());
+                for (const auto& point : input_xyzi->points) {
+                    PointXYZILID value{};
+                    value.x = point.x;
+                    value.y = point.y;
+                    value.z = point.z;
+                    value.intensity = point.intensity;
+                    value.label = 0;
+                    value.id = 0;
+                    input->emplace_back(value);
+                }
+                travel::TravelGroundSeg<PointXYZILID> tgs;
+                tgs.setParams(
+                    80.0, 1.0, 8.0, 3, 5, 10, 0.5, 0.125, 0.3, 0.940,
+                    200.0, 0.03, 0.1, 1.0, true, false);
+                travel::PointCloud<PointXYZILID> ground;
+                travel::PointCloud<PointXYZILID> nonground;
+                tgs.estimateGround(*input, ground, nonground, tgs_seconds);
+                input_points = input->size();
+                ground_points = ground.size();
+                nonground_points = nonground.size();
+                const std::string base = output_dir + "/" + std::to_string(row.timeline_frame_index);
+                writeXYZI(base + "_ground.bin", ground);
+                writeXYZI(base + "_nonground.bin", nonground);
+                ++expected_slot;
+            }
+            const auto completed = Clock::now();
+            timing << row.timeline_frame_index << '\t' << row.source_frame_index << '\t'
+                   << row.session_seconds << '\t' << (row.available_slot >= 0 ? 1 : 0) << '\t'
+                   << row.available_slot << '\t' << input_points << '\t' << ground_points << '\t'
+                   << nonground_points << '\t' << (tgs_seconds * 1000.0) << '\t'
+                   << milliseconds(completed - stage_started) << '\t' << queue_delay_ms << '\t'
+                   << std::max(0.0, milliseconds(completed - target)) << "\t0\n";
+            if ((row.timeline_frame_index + 1) % 100 == 0) {
+                timing.flush();
+                std::cout << "[TGS-FULL] frame=" << (row.timeline_frame_index + 1)
+                          << "/4489 available=" << expected_slot << "/3928\n";
+            }
+        }
+        timing.flush();
+        if (expected_slot != 3928) {
+            throw std::runtime_error("full-shadow available frame accounting changed");
+        }
+        std::cout << "[TGS-FULL] complete timeline=4489 available=3928\n";
+        return 0;
+    } catch (const std::exception& error) {
+        std::cerr << "[TGS-FULL] " << error.what() << '\n';
+        return 2;
+    }
+}

experiments/perception/worker/m49_t3_travel/run_tgs_full_shadow.sh

@@ -0,0 +1,23 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+readonly INPUT_ROOT=/tgs/inputs
+readonly OUTPUT_ROOT=/tgs/outputs/causal_rolling_1s
+readonly TIMING_PATH=/tgs/tgs-full-timing.tsv
+readonly BINARY=/tmp/run_tgs_full_shadow
+
+test -f "${INPUT_ROOT}/input-manifest.json"
+test -f "${INPUT_ROOT}/schedule.tsv"
+test ! -e /tgs/outputs
+test ! -e "${TIMING_PATH}"
+g++ -std=c++17 -O3 -DNDEBUG -pthread \
+  -I/opt/travel/src/TRAVEL/cpp/travel/core \
+  -I/usr/include/eigen3 \
+  /release/run_tgs_full_shadow.cpp \
+  -o "${BINARY}"
+mkdir -p "${OUTPUT_ROOT}"
+exec /usr/bin/time -v "${BINARY}" \
+  "${INPUT_ROOT}/profiles/causal_rolling_1s" \
+  "${INPUT_ROOT}/schedule.tsv" \
+  "${OUTPUT_ROOT}" \
+  "${TIMING_PATH}"

scripts/build_m49_tgs_full_shadow_worker_artifact.py

@@ -0,0 +1,156 @@
+#!/usr/bin/env python3
+"""Build the deterministic Worker 006 release for the complete TGS shadow."""
+
+from __future__ import annotations
+
+import argparse
+import gzip
+import hashlib
+import io
+import json
+import re
+import subprocess
+import tarfile
+import tempfile
+from pathlib import Path
+
+REPOSITORY_ROOT = Path(__file__).resolve().parents[1]
+SOURCES = (
+    Path("experiments/perception/worker/m49_t3_travel/prepare_tgs_fail_closed_inputs.py"),
+    Path("experiments/perception/worker/m49_t3_travel/build_tgs_fail_closed_evidence.py"),
+    Path("experiments/perception/worker/m49_t3_travel/prepare_tgs_full_shadow_inputs.py"),
+    Path("experiments/perception/worker/m49_t3_travel/run_tgs_full_shadow.cpp"),
+    Path("experiments/perception/worker/m49_t3_travel/run_tgs_full_shadow.sh"),
+    Path("experiments/perception/worker/m49_t3_travel/build_tgs_full_shadow_evidence.py"),
+    Path("experiments/perception/worker/Invoke-M49TgsFullShadow.ps1"),
+    Path("experiments/perception/worker/Invoke-M49TgsFullShadowAsInteractiveUser.ps1"),
+    Path("config/perception/m49-tgs-full-shadow-v1.json"),
+)
+PATCH_ID = re.compile(r"^[A-Za-z0-9._-]{1,96}$")
+
+
+class ArtifactBuildError(RuntimeError):
+    """The full-shadow release cannot be built from the declared source."""
+
+
+def sha256_file(path: Path) -> str:
+    digest = hashlib.sha256()
+    with path.open("rb") as stream:
+        for chunk in iter(lambda: stream.read(1024 * 1024), b""):
+            digest.update(chunk)
+    return digest.hexdigest()
+
+
+def git_revision() -> str:
+    result = subprocess.run(
+        ["git", "rev-parse", "HEAD"], cwd=REPOSITORY_ROOT, check=True, capture_output=True, text=True
+    )
+    return result.stdout.strip()
+
+
+def tar_info(path: Path, arcname: str) -> tarfile.TarInfo:
+    info = tarfile.TarInfo(arcname)
+    info.uid = info.gid = 0
+    info.uname = info.gname = "root"
+    info.mtime = 0
+    if path.is_dir():
+        info.type = tarfile.DIRTYPE
+        info.mode = 0o755
+    else:
+        info.type = tarfile.REGTYPE
+        info.mode = 0o755 if path.suffix in {".sh", ".ps1", ".py"} else 0o644
+        info.size = path.stat().st_size
+    return info
+
+
+def write_archive(stage: Path, target: Path) -> None:
+    members = [stage / "manifest.env", stage / "files.txt", stage / "payload"]
+    members.extend(sorted((stage / "payload").rglob("*")))
+    target.parent.mkdir(parents=True, exist_ok=True)
+    with (
+        target.open("wb") as raw,
+        gzip.GzipFile(filename="", mode="wb", fileobj=raw, mtime=0) as compressed,
+        tarfile.open(fileobj=compressed, mode="w", format=tarfile.PAX_FORMAT) as archive,
+    ):
+        for path in members:
+            info = tar_info(path, path.relative_to(stage).as_posix())
+            if path.is_file():
+                with path.open("rb") as stream:
+                    archive.addfile(info, stream)
+            else:
+                archive.addfile(info, io.BytesIO())
+
+
+def build(patch_id: str, output_directory: Path, *, revision: str | None = None) -> dict[str, object]:
+    if PATCH_ID.fullmatch(patch_id) is None:
+        raise ArtifactBuildError("patch id is invalid")
+    sources = tuple(REPOSITORY_ROOT / source for source in SOURCES)
+    if any(path.is_symlink() or not path.is_file() for path in sources):
+        raise ArtifactBuildError("release input is not a regular file")
+    selected_revision = revision or git_revision()
+    if re.fullmatch(r"[a-f0-9]{40}", selected_revision) is None:
+        raise ArtifactBuildError("artifact revision is invalid")
+    with tempfile.TemporaryDirectory(prefix="mission-core-m49-tgs-full-") as directory:
+        stage = Path(directory)
+        payload = stage / "payload"
+        payload.mkdir()
+        files: dict[str, dict[str, object]] = {}
+        for source in sources:
+            destination = payload / source.name
+            destination.write_bytes(source.read_bytes())
+            files[destination.name] = {"bytes": destination.stat().st_size, "sha256": sha256_file(destination)}
+        release = {
+            "schema_version": "missioncore.m49-tgs-full-shadow-worker-release/v1",
+            "patch_id": patch_id,
+            "code_revision": selected_revision,
+            "worker_id": "worker-006",
+            "candidate_id": "travel-tgs-full-shadow",
+            "license": "GPL-3.0-or-later",
+            "source_pack_sha256": "0685d24219d8236caf8b7f1685e93f6d6b59e7fd015a768d88a92bbe8b154944",
+            "images": {
+                "travel": "sha256:7b412020f4d8392d1d1ed1b33beadc44140f0ea8f781e62dd69796042334300f",
+                "parity": "sha256:ceb13548617e4bd3f619766bfdff00af3fa5160946b367828da6d2233dcdcba0",
+            },
+            "authority": {
+                "visual_quality_accepted": False,
+                "traversability_accepted": False,
+                "realtime_accepted": False,
+                "integrated_graph_performance_accepted": False,
+                "navigation_or_actuation_allowed": False,
+            },
+            "files": files,
+        }
+        release_path = payload / "release.json"
+        release_path.write_text(json.dumps(release, indent=2, sort_keys=True) + "\n", encoding="utf-8")
+        payload_names = sorted((*files, release_path.name))
+        (stage / "manifest.env").write_text(
+            f"id={patch_id}\ncomponent=mission-core-worker\ntype=qualification-release\n", encoding="utf-8"
+        )
+        (stage / "files.txt").write_text("\n".join(payload_names) + "\n", encoding="utf-8")
+        target = output_directory.resolve() / f"nodedc-{patch_id}.tgz"
+        write_archive(stage, target)
+    return {
+        "ok": True,
+        "artifact": str(target),
+        "sha256": sha256_file(target),
+        "patch_id": patch_id,
+        "code_revision": selected_revision,
+        "payload_files": payload_names,
+    }
+
+
+def main() -> int:
+    parser = argparse.ArgumentParser()
+    parser.add_argument("patch_id")
+    parser.add_argument("--output-directory", type=Path, default=REPOSITORY_ROOT / ".runtime/worker-artifacts")
+    arguments = parser.parse_args()
+    try:
+        result = build(arguments.patch_id, arguments.output_directory)
+    except (ArtifactBuildError, OSError, subprocess.SubprocessError) as exc:
+        parser.error(str(exc))
+    print(json.dumps(result, indent=2, sort_keys=True))
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

