Massive Visual Research Swarm Integration Blueprint V2

00. Executive integration ruling

The recovered swarm material should not be rebuilt as a separate SREIC swarm platform, a second assistant, or a new central orchestrator. Its value is that it supplies a rich domain operating model, swarm decomposition patterns, PMLA/context concepts, cross-domain research patterns, economic feedback ideas, and a highly visual operator experience. The modern Neuruh estate already owns the infrastructure that should execute those ideas.

What is proven vs proposed

Source claim: classifications above are the source document's own rulings, not independently verified facts.

01. Source basis and anti-loss custody

This integration is grounded in both the recovered donor estate and current Neuruh source/canon. The original source binaries are preserved separately; Git stores their exact custody manifest and the derived architecture rather than pretending binary presence proves runtime behavior. Derived companion blueprints already preserved include the SREIC master blueprint and the Research Swarm recovery/handoff. This document converges those with the current Copilot/Organism/Worlds/visual topology rather than replacing them.

Recovered sources (hashed)

Anti-loss rule (source): Originals remain immutable source artifacts. Every derivative records source filename + SHA. Git holds the source-custody manifest. The user-facing archive bundles originals and derivatives together so this work cannot disappear into chat history.

ZIP code / artifact harvest

React + TypeScript + Vite dashboard and compiled dist. Three.js / React Three Fiber / Drei visual components. ActivityFeed, LegionCard, LegionIcon3D, PerformanceChart, StatCard, USMap. Deck imagery and a visual architecture plate. Many displayed metrics/events are simulated or hardcoded; they must be replaced with real Neuruh read models before use.

Source claim: the recovered visual donor is useful as aesthetic/interaction lineage only. The old domination/legion labels and metrics are not current runtime truth.

02. Existing Neuruh infrastructure we should reuse

The current topology already contains the exact separation the donor was reaching toward: shared substrate, bounded roles, event-driven work, authority, independent court, durable custody, outcome learning, and domain Worlds. The integration job is to expose these as one coherent visual operating system.

03. Copilot: make Presence the universal operator shell

Copilot should be a role-aware skin over Presence/LARIC and Cockpit. Current source already demonstrates that the same Presence runtime can present different tenant/operator skins without code forks. That is the leverage point.

Copilot UX law

• The conversation follows the selected World and active mission automatically.
• Every answer can expose the evidence handles behind its state.
• Do it compiles a mission or proposal; it does not bypass authority.
• The shell can switch skins without creating duplicate runtime state.
• A provider/model may change; mission identity, receipts and authority do not.

Copilot skins explorer

Copilot request flow (source): COPILOT REQUEST → context compiler (World + actor + mission + evidence frontier) → BOB / Founder Runtime mission compile → Governance intent classification → Workers / Switchboard → receipts + visible trace → Copilot explains result from durable state.

04. Worlds: domain-scale intelligence without duplicated cores

A World is a manifest that composes shared organs and domain-specific contracts. Real Estate / SREIC is World 1. A World should be reproducible from references to shared organs plus domain overlays. A World may have distinct sources, recipes, policies, projections, terminology and product promises. It should not carry another generic router, memory system, governance engine or receipt format.

Candidate World composition contract

neuruh.world.v1 fields: world_id, name, thesis, owner, lifecycle, source_manifest_refs[], organ_refs[], capability_refs[], source_adapters[], recipes[], policy_overlay, allowed_action_classes[], projection_routes[], copilot_skin, spatial_adapter, evidence_contract, event_contract, outcome_contract, source_refs[], runtime_refs[], unverified[], acceptance_evidence[].

Convergence gate (source): This is a contract candidate, not permission to create a fourth World schema. JGI/LARIC first compare current World registry, public WORLD_MANIFEST, UPL/World Emitter contracts and Cockpit world-registry. Extend the strongest owner; create only irreducible residue.

World roadmap explorer

05. World 1: SREIC / Real Estate becomes the flagship visual World

The SREIC donor is most valuable when treated as a domain operating blueprint layered over the existing real-estate stack. The system can preserve the recognizable acquisition/capital/development/exit/intelligence mental model while replacing unsafe or fictional execution with current governed owners.