src/k1link/laboratory/m49_tgs_fail_closed.py

@@ -0,0 +1,440 @@
+"""Seal and verify bounded gravity-aligned TRAVEL TGS evidence."""
+
+from __future__ import annotations
+
+import hashlib
+import json
+import os
+import shutil
+import tempfile
+from dataclasses import dataclass
+from datetime import UTC, datetime
+from pathlib import Path
+from typing import Any, Final
+
+M49_TGS_RESULT_SCHEMA: Final = "missioncore.m49-tgs-fail-closed-result/v1"
+M49_TGS_REPORT_SCHEMA: Final = "missioncore.m49-tgs-fail-closed-report/v1"
+M49_TGS_WORKER_RESULT_SCHEMA: Final = "missioncore.m49-tgs-fail-closed-evidence-result/v1"
+M49_TGS_PREFIX: Final = "m49-tgs-fail-closed-"
+M49_TGS_PROFILE_ID: Final = "m49-ravnoves00-tgs-fail-closed-evidence/v1"
+M49_TGS_ANCHORS: Final = (171, 306, 368, 402, 450, 509, 525, 744, 1122, 1856)
+M49_TGS_PROFILES: Final = ("current_increment", "causal_rolling_1s")
+_MAX_JSON_BYTES: Final = 1024 * 1024
+_HASH_CHUNK_BYTES: Final = 1024 * 1024
+
+
+class M49TgsFailClosedError(RuntimeError):
+    """The TGS evidence pack is unavailable or failed its immutable contract."""
+
+
+@dataclass(frozen=True, slots=True)
+class M49TgsFailClosedResult:
+    result_id: str
+    root: Path
+    manifest: dict[str, Any]
+    report: dict[str, Any]
+
+    @property
+    def evidence_path(self) -> Path:
+        return self.root / "evidence.npz"
+
+
+def seal_m49_tgs_fail_closed(
+    *,
+    source_root: Path,
+    destination_root: Path,
+    profile_path: Path,
+    linked_visual_result_id: str,
+    created_at_utc: str | None = None,
+) -> M49TgsFailClosedResult:
+    source = _real_directory(source_root, "M49 Worker evidence")
+    destination = destination_root.expanduser().absolute()
+    destination.mkdir(parents=True, exist_ok=True)
+    if destination.is_symlink():
+        raise M49TgsFailClosedError("M49 destination must not be a symlink")
+    profile = _json_file(profile_path, "M49 profile")
+    worker_result = _json_file(source / "result.json", "M49 Worker result")
+    worker_summary = _json_file(source / "worker-summary.json", "M49 Worker summary")
+    input_manifest = _json_file(source / "input-manifest.json", "M49 input manifest")
+    timing = _timing_metrics(source / "tgs-timing.tsv")
+    _validate_source(profile, worker_result, worker_summary, input_manifest, source)
+    if (
+        not linked_visual_result_id.startswith("m4-threat-replay-")
+        or len(linked_visual_result_id) != len("m4-threat-replay-") + 64
+    ):
+        raise M49TgsFailClosedError("M49 linked visual result is invalid")
+
+    evidence_sha = _file_sha256(source / "evidence.npz")
+    identity = {
+        "schema_version": M49_TGS_RESULT_SCHEMA,
+        "source": {
+            "source_id": "RAVNOVES00",
+            "source_session_id": "20260720T065719Z_viewer_live",
+            "source_pack_sha256": worker_result["source_pack_sha256"],
+            "input_manifest_sha256": worker_result["input_manifest_sha256"],
+            "linked_visual_result_id": linked_visual_result_id,
+            "anchor_frame_indices": list(M49_TGS_ANCHORS),
+        },
+        "configuration": {
+            "profile_id": M49_TGS_PROFILE_ID,
+            "config_sha256": worker_result["config_sha256"],
+            "coordinate_frame": "map-gravity-local",
+            "primary_profile": "causal_rolling_1s",
+            "cell_size_m": worker_result["costmap"]["cell_size_m"],
+            "radius_m": worker_result["costmap"]["radius_m"],
+        },
+        "method": {
+            "execution_class": "deterministic",
+            "pipeline_id": "travel-tgs-gravity-aligned-fail-closed/v1",
+            "travel_revision": profile["source"]["travel_revision"],
+            "aos_used": False,
+            "missing_support_means_free": False,
+            "eligible_point_accounting": "exact-multiset-complement",
+        },
+        "evidence": {
+            "sha256": evidence_sha,
+            "byte_length": (source / "evidence.npz").stat().st_size,
+            "state_codes": profile["state_codes"],
+        },
+        "authority": {
+            "commands_enabled": False,
+            "navigation_or_safety_accepted": False,
+            "visual_quality_accepted": False,
+        },
+    }
+    identity_sha256 = _canonical_sha256(identity)
+    result_id = f"{M49_TGS_PREFIX}{identity_sha256}"
+    target = destination / result_id
+    if target.exists():
+        return read_m49_tgs_fail_closed(target)
+
+    created = created_at_utc or datetime.now(tz=UTC).isoformat().replace("+00:00", "Z")
+    anchors = worker_result["anchors"]
+    primary = [row for row in anchors if row["profile_id"] == "causal_rolling_1s"]
+    report = {
+        "schema_version": M49_TGS_REPORT_SCHEMA,
+        "result_id": result_id,
+        "created_at_utc": created,
+        "source": identity["source"],
+        "configuration": {
+            **identity["configuration"],
+            "state_priority": profile["costmap"]["state_priority"],
+            "tgs": profile["tgs"],
+        },
+        "method": {
+            **identity["method"],
+            "components": [
+                {
+                    "kind": "algorithm",
+                    "name": "TRAVEL GroundSeg",
+                    "version": profile["source"]["travel_revision"],
+                    "role": "gravity-aligned ground/nonground separation",
+                },
+                {
+                    "kind": "algorithm",
+                    "name": "fail-closed complement adapter",
+                    "version": "v1",
+                    "role": "retain rejected points and explicit unobserved cells",
+                },
+                {
+                    "kind": "runtime",
+                    "name": "Worker 006 CPU qualification",
+                    "version": worker_summary["code_revision"],
+                    "role": "20 bounded TGS invocations without GPU",
+                },
+            ],
+        },
+        "execution": {
+            "worker": "Worker 006",
+            "device": "cpu",
+            "gpu_used": False,
+            "wrapper_elapsed_seconds": worker_summary["wall_seconds"],
+            "canonical_triton_id": worker_summary["canonical_triton_id"],
+            "canonical_triton_health": worker_summary["canonical_triton_health"],
+            "free_memory_gib_before": worker_summary["free_memory_gib_before"],
+        },
+        "metrics": {
+            "anchor_count": len(M49_TGS_ANCHORS),
+            "anchor_profile_count": len(anchors),
+            "all_eligible_points_accounted": True,
+            "primary_profile": "causal_rolling_1s",
+            "primary": primary,
+            "costmap_cell_count": worker_result["costmap"]["cell_count"],
+            "process_wall_current_p50_ms": timing["current_increment"]["p50_ms"],
+            "process_wall_current_max_ms": timing["current_increment"]["max_ms"],
+            "process_wall_rolling_p50_ms": timing["causal_rolling_1s"]["p50_ms"],
+            "process_wall_rolling_max_ms": timing["causal_rolling_1s"]["max_ms"],
+            "process_max_rss_kib": timing["max_rss_kib"],
+        },
+        "acceptance": {
+            "representation_complete": True,
+            "all_points_accounted": True,
+            "aos_absent": True,
+            "gpu_absent": True,
+            "visual_quality_accepted": False,
+            "traversability_accepted": False,
+        },
+        "decision": {
+            "state": "visual-review-required",
+            "candidate_retained": True,
+            "next_action": (
+                "Review exact gravity-aligned points and fail-closed costmap "
+                "on the ten immutable anchors."
+            ),
+        },
+        "limitations": [
+            "The ten anchors are bounded diagnostic evidence, not a full 4,489-frame replay.",
+            "No independent terrain or traversability truth is available.",
+            "No vehicle envelope exists, so occupied cells do not grant or deny physical passage.",
+            "Visual quality, realtime integration, navigation and actuation remain unaccepted.",
+        ],
+        "authority": {
+            "mode": "replay-simulated",
+            "commands_enabled": False,
+            "navigation_or_safety_accepted": False,
+            "visual_quality_accepted": False,
+        },
+        "visual_review": {
+            "instrument": "m4-canonical-reference-graph",
+            "linked_visual_result_id": linked_visual_result_id,
+            "anchors": list(M49_TGS_ANCHORS),
+            "profiles": list(M49_TGS_PROFILES),
+            "default_profile": "causal_rolling_1s",
+            "point_states": profile["state_codes"],
+        },
+    }
+
+    with tempfile.TemporaryDirectory(prefix="mission-core-m49-tgs-", dir=destination) as raw:
+        staging = Path(raw) / result_id
+        staging.mkdir()
+        for name in (
+            "evidence.npz",
+            "worker-summary.json",
+            "input-manifest.json",
+            "tgs-timing.tsv",
+        ):
+            shutil.copyfile(source / name, staging / name)