The complete Real Estate World loop

SEE THE MARKET → FIND HIDDEN VALUE → PROVE THE SIGNAL → RESOLVE PROPERTY + ENTITY → CREATE OPPORTUNITY → RUN MULTI-STRATEGY UNDERWRITING → RANK ECONOMIC ROUTES → GOVERN ACTION → EXECUTE / DELIVER → CREATE / OPERATE / EXIT VALUE → RECORD OUTCOME → ATTRIBUTE + CALIBRATE → LEARN → REPEAT

What the UI should feel like

• A map/terrain of real properties and opportunity clusters, not a generic CRM table.
• Live mission traces branch from an opportunity into research, evidence, underwriting, contact, capital and disposition work.
• The old Legion cards survive as group-level views, but counts come from active worker leases/queues.
• Every property/opportunity can expand into an evidence graph, recipe decisions, authority gates and economic alternatives.
• One click switches from geographic World view to Galaxy graph to mission Swarm trace without leaving the same case.

Historical legions mapped to modern visual groups

Source claim: legion names are historical donor labels. Their modern value is capability composition, not permanent units.

06. Swarm runtime and count discipline

Do not run thousands of permanent model personalities. Treat swarm size as an elastic execution plan. BOB compiles the work. Workers lease bounded units. Role/capability registries describe what each worker can do. Switchboard preserves transitions. Completed branches seal receipts and court as required.

Worker visualization grammar

• Node shape = role/capability class. Border = World. Fill state = queued / running / waiting / court / blocked / complete / defect / unverified.
• Edges = actual parent/child mission or receipt transition. Pulse/particle = real event, not animation timer.
• Click = mission capsule, evidence frontier, source SHA/tree, latest receipt, authority, next action, do-not-rerun.

Calculator 1: Swarm count discipline

Enter live worker-lease counts from Workers/Switchboard state. The calculator keeps the three count classes separate and shows live leases only as a small fraction of the historical 5,150 donor capacity figure.

07. Massive visual command system

The visual system should be a read-model compiler: normalize current source/runtime/receipt/outcome state into view models, then let multiple renderers consume it. The read model is allowed to simplify for comprehension; it is not allowed to manufacture status. One read-only projection compiler feeds 2D Cockpit, 3D Spatial Host and Copilot.

Top-level surfaces explorer

Canonical visual truth rules

1. Every metric has a source and as-of timestamp.
2. Every state has a lifecycle vocabulary and strongest evidence class.
3. Every animation is triggered by an observed event/transition.
4. Every success/failure is traceable to a receipt, court or provider state.
5. UNKNOWN is visible, never coerced into zero/offline/healthy.
6. Candidate vs merged vs deployed vs healthy vs revenue remain separate.
7. Historical donor visuals are labeled historical.

Spatial modes

08. Event and projection contracts

The fastest route to a living system is not a new frontend first. It is a stable, read-only event/view vocabulary that lets Cockpit, Spatial Host and Copilot tell the same story from the same evidence.

Candidate visual event envelope

neuruh.visual-event.v1: event_id, occurred_at, source_system, mission_id, parent_mission_id, world_id, object_type, object_id, event_type, previous_state, new_state, truth_class, receipt_ref, evidence_refs[], authority_ref, visual_hint (non-authoritative).

Priority event vocabulary explorer

Read-model strategy

09. Opportunity is the cross-World organizing object

The biggest architecture improvement is to stop organizing the visual system around chats or generic tasks. An Opportunity can carry a source-backed thesis from signal through strategy, mission, outcome and learning across any World.

Opportunity chain (source): SIGNAL → EVIDENCE → ENTITY → OPPORTUNITY → ALTERNATIVES → EXPECTED VALUE + UNCERTAINTY → MISSION → GOVERNED ACTION → OUTCOME → ATTRIBUTION → CALIBRATION

Opportunity panel

Economic circuit-breaker integration

The recovered Economic Guardian is best implemented as a converged economic evidence/control seam, not another agent. A cost/outcome ledger can feed RICHIE and Governance while Cockpit visualizes budget and risk. Cost events by mission/tool/provider. Budget and threshold references. Attributable outcome/revenue/value with evidence class. Rolling cost-per-result / ROI where mathematically valid. HALT / ESCALATE recommendation; Governance decides. No synthetic revenue or expected value rendered as cash received.

Calculator 2: Economic circuit-breaker (cost-per-result and ROI)

10. Learning and Research Swarm complete the loop

Research Swarm and the learning spine serve different roles. Research Swarm discovers and proposes. Outcome/calibration learns from what actually happened. The connection is where the system begins to improve without becoming self-authorizing.

Research Copilot view

• Inbox of new sources/donors/research feeds.
• Source SHA, provenance, date and scope.
• Current owner match / duplicate / conflict / unique residue.
• Concept lineage map into current organs and Worlds.
• Candidate experiments and economic leverage score.
• Build/court/custody status if residue becomes code.
• Rejected anti-patterns retained for defense and lineage.

The learning loop

Exact chain (V2, section 25): Outcome → 017 calibration ledger → 019 Learning Update Proposal (LUP) → N-Lab evaluation → canary → Promotion Gate → revision receipt. A founder override should create an evidence-bearing override/decision record that may feed a Learning Update Proposal. It must not silently rewrite policy.

11. Harvest the donor UI without importing its false state

The recovered ZIP already gives us a visual vocabulary. The shortest path is a component-harvest mission against Cockpit/Spatial Host: preserve distinctive interactions and 3D language, throw away simulated metrics and unsafe commands, and bind every surviving component to current read models.

Donor component harvest explorer

Visual signature to preserve (source): Biological-architectural depth, spatial constellation, map/radar logic, 3D role glyphs, live traces, dark operational depth, restrained earned illumination. The UI should feel like an organism being observed, not a SaaS admin template. This demo renders that signature in the light editorial document style required for this build, with the same spatial grammar in SVG figures.

12. Spatial Host: the massive 3D organism

The historical Neuruh World Engine already articulated the right abstraction: data adapter to spatial layout to render layer. Current Cockpit source says reuse the existing spatial host/R3F rather than rebuild it. The 3D system should therefore consume the same read model as the 2D Cockpit.

Camera / interaction model

• Wide: whole organism and Worlds.
• Vertical: selected World and its shared-organ dependencies.
• Mid: Opportunity or mission system.
• Close: evidence/court/receipt object.
• Core: Governance gate / decision trace.
• Clicking an object opens proof before action controls.

13. Build program: the order that avoids a giant rewrite

Do not start with

• a new giant frontend repo;
• a new swarm server;
• 5,150 always-on agents;
• a new database for Worlds;
• a new assistant called Copilot with its own state;
• a new 3D engine;
• World Emitter deployment before its current UNWIRED state is resolved.

Phases 0 to 10

14. Repository ownership and implementation placement

Git custody already updated in this pass (source claim)

• NeuruhAI/neuruh-vault PR #8 remains the Research Swarm candidate lane.
• Added 08_MASSIVE_VISUAL_WORLD_INTEGRATION_BLUEPRINT.md.
• Added 09_SOURCE_ARTIFACT_CUSTODY.md.
• Added 10_WORLD_VISUAL_CONTRACT_CANDIDATES.md.
• Updated Research Swarm README to index visual integration and source custody.

Notion / canon update plan

The correct Notion approach is small cross-links/amendments into existing owners, not a new parallel notebook. The connector was asked to insert the first amendment during this pass but returned an internal write failure twice, so this document does not claim Notion was changed. Current Notion status: WRITE NOT PROVEN. Two targeted insert attempts returned connector internal errors. Git/Vault is therefore the durable current custody. Retry Notion only as a small projection update when the connector/workspace accepts writes; do not recreate the material manually in a new page.

Canonical owners needing amendment: Execution Topology Canon (Research Swarm/Copilot/Worlds placement + Git PR pointer); Chapter 22 Worlds (World composition + Copilot/Research Swarm relationship); Chapter 37 Visual Atlas (candidate new overlay views without falsifying the existing 25-plate canon); Chapter 39 Source/Runtime/Receipt (add Research Swarm and visual projection rows after contract ownership is ruled); Neuruh World Engine IP entry (record historical donor to Spatial Host/World projection ruling).

15. Acceptance courts for the massive visual system

16. Exact next missions

Founder interruption policy

• Do not ask Jeramie to manually shuttle context between roles.
• Interrupt only for consequential authority, unresolved product/economic choice, external account/permission block, or true source collision.
• VIBIN/Copilot should summarize exact decision needed, evidence, options and consequence.

Detailed handoff: read first

Research Swarm README and files 08-10 in Vault PR #8. Current Execution Topology Canon. Chapter 22 Worlds. Chapter 37 Visual Atlas. Chapter 39 Source to Runtime to Receipt to Consumer. Cockpit presence-runtime world-registry and OWNERSHIP.md. Current Workers/Switchboard contracts before inventing visual events.