+        _write_json(staging / "report.json", report)
+        artifacts = [
+            _artifact(staging / "report.json", "report", M49_TGS_REPORT_SCHEMA, "application/json"),
+            _artifact(
+                staging / "evidence.npz", "visual-spatial-evidence", None, "application/x-npz"
+            ),
+            _artifact(staging / "worker-summary.json", "runtime-summary", None, "application/json"),
+            _artifact(staging / "input-manifest.json", "source-manifest", None, "application/json"),
+            _artifact(
+                staging / "tgs-timing.tsv", "runtime-timing", None, "text/tab-separated-values"
+            ),
+        ]
+        manifest = {
+            "schema_version": M49_TGS_RESULT_SCHEMA,
+            "result_id": result_id,
+            "created_at_utc": created,
+            "identity_sha256": identity_sha256,
+            "identity": identity,
+            "artifacts": artifacts,
+            "authority": report["authority"],
+            "ground_truth": False,
+        }
+        _write_json(staging / "manifest.json", manifest)
+        os.replace(staging, target)
+    return read_m49_tgs_fail_closed(target)
+
+
+def read_m49_tgs_fail_closed(root: Path) -> M49TgsFailClosedResult:
+    candidate = _real_directory(root, "M49 result")
+    if not candidate.name.startswith(M49_TGS_PREFIX):
+        raise M49TgsFailClosedError("M49 result identity is invalid")
+    manifest = _json_file(candidate / "manifest.json", "M49 manifest")
+    report = _json_file(candidate / "report.json", "M49 report")
+    if (
+        manifest.get("schema_version") != M49_TGS_RESULT_SCHEMA
+        or manifest.get("result_id") != candidate.name
+        or report.get("schema_version") != M49_TGS_REPORT_SCHEMA
+        or report.get("result_id") != candidate.name
+    ):
+        raise M49TgsFailClosedError("M49 result contract changed")
+    identity = manifest.get("identity")
+    identity_sha = manifest.get("identity_sha256")
+    if (
+        not isinstance(identity, dict)
+        or not isinstance(identity_sha, str)
+        or _canonical_sha256(identity) != identity_sha
+        or candidate.name != f"{M49_TGS_PREFIX}{identity_sha}"
+    ):
+        raise M49TgsFailClosedError("M49 identity proof changed")
+    artifacts = manifest.get("artifacts")
+    if not isinstance(artifacts, list) or len(artifacts) != 5:
+        raise M49TgsFailClosedError("M49 artifact manifest changed")
+    for item in artifacts:
+        if not isinstance(item, dict):
+            raise M49TgsFailClosedError("M49 artifact descriptor changed")
+        path = candidate / str(item.get("path", ""))
+        if (
+            path.is_symlink()
+            or not path.is_file()
+            or path.parent != candidate
+            or path.stat().st_size != item.get("byte_length")
+            or _file_sha256(path) != item.get("sha256")
+        ):
+            raise M49TgsFailClosedError("M49 artifact proof changed")
+    return M49TgsFailClosedResult(candidate.name, candidate, manifest, report)
+
+
+def _validate_source(
+    profile: dict[str, Any],
+    worker_result: dict[str, Any],
+    worker_summary: dict[str, Any],
+    input_manifest: dict[str, Any],
+    source: Path,
+) -> None:
+    if (
+        profile.get("schema_version") != "missioncore.m49-tgs-fail-closed-evidence-profile/v1"
+        or profile.get("profile_id") != M49_TGS_PROFILE_ID
+        or tuple(profile.get("anchors", ())) != M49_TGS_ANCHORS
+        or profile.get("invariants", {}).get("aos_allowed") is not False
+        or profile.get("invariants", {}).get("missing_support_means_free") is not False
+    ):
+        raise M49TgsFailClosedError("M49 profile changed")
+    if (
+        worker_result.get("schema_version") != M49_TGS_WORKER_RESULT_SCHEMA
+        or worker_result.get("status") != "passed"
+        or worker_result.get("summary", {}).get("aos_used") is not False
+        or worker_result.get("summary", {}).get("all_eligible_points_accounted") is not True
+        or worker_result.get("summary", {}).get("primary_profile") != "causal_rolling_1s"
+        or len(worker_result.get("anchors", ())) != 20
+    ):
+        raise M49TgsFailClosedError("M49 Worker result changed")
+    if (
+        input_manifest.get("schema_version") != "missioncore.m49-tgs-fail-closed-input/v1"
+        or input_manifest.get("coordinate_frame") != "map-gravity-local"
+        or len(input_manifest.get("records", ())) != 20
+    ):
+        raise M49TgsFailClosedError("M49 input manifest changed")
+    if (
+        worker_summary.get("gpu_requested") is not False
+        and worker_summary.get("gpu_requested") is not None
+    ):
+        raise M49TgsFailClosedError("M49 Worker GPU contract changed")
+    if (
+        worker_summary.get("aos_used") is not False
+        or worker_summary.get("all_eligible_points_accounted") is not True
+        or worker_summary.get("canonical_triton_health") != "healthy"
+    ):
+        raise M49TgsFailClosedError("M49 Worker cleanup changed")
+    evidence = worker_result.get("evidence", {})
+    if (
+        evidence.get("path") != "evidence.npz"
+        or evidence.get("bytes") != (source / "evidence.npz").stat().st_size
+        or evidence.get("sha256") != _file_sha256(source / "evidence.npz")
+        or worker_result.get("input_manifest_sha256")
+        != _file_sha256(source / "input-manifest.json")
+    ):
+        raise M49TgsFailClosedError("M49 evidence proof changed")
+
+
+def _artifact(
+    path: Path,
+    role: str,
+    schema_version: str | None,
+    media_type: str,
+) -> dict[str, object]:
+    result: dict[str, object] = {
+        "role": role,
+        "path": path.name,
+        "byte_length": path.stat().st_size,
+        "sha256": _file_sha256(path),
+        "media_type": media_type,
+    }
+    if schema_version is not None:
+        result["schema_version"] = schema_version
+    return result
+
+
+def _timing_metrics(path: Path) -> dict[str, Any]:
+    if path.is_symlink() or not path.is_file():
+        raise M49TgsFailClosedError("M49 timing evidence is unavailable")
+    profiles: dict[str, list[float]] = {name: [] for name in M49_TGS_PROFILES}
+    maximum_rss = 0
+    lines = path.read_text(encoding="utf-8-sig").splitlines()
+    if not lines or lines[0] != "profile\tslot\twall_seconds\tmax_rss_kib":
+        raise M49TgsFailClosedError("M49 timing evidence changed")
+    for line in lines[1:]:
+        fields = line.split("\t")
+        if len(fields) != 4 or fields[0] not in profiles:
+            raise M49TgsFailClosedError("M49 timing row changed")
+        profiles[fields[0]].append(float(fields[2]))
+        maximum_rss = max(maximum_rss, int(fields[3]))
+    if any(len(values) != len(M49_TGS_ANCHORS) for values in profiles.values()):
+        raise M49TgsFailClosedError("M49 timing coverage changed")
+    result: dict[str, Any] = {"max_rss_kib": maximum_rss}
+    for name, values in profiles.items():
+        ordered = sorted(values)
+        result[name] = {
+            "p50_ms": ordered[len(ordered) // 2] * 1000,
+            "max_ms": max(ordered) * 1000,
+        }
+    return result
+
+
+def _real_directory(path: Path, label: str) -> Path:
+    candidate = path.expanduser().absolute()
+    if candidate.is_symlink():
+        raise M49TgsFailClosedError(f"{label} must not be a symlink")
+    try:
+        resolved = candidate.resolve(strict=True)
+    except OSError as exc:
+        raise M49TgsFailClosedError(f"{label} is unavailable") from exc
+    if not resolved.is_dir():
+        raise M49TgsFailClosedError(f"{label} is unavailable")
+    return resolved
+
+
+def _json_file(path: Path, label: str) -> dict[str, Any]:
+    if path.is_symlink() or not path.is_file() or path.stat().st_size > _MAX_JSON_BYTES:
+        raise M49TgsFailClosedError(f"{label} is unavailable")
+    try:
+        value = json.loads(path.read_text(encoding="utf-8-sig"))
+    except (json.JSONDecodeError, OSError) as exc:
+        raise M49TgsFailClosedError(f"{label} is invalid") from exc
+    if not isinstance(value, dict):
+        raise M49TgsFailClosedError(f"{label} is invalid")
+    return value
+
+
+def _write_json(path: Path, value: object) -> None:
+    path.write_bytes(_canonical_json(value) + b"\n")
+
+
+def _canonical_json(value: object) -> bytes:
+    return json.dumps(
+        value,
+        ensure_ascii=False,
+        sort_keys=True,
+        separators=(",", ":"),
+        allow_nan=False,
+    ).encode("utf-8")
+
+
+def _canonical_sha256(value: object) -> str:
+    return hashlib.sha256(_canonical_json(value)).hexdigest()
+
+
+def _file_sha256(path: Path) -> str:
+    digest = hashlib.sha256()
+    with path.open("rb") as stream:
+        for chunk in iter(lambda: stream.read(_HASH_CHUNK_BYTES), b""):
+            digest.update(chunk)
+    return digest.hexdigest()
+
+
+__all__ = [
+    "M49_TGS_ANCHORS",
+    "M49_TGS_PREFIX",
+    "M49_TGS_REPORT_SCHEMA",
+    "M49_TGS_RESULT_SCHEMA",
+    "M49TgsFailClosedError",
+    "M49TgsFailClosedResult",
+    "read_m49_tgs_fail_closed",
+    "seal_m49_tgs_fail_closed",
+]