Stop conditions

• If a current contract already owns a field/model, adapt to it; stop schema invention.
• If source/runtime identity conflicts, classify CONFLICTED_CUSTODY and stop mutation.
• If a visual metric cannot be traced, omit it.
• If World Emitter is still unwired, do not stub it as live.
• If a skin requires new authority, the design is wrong; route through Governance.

First build-session objective (source): Produce a field-level comparison of the donor/Kimi/SREIC visual and World requirements against the current Cockpit neuruh.spatial-organism.v4, neuruh.operating-model.v1, Workers/Switchboard mission/receipt contracts, Opportunity Case, N-Cell/effective-state and World Emitter specs. The output is an adapter/extension plan, not a new schema by default. Do not write UI first. The first accepted code change should be a read-only adapter/contract extension into Cockpit, not a new visual shell.

17. V2 gap closure: the complete organism loop

Correction (source): the V1 direction was sound, but it treated several already-built Neuruh subsystems as future concepts and compressed major donor residue. V2 makes current owners the foundation. The first blueprint was missing substantial integration depth. The problem was not simply more features. It failed to distinguish three different categories: systems already built and therefore requiring extension; existing lineages that needed binding; and historical donor concepts that still require a present-owner court. This supplement corrects all three.

Gap classes

V2 completion ruling (source): The larger Neuruh system should not be described as a swarm dashboard attached to an organism. The swarm is one elastic execution expression inside a much larger evidence/state/capability/World/learning architecture. The visual product is Cockpit/Spatial Organism enriched by all of those existing contracts.

18. Cockpit, World Factory, N-Cell and UPL

Cockpit / Spatial Organism is already the visual operating substrate

V1 proposed a read-model/event layer as if it had to be invented. Current Cockpit source already contains most of it. This changes the implementation program materially: the job is to extend and enrich an existing truth-backed spatial dataset, not build another visualization state service. Spatial Organism V4 and the older rich organism visualization are implementation lineage, not mockups. Founder Command Stage A/B already carried focus/mission layout and mobile/compact behavior. The known stripped/minimal view was a regression/partial projection; the canonical recovery law is DO NOT BUILD COCKPIT V2 and DO NOT BUILD ANOTHER GALAXY. Existing adapters already cover Galaxy, Continuity, Qdrant, Vault, DATUM, LandOS, UnderwriteX, Governance and Laric mission context.

V2 build rule: Before adding any visual-event schema, compare the donor field to organism-dataset.js, operating-model.js, flow trace, mission projection, receipts and current World objects. Extend only what is truly absent.

World Factory is already split across Factory + World Emitter + Galaxy

V1 described World Emitter as a later wiring task. That understated what is already implemented. The current World Emitter is a real multi-tenant, white-label, Notion/Vault-driven world projection engine with seed, refresh, batch and publish-queue behavior. Factory already supplies the governed build/evaluation lane. Real Estate/SREIC is not only a Cockpit view. It should have a World spec, capability references, policy overlay, product/client projections and World-level outcome contract. Once the same machinery creates a second World without copying the substrate, Neuruh has a real generality proof.

World Factory tape (source): GALAXY / CANONICAL NODE → WORLD SPEC (VAULT) → FACTORY CONTRACT / REUSE / EVALUATION → WORLD EMITTER: seed / tenantize / theme / build → LOCAL PREVIEW → SENTINEL + PUBLIC PROOF + POLICY → PROMOTION AUTHORITY → CLIENT / PUBLIC WORLD

N-Cell, UPL and the effective-state spine

The first blueprint largely omitted the state/revision machinery that makes a living visual system trustworthy. These components should be first-class inputs to Cockpit rather than invisible backend details. New visual layers this unlocks: CURRENT vs CANDIDATE vs EFFECTIVE vs RUNTIME overlays; at-time reconstruction of topology and mission state; drift heatmap (what the system expected vs what providers actually show); why-this-state drawer (evidence to reconciliation to revision to effective state); rollback and supersession lineage visible without flattening history.

19. Capability genome and signal grid

Capability Registry + Atomic Registry are the reusable genome

Historical legions should not survive as permanent infrastructure units. Their useful boundary is capability composition. Current registries give Neuruh a cleaner way to express this. Proton/Neutron/Electron should remain registry metadata describing logic, state/stability and external execution characteristics. It must not create a second orchestration or authority system.