src/k1link/laboratory/m49_tgs_full_shadow.py

@@ -0,0 +1,304 @@
+"""Seal and verify the complete source-paced TRAVEL TGS shadow."""
+
+from __future__ import annotations
+
+import hashlib
+import json
+import shutil
+import tempfile
+from dataclasses import dataclass
+from datetime import UTC, datetime
+from pathlib import Path
+from typing import Any, Final
+
+RESULT_SCHEMA: Final = "missioncore.m49-tgs-full-shadow-lab/v1"
+REPORT_SCHEMA: Final = "missioncore.m49-tgs-full-shadow-report/v1"
+WORKER_SCHEMA: Final = "missioncore.m49-tgs-full-shadow-result/v1"
+PREFIX: Final = "m49-tgs-full-shadow-"
+PROFILE_SCHEMA: Final = "missioncore.m49-tgs-full-shadow-profile/v1"
+EVIDENCE_FILES: Final = (
+    "costmap-cell-centers-xy-m.npy",
+    "costmap-cell-indices-xy.npy",
+    "costmap-states.npy",
+    "costmap-z-bounds-m.npy",
+    "frames.ndjson",
+)
+_HASH_CHUNK_BYTES: Final = 1024 * 1024
+_MAX_JSON_BYTES: Final = 4 * 1024 * 1024
+
+
+class M49TgsFullShadowError(RuntimeError):
+    """The full TGS shadow is unavailable or violates its immutable contract."""
+
+
+@dataclass(frozen=True, slots=True)
+class M49TgsFullShadowResult:
+    result_id: str
+    root: Path
+    manifest: dict[str, Any]
+    report: dict[str, Any]
+
+
+def _sha256(path: Path) -> str:
+    digest = hashlib.sha256()
+    with path.open("rb") as stream:
+        for chunk in iter(lambda: stream.read(_HASH_CHUNK_BYTES), b""):
+            digest.update(chunk)
+    return digest.hexdigest()
+
+
+def _canonical_sha256(value: object) -> str:
+    content = json.dumps(
+        value, ensure_ascii=False, sort_keys=True, separators=(",", ":"), allow_nan=False
+    ).encode("utf-8")
+    return hashlib.sha256(content).hexdigest()
+
+
+def _json(path: Path, label: str) -> dict[str, Any]:
+    if path.is_symlink() or not path.is_file() or path.stat().st_size > _MAX_JSON_BYTES:
+        raise M49TgsFullShadowError(f"{label} is unavailable")
+    value = json.loads(path.read_text(encoding="utf-8-sig"))
+    if not isinstance(value, dict):
+        raise M49TgsFullShadowError(f"{label} is invalid")
+    return value
+
+
+def _artifact(path: Path, role: str, media_type: str) -> dict[str, object]:
+    return {
+        "path": path.name,
+        "role": role,
+        "media_type": media_type,
+        "byte_length": path.stat().st_size,
+        "sha256": _sha256(path),
+    }
+
+
+def seal_m49_tgs_full_shadow(
+    *,
+    source_root: Path,
+    destination_root: Path,
+    profile_path: Path,
+    linked_visual_result_id: str,
+    linked_semantic_result_id: str,
+    created_at_utc: str | None = None,
+) -> M49TgsFullShadowResult:
+    source = source_root.expanduser().resolve(strict=True)
+    if source.is_symlink() or not source.is_dir():
+        raise M49TgsFullShadowError("Worker evidence root is unavailable")
+    destination = destination_root.expanduser().absolute()
+    destination.mkdir(parents=True, exist_ok=True)
+    if destination.is_symlink():
+        raise M49TgsFullShadowError("destination must not be a symlink")
+    profile = _json(profile_path, "full-shadow profile")
+    worker = _json(source / "result.json", "Worker result")
+    summary = _json(source / "worker-summary.json", "Worker summary")
+    timeline = worker.get("timeline")
+    if (
+        profile.get("schema_version") != PROFILE_SCHEMA
+        or worker.get("schema_version") != WORKER_SCHEMA
+        or not isinstance(timeline, dict)
+        or (
+            timeline.get("frame_count") != 4489
+            or timeline.get("available_lidar_frame_count") != 3928
+            or timeline.get("missing_lidar_frame_count") != 561
+        )
+        or worker.get("point_accounting", {}).get("unaccounted") != 0
+        or summary.get("schema_version") != "missioncore.m49-tgs-full-shadow-worker-summary/v1"
+        or summary.get("gpu_requested") is not False
+        or summary.get("aos_used") is not False
+        or summary.get("all_timeline_frames_accounted") is not True
+        or summary.get("all_eligible_points_accounted") is not True
+        or summary.get("canonical_triton_health") != "healthy"
+    ):
+        raise M49TgsFullShadowError("Worker full-shadow contract changed")
+    if (
+        not linked_visual_result_id.startswith("m4-threat-replay-")
+        or len(linked_visual_result_id) != len("m4-threat-replay-") + 64
+    ):
+        raise M49TgsFullShadowError("linked visual result is invalid")
+    if (
+        not linked_semantic_result_id.startswith("e47-semantic-slam-")
+        or len(linked_semantic_result_id) != len("e47-semantic-slam-") + 64
+    ):
+        raise M49TgsFullShadowError("linked semantic result is invalid")
+    for name in EVIDENCE_FILES:
+        path = source / name
+        proof = worker.get("files", {}).get(name, {})
+        if (
+            path.is_symlink()
+            or not path.is_file()
+            or proof.get("bytes") != path.stat().st_size
+            or proof.get("sha256") != _sha256(path)
+        ):
+            raise M49TgsFullShadowError(f"Worker evidence changed: {name}")
+    identity = {
+        "schema_version": RESULT_SCHEMA,
+        "source_pack_sha256": worker["source_pack_sha256"],
+        "input_manifest_sha256": worker["input_manifest_sha256"],
+        "config_sha256": worker["config_sha256"],
+        "linked_visual_result_id": linked_visual_result_id,
+        "linked_semantic_result_id": linked_semantic_result_id,
+        "files": {name: worker["files"][name]["sha256"] for name in EVIDENCE_FILES},
+        "authority": {
+            "commands_enabled": False,
+            "navigation_or_safety_accepted": False,
+            "visual_quality_accepted": False,
+        },
+    }
+    identity_sha256 = _canonical_sha256(identity)
+    result_id = f"{PREFIX}{identity_sha256}"
+    target = destination / result_id
+    if target.exists():
+        return read_m49_tgs_full_shadow(target)
+    created = created_at_utc or datetime.now(tz=UTC).isoformat().replace("+00:00", "Z")
+    performance_accepted = worker.get("status") == "passed"
+    report = {
+        "schema_version": REPORT_SCHEMA,
+        "result_id": result_id,
+        "created_at_utc": created,
+        "source": {
+            "source_id": "RAVNOVES00",
+            "source_session_id": "20260720T065719Z_viewer_live",
+            "source_pack_sha256": worker["source_pack_sha256"],
+            "linked_visual_result_id": linked_visual_result_id,
+            "linked_semantic_result_id": linked_semantic_result_id,
+        },
+        "configuration": {
+            "profile_id": profile["profile_id"],
+            "config_sha256": worker["config_sha256"],
+            "coordinate_frame": "map-gravity-local",
+            "history_seconds": profile["profile"]["history_seconds"],
+            "cell_size_m": worker["costmap"]["cell_size_m"],
+            "radius_m": worker["costmap"]["radius_m"],
+            "state_priority": profile["costmap"]["state_priority"],
+        },
+        "execution": {
+            "worker": "Worker 006",
+            "device": "cpu",
+            "gpu_used": False,
+            "aos_used": False,
+            "wrapper_elapsed_seconds": summary["wall_seconds"],
+            "canonical_triton_id": summary["canonical_triton_id"],
+            "canonical_triton_health": summary["canonical_triton_health"],
+        },
+        "timeline": worker["timeline"],
+        "point_accounting": worker["point_accounting"],
+        "performance": worker["performance"],
+        "acceptance": {
+            **worker["acceptance"],
+            "representation_complete": True,
+            "visual_quality_accepted": False,
+            "integrated_graph_performance_accepted": False,
+        },
+        "decision": {
+            "state": (
+                "source-paced-qualified-visual-review-required"
+                if performance_accepted
+                else "performance-rejected"
+            ),
+            "candidate_retained": performance_accepted,
+            "next_action": (
+                "Review the complete camera-synchronised TGS costmap timeline; "
+                "then measure the integrated graph regression separately."
+            ),
+        },
+        "limitations": [
+            "The run proves recorded source-paced CPU shadow performance, not live sensor transport.",
+            "No independent traversability truth or vehicle envelope is present.",
+            "Missing LiDAR frames are explicit all-cell UNOBSERVED and never inferred free.",
+            "Camera projection, navigation and actuation remain disabled.",
+        ],
+        "authority": {
+            "mode": "replay-simulated",
+            "commands_enabled": False,
+            "realtime_shadow_accepted": performance_accepted,
+            "integrated_graph_performance_accepted": False,
+            "navigation_or_safety_accepted": False,
+            "visual_quality_accepted": False,
+        },
+        "visual_review": {
+            "instrument": "m4-canonical-reference-graph",
+            "linked_visual_result_id": linked_visual_result_id,
+            "linked_semantic_result_id": linked_semantic_result_id,
+            "frame_count": 4489,
+            "state_codes": profile["state_codes"],
+        },
+    }
+    with tempfile.TemporaryDirectory(prefix="mission-core-m49-tgs-full-", dir=destination) as raw:
+        staging = Path(raw) / result_id
+        staging.mkdir()
+        for name in EVIDENCE_FILES:
+            shutil.copyfile(source / name, staging / name)
+        shutil.copyfile(source / "worker-summary.json", staging / "worker-summary.json")
+        (staging / "report.json").write_text(
+            json.dumps(report, indent=2, sort_keys=True) + "\n", encoding="utf-8"
+        )
+        artifacts = [
+            _artifact(staging / "report.json", "report", "application/json"),
+            _artifact(staging / "worker-summary.json", "runtime-summary", "application/json"),
+        ]
+        artifacts.extend(
+            _artifact(
+                staging / name,
+                "frame-catalog" if name == "frames.ndjson" else "spatial-evidence",
+                "application/x-ndjson" if name == "frames.ndjson" else "application/x-npy",
+            )
+            for name in EVIDENCE_FILES
+        )
+        manifest = {
+            "schema_version": RESULT_SCHEMA,
+            "result_id": result_id,
+            "created_at_utc": created,
+            "identity_sha256": identity_sha256,
+            "identity": identity,
+            "artifacts": artifacts,
+        }
+        (staging / "manifest.json").write_text(
+            json.dumps(manifest, indent=2, sort_keys=True) + "\n", encoding="utf-8"
+        )
+        staging.replace(target)
+    return read_m49_tgs_full_shadow(target)
+
+
+def read_m49_tgs_full_shadow(root: Path) -> M49TgsFullShadowResult:
+    candidate = root.expanduser().resolve(strict=True)
+    if candidate.is_symlink() or not candidate.is_dir() or not candidate.name.startswith(PREFIX):
+        raise M49TgsFullShadowError("full-shadow result root is invalid")
+    manifest = _json(candidate / "manifest.json", "full-shadow manifest")
+    report = _json(candidate / "report.json", "full-shadow report")
+    identity = manifest.get("identity")
+    if (
+        manifest.get("schema_version") != RESULT_SCHEMA
+        or report.get("schema_version") != REPORT_SCHEMA
+        or manifest.get("result_id") != candidate.name
+        or report.get("result_id") != candidate.name
+        or not isinstance(identity, dict)
+        or manifest.get("identity_sha256") != _canonical_sha256(identity)
+        or candidate.name != f"{PREFIX}{manifest['identity_sha256']}"
+    ):
+        raise M49TgsFullShadowError("full-shadow identity changed")
+    artifacts = manifest.get("artifacts")
+    if not isinstance(artifacts, list):
+        raise M49TgsFullShadowError("full-shadow artifact catalog changed")
+    for artifact in artifacts:
+        if not isinstance(artifact, dict) or not isinstance(artifact.get("path"), str):
+            raise M49TgsFullShadowError("full-shadow artifact entry changed")
+        path = candidate / artifact["path"]
+        if (
+            path.parent != candidate
+            or path.is_symlink()
+            or not path.is_file()
+            or artifact.get("byte_length") != path.stat().st_size
+            or artifact.get("sha256") != _sha256(path)
+        ):
+            raise M49TgsFullShadowError("full-shadow artifact digest changed")
+    return M49TgsFullShadowResult(candidate.name, candidate, manifest, report)
+
+
+__all__ = [
+    "M49TgsFullShadowError",
+    "M49TgsFullShadowResult",
+    "PREFIX",
+    "read_m49_tgs_full_shadow",
+    "seal_m49_tgs_full_shadow",
+]

src/k1link/web/app.py

@@ -133,6 +133,8 @@ from k1link.web.m48s_fixed_class_detector_lab_api import (
     build_m48s_fixed_class_detector_lab_router,
 )
 from k1link.web.m48t_risk_quality_lab_api import build_m48t_risk_quality_lab_router
+from k1link.web.m49_tgs_fail_closed_api import build_m49_tgs_fail_closed_router
+from k1link.web.m49_tgs_full_shadow_api import build_m49_tgs_full_shadow_router
 from k1link.web.map_api import (
     MapGatewayConfiguration,
     MapGatewayProxy,
@@ -992,6 +994,28 @@ app.include_router(
         ),
     )
 )
+app.include_router(
+    build_m49_tgs_fail_closed_router(
+        root_provider=lambda: (
+            REPOSITORY_ROOT
+            / ".runtime"
+            / "compute-experiments"
+            / "m49"
+            / "tgs-fail-closed-results"
+        ),
+    )
+)
+app.include_router(
+    build_m49_tgs_full_shadow_router(
+        root_provider=lambda: (
+            REPOSITORY_ROOT
+            / ".runtime"
+            / "compute-experiments"
+            / "m49"
+            / "tgs-full-shadow-results"
+        ),
+    )
+)
 app.include_router(
     build_m48s_fixed_class_detector_lab_router(
         root_provider=lambda: (

src/k1link/web/m49_tgs_fail_closed_api.py

@@ -0,0 +1,284 @@
+"""Read-only API for sealed gravity-aligned M49 TGS evidence."""
+
+from __future__ import annotations
+
+import copy
+import json
+import math
+import re
+from collections.abc import Callable
+from functools import lru_cache
+from pathlib import Path
+from typing import Final
+
+import numpy as np
+from fastapi import APIRouter, HTTPException, Query, Response
+
+from k1link.laboratory.m49_tgs_fail_closed import (
+    M49_TGS_ANCHORS,
+    M49_TGS_PREFIX,
+    M49TgsFailClosedError,
+    M49TgsFailClosedResult,
+    read_m49_tgs_fail_closed,
+)
+
+RootProvider = Callable[[], Path | None]
+
+RESULT_ID: Final = re.compile(rf"^{re.escape(M49_TGS_PREFIX)}[a-f0-9]{{64}}$")
+RESULT_VIEW_SCHEMA: Final = "missioncore.m49-tgs-fail-closed-view/v1"
+RESULT_CATALOG_SCHEMA: Final = "missioncore.m49-tgs-fail-closed-catalog/v1"
+ANCHOR_CATALOG_SCHEMA: Final = "missioncore.m49-tgs-anchor-catalog/v1"
+ANCHOR_SPATIAL_SCHEMA: Final = "missioncore.m49-tgs-anchor-spatial/v1"
+ENDPOINT_ROOT: Final = "/api/v1/laboratory/m49/tgs-fail-closed"
+PROFILES: Final = ("current_increment", "causal_rolling_1s")
+
+
+def build_m49_tgs_fail_closed_router(*, root_provider: RootProvider = lambda: None) -> APIRouter:
+    router = APIRouter(prefix=ENDPOINT_ROOT, tags=["laboratory"])
+
+    def result(result_id: str) -> M49TgsFailClosedResult:
+        candidate = _resolve_candidate(root_provider, result_id)
+        try:
+            return _read_result_cached(str(candidate), _signature(candidate))
+        except (M49TgsFailClosedError, OSError, ValueError):
+            raise HTTPException(status_code=404, detail="M49 TGS result not found") from None
+
+    @router.get("/results")
+    def list_results(limit: int = Query(default=1, ge=1, le=10)) -> dict[str, object]:
+        root = _configured_root(root_provider)
+        if root is None:
+            return _empty_catalog(configured=False)
+        items: list[dict[str, object]] = []
+        invalid_total = 0
+        for candidate in sorted(root.iterdir()):
+            if not candidate.is_dir() or RESULT_ID.fullmatch(candidate.name) is None:
+                continue
+            try:
+                sealed = _read_result_cached(str(candidate.resolve()), _signature(candidate))
+                items.append(_project_result(sealed))
+            except (M49TgsFailClosedError, OSError, ValueError):
+                invalid_total += 1
+        items.sort(
+            key=lambda item: (str(item["created_at_utc"]), str(item["result_id"])),
+            reverse=True,
+        )
+        return {
+            "schema_version": RESULT_CATALOG_SCHEMA,
+            "configured": True,
+            "items": items[:limit],
+            "candidate_total": len(items) + invalid_total,
+            "invalid_total": invalid_total,
+            "access": "read-only",
+        }
+
+    @router.get("/{result_id}")
+    def get_result(result_id: str) -> dict[str, object]:
+        return _project_result(result(result_id))
+
+    @router.get("/{result_id}/anchors")
+    def get_anchors(result_id: str) -> dict[str, object]:
+        sealed = result(result_id)
+        return {
+            "schema_version": ANCHOR_CATALOG_SCHEMA,
+            "result_id": result_id,
+            "linked_visual_result_id": sealed.report["visual_review"]["linked_visual_result_id"],
+            "anchors": copy.deepcopy(sealed.report["metrics"]["primary"]),
+            "anchor_count": len(M49_TGS_ANCHORS),
+            "profiles": list(PROFILES),
+            "default_profile": "causal_rolling_1s",
+            "access": "read-only",
+        }
+
+    @router.get("/{result_id}/anchors/{anchor_frame_index}/spatial")
+    def get_anchor_spatial(
+        result_id: str,
+        anchor_frame_index: int,
+        profile: str = Query(default="causal_rolling_1s"),
+    ) -> Response:
+        if profile not in PROFILES or anchor_frame_index not in M49_TGS_ANCHORS:
+            raise HTTPException(status_code=404, detail="M49 TGS anchor not found")
+        sealed = result(result_id)
+        try:
+            content = _anchor_json_cached(
+                str(sealed.evidence_path),
+                result_id,
+                anchor_frame_index,
+                profile,
+                _evidence_signature(sealed.evidence_path),
+            )
+        except (KeyError, OSError, ValueError):
+            raise HTTPException(
+                status_code=503, detail="M49 TGS spatial evidence failed verification"
+            ) from None
+        return Response(
+            content=content,
+            media_type="application/json",
+            headers={
+                "Cache-Control": "private, max-age=31536000, immutable",
+                "X-Content-Type-Options": "nosniff",
+            },
+        )
+
+    return router
+
+
+@lru_cache(maxsize=4)
+def _read_result_cached(result_root: str, signature: tuple[int, ...]) -> M49TgsFailClosedResult:
+    del signature
+    return read_m49_tgs_fail_closed(Path(result_root))
+
+
+@lru_cache(maxsize=24)
+def _anchor_json_cached(
+    evidence_path: str,
+    result_id: str,
+    anchor_frame_index: int,
+    profile: str,
+    signature: tuple[int, int],
+) -> bytes:
+    del signature
+    slot = M49_TGS_ANCHORS.index(anchor_frame_index)
+    with np.load(evidence_path, allow_pickle=False) as evidence:
+        offsets = evidence[f"{profile}_point_offsets"]
+        start = int(offsets[slot])
+        end = int(offsets[slot + 1])
+        points = evidence[f"{profile}_points_xyz_m"][start:end]
+        point_states = evidence[f"{profile}_point_states"][start:end]
+        centers = evidence["costmap_cell_centers_xy_m"]
+        cell_states = evidence[f"{profile}_costmap_states"][slot]
+        z_bounds = evidence[f"{profile}_costmap_z_bounds_m"][slot]
+        if (
+            points.shape[1:] != (3,)
+            or point_states.shape != (points.shape[0],)
+            or centers.shape != (2244, 2)
+            or cell_states.shape != (2244,)
+            or z_bounds.shape != (2244, 2)
+            or not np.isfinite(points).all()
+            or not np.isfinite(centers).all()
+            or not np.isin(point_states, np.asarray([1, 2, 3], dtype=np.uint8)).all()
+            or not np.isin(cell_states, np.asarray([0, 1, 2, 3], dtype=np.uint8)).all()
+        ):
+            raise ValueError("M49 TGS spatial shape changed")
+        payload = {
+            "schema_version": ANCHOR_SPATIAL_SCHEMA,
+            "result_id": result_id,
+            "anchor_frame_index": anchor_frame_index,
+            "source_sequence": anchor_frame_index,
+            "profile": profile,
+            "coordinate_frame": "map-gravity-local",
+            "points_xyz_m": points.astype(float).tolist(),
+            "point_states": point_states.astype(int).tolist(),
+            "costmap": {
+                "cell_size_m": 0.45,
+                "radius_m": 12.0,
+                "centers_xy_m": centers.astype(float).tolist(),
+                "states": cell_states.astype(int).tolist(),
+                "z_bounds_m": [
+                    [
+                        float(row[0]) if math.isfinite(float(row[0])) else None,
+                        float(row[1]) if math.isfinite(float(row[1])) else None,
+                    ]
+                    for row in z_bounds
+                ],
+            },
+            "state_codes": {
+                "UNOBSERVED": 0,
+                "GROUND_SUPPORT": 1,
+                "NONGROUND_OCCUPIED": 2,
+                "UNKNOWN_REJECTED": 3,
+            },
+            "all_points_accounted": True,
+            "aos_used": False,
+            "authority": {
+                "visual_quality_accepted": False,
+                "navigation_or_safety_accepted": False,
+            },
+            "access": "read-only",
+        }
+    return json.dumps(
+        payload,
+        ensure_ascii=False,
+        sort_keys=True,
+        separators=(",", ":"),
+        allow_nan=False,
+    ).encode("utf-8")
+
+
+def _project_result(result: M49TgsFailClosedResult) -> dict[str, object]:
+    report = result.report
+    return {
+        "schema_version": RESULT_VIEW_SCHEMA,
+        "result_id": result.result_id,
+        "created_at_utc": result.manifest["created_at_utc"],
+        "source": copy.deepcopy(report["source"]),
+        "configuration": copy.deepcopy(report["configuration"]),
+        "method": copy.deepcopy(report["method"]),
+        "execution": copy.deepcopy(report["execution"]),
+        "metrics": copy.deepcopy(report["metrics"]),
+        "acceptance": copy.deepcopy(report["acceptance"]),
+        "decision": copy.deepcopy(report["decision"]),
+        "limitations": copy.deepcopy(report["limitations"]),
+        "authority": copy.deepcopy(report["authority"]),
+        "visual_review": copy.deepcopy(report["visual_review"]),
+        "ground_truth": False,
+        "access": "read-only",
+    }
+
+
+def _resolve_candidate(provider: RootProvider, result_id: str) -> Path:
+    if RESULT_ID.fullmatch(result_id) is None:
+        raise HTTPException(status_code=404, detail="M49 TGS result not found")
+    root = _configured_root(provider)
+    if root is None:
+        raise HTTPException(status_code=404, detail="M49 TGS result not found")
+    candidate = root / result_id
+    if candidate.is_symlink() or not candidate.is_dir():
+        raise HTTPException(status_code=404, detail="M49 TGS result not found")
+    resolved = candidate.resolve(strict=True)
+    if resolved.parent != root:
+        raise HTTPException(status_code=404, detail="M49 TGS result not found")
+    return resolved
+
+
+def _configured_root(provider: RootProvider) -> Path | None:
+    value = provider()
+    if value is None or value.is_symlink() or not value.is_dir():
+        return None
+    return value.resolve(strict=True)
+
+
+def _signature(candidate: Path) -> tuple[int, ...]:
+    result: list[int] = []
+    for name in (
+        "manifest.json",
+        "report.json",
+        "evidence.npz",
+        "worker-summary.json",
+        "input-manifest.json",
+        "tgs-timing.tsv",
+    ):
+        path = candidate / name
+        if path.is_symlink() or not path.is_file():
+            raise ValueError("M49 TGS artifact unavailable")
+        stat = path.stat()
+        result.extend((stat.st_size, stat.st_mtime_ns))
+    return tuple(result)
+
+
+def _evidence_signature(path: Path) -> tuple[int, int]:
+    stat = path.stat()
+    return stat.st_size, stat.st_mtime_ns
+
+
+def _empty_catalog(*, configured: bool) -> dict[str, object]:
+    return {
+        "schema_version": RESULT_CATALOG_SCHEMA,
+        "configured": configured,
+        "items": [],
+        "candidate_total": 0,
+        "invalid_total": 0,
+        "access": "read-only",
+    }
+
+
+__all__ = ["build_m49_tgs_fail_closed_router"]