Donor residue path (source): RESEARCH / DONOR RESIDUE → CAPABILITY DECOMPOSITION (ESTATE OFFICE) → CAPABILITY REGISTRY + ATOMIC CLASSIFICATION → WORLD REFERENCES CAPABILITY → MISSION REQUESTS CAPABILITY → WORKER LEASES PROVIDER / TOOL / MODEL → OUTCOME CALIBRATES CAPABILITY → FACTORY/N-LAB EVALUATES CANDIDATE CHANGE

Estate Office to Research Swarm: the donor recovery pipeline

Research Swarm should not become a second archaeology/custody system. Estate Office already owns technical/IP estate classification. Research Swarm begins after custody/identity or calls Estate Office when identity is unresolved. The donor AutonomousDirector contained a useful file/document intake capability: directory monitoring, hash dedupe, classification, entity/opportunity extraction and research routing. That should be implemented only as a bounded intake adapter feeding Estate Office/Research Swarm/Vault, not as a new central director.

Signal Intelligence Grid

The donor cross-domain signal scanner was not isolated invention. Neuruh already has substantial signal/discovery lineage. V2 joins those historical sources to the modern Evidence to Opportunity to Mission path rather than starting another signal database.

Signal path (source): RAW SOURCE → SOURCE / LAWFULNESS / FRESHNESS EVIDENCE → SIGNAL → ENTITY / IDENTITY → OPPORTUNITY GENOME / CASE → RECIPE ELIGIBILITY + UNKNOWN / NEXT-BEST-EVIDENCE → RICHIE PRIORITY → BOB MISSION → GOVERNED ACTION / OUTCOME / CALIBRATION

Calculator 3: Demand Radar score

Source formula (Table 46): score = fit x intent x trigger x timing x value x reachability - risk. Each factor is scored 0 to 10, so the maximum product is 1,000,000. Risk is subtracted in the same points. Weights must be calibrated against outcomes; an uncalibrated score is a research ranking, not a value claim.

20. Security, autonomy and reasoning

Security, Zero Trust and graduated autonomy

The donor security and autonomy layers were underrepresented. Their strongest modern value is not a separate Constitutional Guardian. It is a tighter security perimeter, input quarantine, tenant isolation and evidence-earned autonomy policy over the existing Governance stack.

Autonomy is a matrix, not a global level

Causal, counterfactual, uncertainty and strategic reasoning

The donor 26-gap and Foundational Intelligence sections contain cognition work that V1 almost entirely omitted. These should become research/evaluation capabilities, never an unbounded wisdom brain.

Psychology / persuasion boundary (source): Historical psychological profiling/persuasion concepts may be studied as decision or user-research inputs, but they do not justify deceptive, coercive or manipulative external actions. Governance and public-proof boundaries remain decisive.

Learning chain detail (Outcome to 017 to 019 to N-Lab)

21. Channels and public layers

Multimodal, mobile, field and physical-digital channels

The integrated system should not be desktop-only. Existing donor and Neuruh lineage supports a single mission/context identity moving across founder desktop, mobile, field evidence and external connectors.

Public, client and distribution layers

V1 centered founder Cockpit. The larger system also needs explicit boundaries for tenant/client experiences, public proof and developer/plugin distribution. One organism, multiple projections.

Real Estate / SREIC binds the full Opportunity and QuerySpec machinery

V1 listed the major real-estate organs but did not integrate the contracts that make World 1 operational. V2 treats these as core World state.

Modern SREIC navigation (source): MAP / SIGNAL FIELD → PROPERTY / ENTITY → OPPORTUNITY CASE → RECIPE ELIGIBILITY + WHY MATCHED + UNKNOWN → EVIDENCE / NEXT-BEST-EVIDENCE → ACQUIRE | CAPITAL | DEVELOP | OPERATE | EXIT | REJECT → MISSION DAG / AUTHORITY / EXECUTION → DISPOSITION / DELIVERY / OUTCOME / CALIBRATION

22. What remains donor-only, research, held or rejected

Preserve without activating. Rejected or held concepts remain in donor provenance because they explain architecture history and can serve red-team/security analysis. Preservation is not adoption.

Visual Genome: Kimi was only one donor

V1 overfocused the newly recovered Kimi dashboard. Current canon already identifies a much larger visual donor genome. V2 requires one component-harvest court across all of it.