src/k1link/web/m49_tgs_full_shadow_api.py

@@ -0,0 +1,314 @@
+"""Read-only API for the sealed complete TGS shadow."""
+
+from __future__ import annotations
+
+import copy
+import json
+import math
+import re
+from collections.abc import Callable
+from functools import lru_cache
+from pathlib import Path
+from typing import Final
+
+import numpy as np
+from fastapi import APIRouter, HTTPException, Query, Response
+
+from k1link.laboratory.m49_tgs_full_shadow import (
+    M49TgsFullShadowError,
+    M49TgsFullShadowResult,
+    PREFIX,
+    read_m49_tgs_full_shadow,
+)
+
+RootProvider = Callable[[], Path | None]
+RESULT_ID: Final = re.compile(rf"^{re.escape(PREFIX)}[a-f0-9]{{64}}$")
+ENDPOINT_ROOT: Final = "/api/v1/laboratory/m49/tgs-full-shadow"
+
+
+def build_m49_tgs_full_shadow_router(*, root_provider: RootProvider = lambda: None) -> APIRouter:
+    router = APIRouter(prefix=ENDPOINT_ROOT, tags=["laboratory"])
+
+    def sealed(result_id: str) -> M49TgsFullShadowResult:
+        root = _configured_root(root_provider)
+        if root is None or RESULT_ID.fullmatch(result_id) is None:
+            raise HTTPException(status_code=404, detail="M49 TGS full shadow not found")
+        candidate = root / result_id
+        if candidate.is_symlink() or not candidate.is_dir():
+            raise HTTPException(status_code=404, detail="M49 TGS full shadow not found")
+        try:
+            resolved = candidate.resolve(strict=True)
+            if resolved.parent != root:
+                raise ValueError("result escaped configured root")
+            return _read_cached(str(resolved), _signature(resolved))
+        except (M49TgsFullShadowError, OSError, ValueError):
+            raise HTTPException(status_code=404, detail="M49 TGS full shadow not found") from None
+
+    @router.get("/results")
+    def list_results(limit: int = Query(default=1, ge=1, le=10)) -> dict[str, object]:
+        root = _configured_root(root_provider)
+        if root is None:
+            return _catalog([], configured=False, invalid_total=0)
+        results: list[dict[str, object]] = []
+        invalid = 0
+        for candidate in sorted(root.iterdir()):
+            if not candidate.is_dir() or RESULT_ID.fullmatch(candidate.name) is None:
+                continue
+            try:
+                results.append(_project(_read_cached(str(candidate.resolve()), _signature(candidate))))
+            except (M49TgsFullShadowError, OSError, ValueError):
+                invalid += 1
+        results.sort(key=lambda row: (str(row["created_at_utc"]), str(row["result_id"])), reverse=True)
+        return _catalog(results[:limit], configured=True, invalid_total=invalid)
+
+    @router.get("/{result_id}")
+    def get_result(result_id: str) -> dict[str, object]:
+        return _project(sealed(result_id))
+
+    @router.get("/{result_id}/frames/{source_sequence}/spatial")
+    def get_spatial(result_id: str, source_sequence: int) -> Response:
+        if source_sequence < 0 or source_sequence >= 4489:
+            raise HTTPException(status_code=404, detail="M49 TGS full-shadow frame not found")
+        result = sealed(result_id)
+        try:
+            content = _frame_json_cached(
+                str(result.root), result_id, source_sequence, _evidence_signature(result.root)
+            )
+        except (OSError, ValueError, KeyError, json.JSONDecodeError):
+            raise HTTPException(status_code=503, detail="M49 TGS full-shadow spatial evidence failed verification") from None
+        return Response(
+            content=content,
+            media_type="application/json",
+            headers={"Cache-Control": "private, max-age=31536000, immutable", "X-Content-Type-Options": "nosniff"},
+        )
+
+    @router.get("/{result_id}/spatial/chunk")
+    def get_spatial_chunk(
+        result_id: str,
+        start: int = Query(ge=0, lt=4489),
+        count: int = Query(default=24, ge=1, le=24),
+    ) -> Response:
+        result = sealed(result_id)
+        bounded_count = min(count, 4489 - start)
+        try:
+            content = _chunk_json_cached(
+                str(result.root), result_id, start, bounded_count, _evidence_signature(result.root)
+            )
+        except (OSError, ValueError, KeyError, json.JSONDecodeError):
+            raise HTTPException(
+                status_code=503,
+                detail="M49 TGS full-shadow spatial chunk failed verification",
+            ) from None
+        return Response(
+            content=content,
+            media_type="application/json",
+            headers={
+                "Cache-Control": "private, max-age=31536000, immutable",
+                "X-Content-Type-Options": "nosniff",
+            },
+        )
+
+    return router
+
+
+@lru_cache(maxsize=4)
+def _read_cached(root: str, signature: tuple[int, ...]) -> M49TgsFullShadowResult:
+    del signature
+    return read_m49_tgs_full_shadow(Path(root))
+
+
+@lru_cache(maxsize=4)
+def _frames(root: str, signature: tuple[int, int]) -> tuple[dict[str, object], ...]:
+    del signature
+    path = Path(root) / "frames.ndjson"
+    rows = tuple(json.loads(line) for line in path.read_text(encoding="utf-8").splitlines())
+    if len(rows) != 4489:
+        raise ValueError("full-shadow frame catalog changed")
+    return rows
+
+
+@lru_cache(maxsize=96)
+def _frame_json_cached(
+    root: str, result_id: str, source_sequence: int, signature: tuple[int, ...]
+) -> bytes:
+    root_path = Path(root)
+    frame_signature = (signature[-2], signature[-1])
+    frame = _frames(root, frame_signature)[source_sequence]
+    centers = np.load(root_path / "costmap-cell-centers-xy-m.npy", mmap_mode="r", allow_pickle=False)
+    states_all = np.load(root_path / "costmap-states.npy", mmap_mode="r", allow_pickle=False)
+    z_all = np.load(root_path / "costmap-z-bounds-m.npy", mmap_mode="r", allow_pickle=False)
+    states = states_all[source_sequence]
+    z_bounds = z_all[source_sequence]
+    if (
+        centers.shape != (2244, 2)
+        or states_all.shape != (4489, 2244)
+        or z_all.shape != (4489, 2244, 2)
+        or not np.isfinite(centers).all()
+        or not np.isin(states, np.asarray([0, 1, 2, 3], dtype=np.uint8)).all()
+    ):
+        raise ValueError("full-shadow spatial shape changed")
+    payload = {
+        "schema_version": "missioncore.m49-tgs-full-shadow-spatial/v1",
+        "result_id": result_id,
+        "source_sequence": source_sequence,
+        "source_frame_index": frame["source_frame_index"],
+        "session_seconds": frame["session_seconds"],
+        "sample_available": frame["sample_available"],
+        "coordinate_frame": "map-gravity-local",
+        "costmap": {
+            "cell_size_m": 0.45,
+            "radius_m": 12.0,
+            "centers_xy_m": centers.astype(float).tolist(),
+            "states": states.astype(int).tolist(),
+            "z_bounds_m": [
+                [
+                    float(row[0]) if math.isfinite(float(row[0])) else None,
+                    float(row[1]) if math.isfinite(float(row[1])) else None,
+                ]
+                for row in z_bounds
+            ],
+        },
+        "metrics": copy.deepcopy(frame),
+        "state_codes": {"UNOBSERVED": 0, "GROUND_SUPPORT": 1, "NONGROUND_OCCUPIED": 2, "UNKNOWN_REJECTED": 3},
+        "aos_used": False,
+        "gpu_used": False,
+        "authority": {"navigation_or_safety_accepted": False, "visual_quality_accepted": False},
+        "access": "read-only",
+    }
+    return json.dumps(payload, ensure_ascii=False, sort_keys=True, separators=(",", ":"), allow_nan=False).encode("utf-8")
+
+
+@lru_cache(maxsize=16)
+def _chunk_json_cached(
+    root: str,
+    result_id: str,
+    start: int,
+    count: int,
+    signature: tuple[int, ...],
+) -> bytes:
+    root_path = Path(root)
+    frame_signature = (signature[-2], signature[-1])
+    frames = _frames(root, frame_signature)
+    centers = np.load(root_path / "costmap-cell-centers-xy-m.npy", mmap_mode="r", allow_pickle=False)
+    states_all = np.load(root_path / "costmap-states.npy", mmap_mode="r", allow_pickle=False)
+    z_all = np.load(root_path / "costmap-z-bounds-m.npy", mmap_mode="r", allow_pickle=False)
+    if (
+        centers.shape != (2244, 2)
+        or states_all.shape != (4489, 2244)
+        or z_all.shape != (4489, 2244, 2)
+        or not np.isfinite(centers).all()
+    ):
+        raise ValueError("full-shadow spatial chunk shape changed")
+    rows: list[dict[str, object]] = []
+    for source_sequence in range(start, start + count):
+        frame = frames[source_sequence]
+        states = states_all[source_sequence]
+        z_bounds = z_all[source_sequence]
+        if not np.isin(states, np.asarray([0, 1, 2, 3], dtype=np.uint8)).all():
+            raise ValueError("full-shadow spatial state changed")
+        rows.append(
+            {
+                "source_sequence": source_sequence,
+                "source_frame_index": frame["source_frame_index"],
+                "session_seconds": frame["session_seconds"],
+                "sample_available": frame["sample_available"],
+                "states": states.astype(int).tolist(),
+                "z_bounds_m": [
+                    [
+                        float(row[0]) if math.isfinite(float(row[0])) else None,
+                        float(row[1]) if math.isfinite(float(row[1])) else None,
+                    ]
+                    for row in z_bounds
+                ],
+                "metrics": copy.deepcopy(frame),
+            }
+        )
+    payload = {
+        "schema_version": "missioncore.m49-tgs-full-shadow-spatial-chunk/v1",
+        "result_id": result_id,
+        "start": start,
+        "count": count,
+        "coordinate_frame": "map-gravity-local",
+        "costmap": {
+            "cell_size_m": 0.45,
+            "radius_m": 12.0,
+            "centers_xy_m": centers.astype(float).tolist(),
+        },
+        "frames": rows,
+        "state_codes": {
+            "UNOBSERVED": 0,
+            "GROUND_SUPPORT": 1,
+            "NONGROUND_OCCUPIED": 2,
+            "UNKNOWN_REJECTED": 3,
+        },
+        "aos_used": False,
+        "gpu_used": False,
+        "authority": {
+            "navigation_or_safety_accepted": False,
+            "visual_quality_accepted": False,
+        },
+        "access": "read-only",
+    }
+    return json.dumps(
+        payload,
+        ensure_ascii=False,
+        sort_keys=True,
+        separators=(",", ":"),
+        allow_nan=False,
+    ).encode("utf-8")
+
+
+def _project(result: M49TgsFullShadowResult) -> dict[str, object]:
+    return {
+        **copy.deepcopy(result.report),
+        "schema_version": "missioncore.m49-tgs-full-shadow-view/v1",
+        "result_id": result.result_id,
+        "created_at_utc": result.manifest["created_at_utc"],
+        "ground_truth": False,
+        "access": "read-only",
+    }
+
+
+def _catalog(items: list[dict[str, object]], *, configured: bool, invalid_total: int) -> dict[str, object]:
+    return {
+        "schema_version": "missioncore.m49-tgs-full-shadow-catalog/v1",
+        "configured": configured,
+        "items": items,
+        "candidate_total": len(items) + invalid_total,
+        "invalid_total": invalid_total,
+        "access": "read-only",
+    }
+
+
+def _configured_root(provider: RootProvider) -> Path | None:
+    value = provider()
+    if value is None or value.is_symlink() or not value.is_dir():
+        return None
+    return value.resolve(strict=True)
+
+
+def _signature(root: Path) -> tuple[int, ...]:
+    values: list[int] = []
+    for name in ("manifest.json", "report.json", "worker-summary.json", *EVIDENCE_FILES):
+        path = root / name
+        if path.is_symlink() or not path.is_file():
+            raise ValueError("full-shadow artifact unavailable")
+        stat = path.stat()
+        values.extend((stat.st_size, stat.st_mtime_ns))
+    return tuple(values)
+
+
+def _evidence_signature(root: Path) -> tuple[int, ...]:
+    return _signature(root)
+
+
+EVIDENCE_FILES: Final = (
+    "costmap-cell-centers-xy-m.npy",
+    "costmap-cell-indices-xy.npy",
+    "costmap-states.npy",
+    "costmap-z-bounds-m.npy",
+    "frames.ndjson",
+)
+
+
+__all__ = ["build_m49_tgs_full_shadow_router"]

tests/test_laboratory_evidence_registry.py

@@ -127,7 +127,7 @@ def test_product_registry_declares_every_advanced_evidence_source() -> None:
         repository_root / "config" / "laboratories"
     )
 
-    assert len(registry.definitions) == 39
+    assert len(registry.definitions) == 42
     assert {item.work_id for item in registry.definitions} >= {
         "e31-source-binding",
         "e46j-raw-fisheye-realtime",
@@ -144,6 +144,8 @@ def test_product_registry_declares_every_advanced_evidence_source() -> None:
         "m48-static-occupancy-qualification",
         "m48s-fixed-class-detector",
         "m48t-risk-quality-temporal",
+        "m49-tgs-fail-closed-evidence",
+        "m49-tgs-full-shadow",
     }
     m48 = next(
         item for item in registry.definitions if item.work_id == "m48-object-centric-quality"

tests/test_laboratory_execution.py

@@ -101,6 +101,8 @@ def test_repository_registry_classifies_every_evidence_definition() -> None:
         "e47-semantic-slam-shadow",
         "m48s-fixed-class-detector",
         "m48t-risk-quality-temporal",
+        "m49-tgs-fail-closed-evidence",
+        "m49-tgs-full-shadow",
     }
     by_work_id = {row.work_id: row for row in execution.definitions}
     assert by_work_id["m48-small-static-passage-regression"].evidence_contract == (
@@ -118,6 +120,10 @@ def test_repository_registry_classifies_every_evidence_definition() -> None:
     assert by_work_id["m48s-fixed-class-detector"].isolation == "bounded-adapter"
     assert by_work_id["m48t-risk-quality-temporal"].lifecycle == "experimental"
     assert by_work_id["m48t-risk-quality-temporal"].isolation == "bounded-adapter"
+    assert by_work_id["m49-tgs-fail-closed-evidence"].lifecycle == "experimental"
+    assert by_work_id["m49-tgs-fail-closed-evidence"].isolation == "bounded-adapter"
+    assert by_work_id["m49-tgs-full-shadow"].lifecycle == "experimental"
+    assert by_work_id["m49-tgs-full-shadow"].isolation == "bounded-adapter"
     assert all(
         row.lifecycle == "canonical"
         for row in execution.definitions
@@ -126,6 +132,8 @@ def test_repository_registry_classifies_every_evidence_definition() -> None:
             "e47-semantic-slam-shadow",
             "m48s-fixed-class-detector",
             "m48t-risk-quality-temporal",
+            "m49-tgs-fail-closed-evidence",
+            "m49-tgs-full-shadow",
         }
     )
     assert len(execution.definitions) + len(execution.legacy_work_ids) == len(

tests/test_laboratory_value_review_registry.py

@@ -80,7 +80,7 @@ def test_product_value_review_registry_covers_reviewed_laboratory_families() ->
         root / "config" / "laboratory-value-review.json"
     )
 
-    assert len(registry.entries) == 38
+    assert len(registry.entries) == 40
     assert {entry.catalog_id for entry in registry.entries} >= {
         "e28-local-surface",
         "e46d-temporal-failure-audit",
@@ -91,4 +91,6 @@ def test_product_value_review_registry_covers_reviewed_laboratory_families() ->
         "m48-static-occupancy-qualification",
         "m48s-fixed-class-detector",
         "m48t-risk-quality-temporal",
+        "m49-tgs-fail-closed-evidence",
+        "m49-tgs-full-shadow",
     }

tests/test_m49_tgs_fail_closed_api.py

@@ -0,0 +1,165 @@
+from __future__ import annotations
+
+import hashlib
+import json
+from pathlib import Path
+
+import numpy as np
+from fastapi import FastAPI
+from fastapi.testclient import TestClient
+
+from k1link.laboratory.m49_tgs_fail_closed import (
+    M49_TGS_ANCHORS,
+    read_m49_tgs_fail_closed,
+    seal_m49_tgs_fail_closed,
+)
+from k1link.web.m49_tgs_fail_closed_api import build_m49_tgs_fail_closed_router
+
+REPOSITORY_ROOT = Path(__file__).resolve().parents[1]
+
+
+def _sha(path: Path) -> str:
+    return hashlib.sha256(path.read_bytes()).hexdigest()
+
+
+def _write_json(path: Path, value: object) -> None:
+    path.write_text(json.dumps(value, sort_keys=True), encoding="utf-8")
+
+
+def _source(root: Path) -> Path:
+    root.mkdir()
+    centers = np.zeros((2244, 2), dtype=np.float32)
+    arrays: dict[str, np.ndarray] = {"costmap_cell_centers_xy_m": centers}
+    summaries: list[dict[str, object]] = []
+    for profile in ("current_increment", "causal_rolling_1s"):
+        arrays[f"{profile}_points_xyz_m"] = np.asarray(
+            [[float(index), 0.0, 0.1] for index in range(10)], dtype=np.float32
+        )
+        arrays[f"{profile}_point_states"] = np.asarray([1, 2] * 5, dtype=np.uint8)
+        arrays[f"{profile}_point_offsets"] = np.arange(11, dtype=np.int64)
+        arrays[f"{profile}_costmap_states"] = np.zeros((10, 2244), dtype=np.uint8)
+        arrays[f"{profile}_costmap_ground_point_counts"] = np.zeros((10, 2244), dtype=np.int32)
+        arrays[f"{profile}_costmap_nonground_point_counts"] = np.zeros((10, 2244), dtype=np.int32)
+        arrays[f"{profile}_costmap_rejected_point_counts"] = np.zeros((10, 2244), dtype=np.int32)
+        arrays[f"{profile}_costmap_z_bounds_m"] = np.full((10, 2244, 2), np.nan, dtype=np.float32)
+        for slot, anchor in enumerate(M49_TGS_ANCHORS):
+            summaries.append(
+                {
+                    "profile_id": profile,
+                    "slot": slot,
+                    "anchor_frame_index": anchor,
+                    "point_count": 1,
+                    "ground_point_count": 1 if slot % 2 == 0 else 0,
+                    "nonground_point_count": 0 if slot % 2 == 0 else 1,
+                    "rejected_point_count": 0,
+                    "ground_cell_count": 0,
+                    "nonground_cell_count": 0,
+                    "rejected_cell_count": 0,
+                    "unobserved_cell_count": 2244,
+                    "all_points_accounted": True,
+                }
+            )
+    np.savez_compressed(root / "evidence.npz", **arrays)
+    records = [
+        {
+            "profile_id": profile,
+            "slot": slot,
+            "anchor_frame_index": anchor,
+        }
+        for profile in ("current_increment", "causal_rolling_1s")
+        for slot, anchor in enumerate(M49_TGS_ANCHORS)
+    ]
+    _write_json(
+        root / "input-manifest.json",
+        {
+            "schema_version": "missioncore.m49-tgs-fail-closed-input/v1",
+            "coordinate_frame": "map-gravity-local",
+            "records": records,
+        },
+    )
+    _write_json(
+        root / "worker-summary.json",
+        {
+            "schema_version": "missioncore.m49-tgs-worker-summary/v1",
+            "code_revision": "a" * 40,
+            "wall_seconds": 1.5,
+            "free_memory_gib_before": 42.0,
+            "canonical_triton_id": "b" * 64,
+            "canonical_triton_health": "healthy",
+            "all_eligible_points_accounted": True,
+            "aos_used": False,
+        },
+    )
+    timing = ["profile\tslot\twall_seconds\tmax_rss_kib"]
+    timing.extend(
+        f"{profile}\t{slot}\t0.02\t7000"
+        for profile in ("current_increment", "causal_rolling_1s")
+        for slot in range(10)
+    )
+    (root / "tgs-timing.tsv").write_text("\n".join(timing) + "\n", encoding="utf-8")
+    _write_json(
+        root / "result.json",
+        {
+            "schema_version": "missioncore.m49-tgs-fail-closed-evidence-result/v1",
+            "status": "passed",
+            "config_sha256": "c" * 64,
+            "source_pack_sha256": "d" * 64,
+            "input_manifest_sha256": _sha(root / "input-manifest.json"),
+            "evidence": {
+                "path": "evidence.npz",
+                "bytes": (root / "evidence.npz").stat().st_size,
+                "sha256": _sha(root / "evidence.npz"),
+            },
+            "costmap": {
+                "coordinate_frame": "map-gravity-local",
+                "cell_size_m": 0.45,
+                "radius_m": 12.0,
+                "cell_count": 2244,
+            },
+            "summary": {
+                "aos_used": False,
+                "all_eligible_points_accounted": True,
+                "primary_profile": "causal_rolling_1s",
+            },
+            "anchors": summaries,
+        },
+    )
+    return root
+
+
+def test_m49_tgs_seal_and_api_are_immutable_and_fail_closed(tmp_path: Path) -> None:
+    source = _source(tmp_path / "source")
+    destination = tmp_path / "results"
+    linked = "m4-threat-replay-" + "e" * 64
+    sealed = seal_m49_tgs_fail_closed(
+        source_root=source,
+        destination_root=destination,
+        profile_path=(
+            REPOSITORY_ROOT / "config" / "perception" / "m49-tgs-fail-closed-evidence-v1.json"
+        ),
+        linked_visual_result_id=linked,
+        created_at_utc="2026-08-26T17:30:00Z",
+    )
+    assert read_m49_tgs_fail_closed(sealed.root).result_id == sealed.result_id
+
+    app = FastAPI()
+    app.include_router(build_m49_tgs_fail_closed_router(root_provider=lambda: destination))
+    client = TestClient(app)
+    catalog = client.get("/api/v1/laboratory/m49/tgs-fail-closed/results")
+    assert catalog.status_code == 200
+    assert catalog.json()["items"][0]["result_id"] == sealed.result_id
+    anchors = client.get(f"/api/v1/laboratory/m49/tgs-fail-closed/{sealed.result_id}/anchors")
+    assert anchors.status_code == 200
+    assert anchors.json()["anchor_count"] == 10
+    spatial = client.get(
+        f"/api/v1/laboratory/m49/tgs-fail-closed/{sealed.result_id}/anchors/171/spatial"
+    )
+    assert spatial.status_code == 200
+    assert spatial.json()["point_states"] == [1]
+    assert spatial.json()["costmap"]["states"] == [0] * 2244
+    assert spatial.json()["authority"]["navigation_or_safety_accepted"] is False
+
+    (sealed.root / "worker-summary.json").write_text("{}", encoding="utf-8")
+    assert (
+        client.get(f"/api/v1/laboratory/m49/tgs-fail-closed/{sealed.result_id}").status_code == 404
+    )