Visual truth court (source): A component can be visually aggressive while remaining epistemically strict: no hardcoded revenue, random activity, fake agent counts, invented health, or animation-triggered state. The visual engine may dramatize an observed transition; it may never create the transition.

23. Revised implementation sequence: build less, bind more

First implementation mission after V2 (source): Produce a field-level comparison of the donor/Kimi/SREIC visual and World requirements against the current Cockpit neuruh.spatial-organism.v4, neuruh.operating-model.v1, Workers/Switchboard mission/receipt contracts, Opportunity Case, N-Cell/effective-state and World Emitter specs. The output is an adapter/extension plan, not a new schema by default.

24. Appendices A to D

Appendix A: Durable custody map

Appendix B: Research Swarm Git package

01-CANON/research-swarm/README.md; 01_DONOR_RECOVERY_CANON.md; 02_SYSTEM_INTEGRATION_TOPOLOGY.md; 03_CAPABILITY_DISPOSITION_MATRIX.md; 04_DETAILED_HANDOFF_2026-09-08.md; 05_EXECUTION_BACKLOG.md; 06_SOURCE_INDEX.md; 07_SECURITY_AND_NONADOPTION_RULINGS.md; 08_MASSIVE_VISUAL_WORLD_INTEGRATION_BLUEPRINT.md; 09_SOURCE_ARTIFACT_CUSTODY.md; 10_WORLD_VISUAL_CONTRACT_CANDIDATES.md.

Appendix C: Final architectural sentence

Appendix D: V2 custody additions

• 01-CANON/research-swarm/11_EXISTING_SYSTEM_INTEGRATION_GAP_CLOSURE.md
• 01-CANON/research-swarm/12_FULL_DONOR_TO_NEURUH_OWNER_CROSSWALK.md
• V2 downloadable DOCX/PDF adds this complete existing-system reconciliation to the original blueprint.

Mission record (source): MISSION: NEURUH-RESEARCH-SWARM-VISUAL-WORLD-INTEGRATION-20260908 | STATUS: CANDIDATE ARCHITECTURE / GIT-CUSTODIED | PRIMARY GIT LANE: NeuruhAI/neuruh-vault PR #8 | RUNTIME MUTATION: NONE AUTHORIZED BY THIS BLUEPRINT | NOTION WRITE: NOT PROVEN (connector internal failure) | SOURCE ORIGINALS: HASHED + PRESERVED | CURRENT IMPLEMENTATION CANDIDATES: AXON PR #6; Atomic Registry PR #1. No merge, deploy, runtime mutation or new authority is implied by this document. Current provider/runtime evidence still outranks architecture prose.

Related Neuruh surfaces: Neuruh Hub (master map of live builds) and the Real Estate Agent system hub at real-estate-agent-c03.pages.dev, the World 1 SREIC sibling surface.

T1. Try-it walkthrough

Walk the blueprint the way an operator would: ruling first, counts second, then World, event, and court. Use the buttons to step through, or jump to any section from the contents.

T2. Blueprint intake, completeness and share link

Draft a blueprint intake the way the source requires: mission identity, World, thesis, source custody, owner, evidence, authority and outcome contract. The completeness meter counts only the eight required fields below. Nothing here writes to any repo, deploys anything, or grants authority. It is a planning record you can share by link.

T3. How it works / What this is not

T4. Honest gaps

• Demo, not runtime. Every count, event and metric on this page is either source text or a sample you enter. Nothing here reads Cockpit, Workers, Switchboard, Galaxy or any live Neuruh system.
• Donor visuals not reproduced. The source DOCX embeds 5 donor images (deck imagery and a visual architecture plate). They are not embedded here; the figures are original SVG diagrams drawn from the source's own descriptions.
• Notion write not proven. The source records two failed connector write attempts; this demo claims no Notion change.
• World Emitter unwired. The source states World Emitter is explicitly unwired in current Cockpit ownership; this demo draws it only as a future path.
• End-to-end auto-Sentinel/result-ingest not live-proven. The source classifies it NOT LIVE-PROVEN; this demo does not claim otherwise.
• Calculators are arithmetic only. Demand Radar weights are uncalibrated in the source, and ROI is valid only where attributable value has an evidence class. The economic verdict is a recommendation; Governance decides.
• Layout not eyeballed in a real browser. Verified in jsdom only; visual polish in a real browser is unconfirmed.
• Source claims labeled as source claims. Ownership, lineage and current-source statements come from the September 8, 2026 document and were not re-verified against live repos for this build.