tests/test_m49_tgs_full_shadow.py

@@ -0,0 +1,178 @@
+from __future__ import annotations
+
+import hashlib
+import importlib.util
+import json
+import sys
+from pathlib import Path
+
+import numpy as np
+
+from k1link.laboratory.m49_tgs_full_shadow import seal_m49_tgs_full_shadow
+from k1link.web import m49_tgs_full_shadow_api as full_shadow_api
+
+REPOSITORY_ROOT = Path(__file__).resolve().parents[1]
+WORKER_ROOT = REPOSITORY_ROOT / "experiments" / "perception" / "worker" / "m49_t3_travel"
+sys.path.insert(0, str(WORKER_ROOT))
+
+EVIDENCE_SPEC = importlib.util.spec_from_file_location(
+    "m49_tgs_full_shadow_evidence", WORKER_ROOT / "build_tgs_full_shadow_evidence.py"
+)
+assert EVIDENCE_SPEC and EVIDENCE_SPEC.loader
+EVIDENCE = importlib.util.module_from_spec(EVIDENCE_SPEC)
+EVIDENCE_SPEC.loader.exec_module(EVIDENCE)
+
+ARTIFACT_SPEC = importlib.util.spec_from_file_location(
+    "m49_tgs_full_shadow_artifact",
+    REPOSITORY_ROOT / "scripts" / "build_m49_tgs_full_shadow_worker_artifact.py",
+)
+assert ARTIFACT_SPEC and ARTIFACT_SPEC.loader
+ARTIFACT = importlib.util.module_from_spec(ARTIFACT_SPEC)
+ARTIFACT_SPEC.loader.exec_module(ARTIFACT)
+
+
+def test_full_shadow_exact_multiset_and_costmap_priority() -> None:
+    native = np.asarray(
+        [[2.0, 0.0, 0.0, 0.0], [2.0, 0.0, 0.0, 0.0], [3.0, 0.0, 1.0, 0.0], [4.0, 0.0, 2.0, 0.0]],
+        dtype=np.float32,
+    )
+    points, states = EVIDENCE.classify_exact_input(native, native[[0]], native[[1, 2]])
+    assert sorted(states.tolist()) == [1, 2, 2, 3]
+    grid = EVIDENCE.costmap_grid(12.0, 0.45)
+    cell_states, _ = EVIDENCE.rasterize(points, states, grid, 0.45)
+    assert np.count_nonzero(cell_states == 2) == 2
+
+
+def test_full_shadow_worker_artifact_is_deterministic_and_cpu_only(tmp_path: Path) -> None:
+    revision = "a" * 40
+    first = ARTIFACT.build("m49-tgs-full-test", tmp_path / "one", revision=revision)
+    second = ARTIFACT.build("m49-tgs-full-test", tmp_path / "two", revision=revision)
+    assert first["sha256"] == second["sha256"]
+    import tarfile
+
+    with tarfile.open(first["artifact"], "r:gz") as archive:
+        release = archive.extractfile("payload/release.json")
+        runner = archive.extractfile("payload/Invoke-M49TgsFullShadow.ps1")
+        assert release is not None and runner is not None
+        release_text = release.read().decode()
+        runner_text = runner.read().decode()
+    assert '"candidate_id": "travel-tgs-full-shadow"' in release_text
+    assert "--gpus" not in runner_text
+    assert "gpu_requested = $false" in runner_text
+
+
+def test_full_shadow_seal_binds_visual_and_semantic_timelines(tmp_path: Path) -> None:
+    source = tmp_path / "worker"
+    source.mkdir()
+    files: dict[str, dict[str, object]] = {}
+    for name in (
+        "costmap-cell-centers-xy-m.npy",
+        "costmap-cell-indices-xy.npy",
+        "costmap-states.npy",
+        "costmap-z-bounds-m.npy",
+        "frames.ndjson",
+    ):
+        payload = f"sealed:{name}\n".encode()
+        (source / name).write_bytes(payload)
+        files[name] = {
+            "bytes": len(payload),
+            "sha256": hashlib.sha256(payload).hexdigest(),
+        }
+    worker = {
+        "schema_version": "missioncore.m49-tgs-full-shadow-result/v1",
+        "status": "passed",
+        "source_pack_sha256": "a" * 64,
+        "input_manifest_sha256": "b" * 64,
+        "config_sha256": "c" * 64,
+        "timeline": {
+            "frame_count": 4489,
+            "available_lidar_frame_count": 3928,
+            "missing_lidar_frame_count": 561,
+        },
+        "point_accounting": {"unaccounted": 0},
+        "costmap": {"cell_size_m": 0.45, "radius_m": 12.0},
+        "performance": {},
+        "acceptance": {},
+        "files": files,
+    }
+    (source / "result.json").write_text(json.dumps(worker), encoding="utf-8")
+    (source / "worker-summary.json").write_text(json.dumps({
+        "schema_version": "missioncore.m49-tgs-full-shadow-worker-summary/v1",
+        "gpu_requested": False,
+        "aos_used": False,
+        "all_timeline_frames_accounted": True,
+        "all_eligible_points_accounted": True,
+        "canonical_triton_health": "healthy",
+        "canonical_triton_id": "triton",
+        "wall_seconds": 1.0,
+    }), encoding="utf-8")
+    profile = tmp_path / "profile.json"
+    profile.write_text(json.dumps({
+        "schema_version": "missioncore.m49-tgs-full-shadow-profile/v1",
+        "profile_id": "test",
+        "profile": {"history_seconds": 1.0},
+        "costmap": {"state_priority": ["NONGROUND_OCCUPIED"]},
+        "state_codes": {"UNOBSERVED": 0},
+    }), encoding="utf-8")
+    visual = "m4-threat-replay-" + "d" * 64
+    semantic = "e47-semantic-slam-" + "e" * 64
+
+    sealed = seal_m49_tgs_full_shadow(
+        source_root=source,
+        destination_root=tmp_path / "results",
+        profile_path=profile,
+        linked_visual_result_id=visual,
+        linked_semantic_result_id=semantic,
+        created_at_utc="2026-08-26T20:27:19Z",
+    )
+
+    assert sealed.report["source"]["linked_visual_result_id"] == visual
+    assert sealed.report["source"]["linked_semantic_result_id"] == semantic
+    assert sealed.manifest["identity"]["linked_semantic_result_id"] == semantic
+
+
+def test_full_shadow_chunk_contract_keeps_missing_lidar_unobserved(
+    monkeypatch,
+    tmp_path: Path,
+) -> None:
+    frames = tuple(
+        {
+            "source_frame_index": index,
+            "session_seconds": index / 10,
+            "sample_available": index != 1,
+            "eligible_point_count": 0 if index == 1 else 1,
+            "ground_point_count": 0,
+            "nonground_point_count": 0 if index == 1 else 1,
+            "rejected_point_count": 0,
+            "occupied_cell_count": 0 if index == 1 else 1,
+        }
+        for index in range(4489)
+    )
+    centers = np.zeros((2244, 2), dtype=np.float32)
+    states = np.broadcast_to(np.zeros((1, 2244), dtype=np.uint8), (4489, 2244))
+    z_bounds = np.broadcast_to(
+        np.full((1, 2244, 2), np.nan, dtype=np.float32),
+        (4489, 2244, 2),
+    )
+
+    monkeypatch.setattr(full_shadow_api, "_frames", lambda *_args: frames)
+    monkeypatch.setattr(
+        full_shadow_api.np,
+        "load",
+        lambda path, **_kwargs: (
+            centers if "centers" in str(path) else states if "states" in str(path) else z_bounds
+        ),
+    )
+    content = full_shadow_api._chunk_json_cached.__wrapped__(
+        str(tmp_path),
+        "m49-tgs-full-shadow-" + "a" * 64,
+        0,
+        2,
+        (0, 0),
+    )
+    payload = json.loads(content)
+    assert payload["schema_version"] == "missioncore.m49-tgs-full-shadow-spatial-chunk/v1"
+    assert payload["count"] == 2
+    assert len(payload["costmap"]["centers_xy_m"]) == 2244
+    assert payload["frames"][1]["sample_available"] is False
+    assert set(payload["frames"][1]["states"]) == {0}