Multi-model analysis
Planned model perspectives propose and challenge interpretations. Agreement is not independent proof; shared errors and unresolved objections must remain visible.
AI INTELLIGENCE · HUMAN CONTROL
DEVELOPMENTAL
What is AUREX / OX?
AUREX is the AI intelligence system being built to turn complex information into evidence-backed decisions.
The design combines multi-model analysis, source checking and governed workflow proposals. OX makes the process and its limits visible; people retain control of consequential decisions. See the NVIDIA research map.
Supply records
Define the question and the permitted sources in scope.
Check discrepancies
Challenge interpretations, check source support and flag missing evidence.
Human review
Review candidate findings; no automatic financial decision.
Technical status & limitations
AUREX is VALEXII’s evidence-governed AI orchestration and workflow system. OX refers to its operational execution layer—the part designed to coordinate bounded jobs, apply authorization checks, and record outcomes with evidence. Together, AUREX / OX connects reasoning and governed execution while keeping humans in control.
OX is a layer designation, not a formally established letter-by-letter acronym. No expanded acronym is asserted for AUREX either.
Evidence → Qualification → Reality → Authority → Effect
Gather observations, check their quality, establish what they support, confirm permission, then record the bounded outcome. Independent verification requires separate evidence; the system must never be its own sole witness. Unknown, missing, stale or unresolved evidence fails closed: HOLD for human review, never PASS. UNKNOWN ≠ SAFE.
Architectural intent describes what we are building. Demo means the fixed synthetic interaction below. Verified live capability requires independent supporting evidence and remains NOT VERIFIED here. Connection checks report transport only; this page does not establish a verified website/backend integration, production readiness or native AI inference.
Development status: sample review available; customer audits and AI models are in development.
Capabilities & applications / Developmental
More perspectives. Better-supported decisions.
AUREX is designed to organize AI analysis around a question, its evidence and the actions a person has actually authorized.
AUREX intelligence capabilities and applications
Evidence checking
Trace conclusions to permitted sources, compare records and surface gaps. The synthetic audit demonstrates fixed checks, not live model reasoning or document extraction.
Governed decisions
Separate suggestions from permission. The intended workflow records scope, limitations and human review before any consequential action; this website grants no financial authority.
Applications under development include research analysis, evidence review and DISPATER.AI industrial operations pilots. Customer workflows, live multi-model inference and independently verified outcomes are not established here. Explore the sample today; discuss a scoped pilot rather than assume a production service.
Discuss a PilotSYNTHETIC DEMOSOURCE REVIEWHUMAN HOLD
DISPATER.AI · INDUSTRIAL EVIDENCE REVIEW · PILOT DESIGN
From dispatch to field proof to a human-reviewed invoice draft.
Oilfield and towing/recovery pilot design · synthetic examples · no customer upload enabled. Audit preparedness, not certified auditing, live model inference or a deployed customer portal.
DISPATER.AI is the proposed managed operations service; AUREX/OX is the underlying intelligence and bounded-workflow design. Together, the aim is to make field records, exceptions and invoice preparation easier to review.
Service companies · revenue protection
Designed to surface unbilled work, missing labor, equipment or material revenue, and underbilling against field records.
Operators · expense verification
Designed to challenge vendor overcharges, duplicate invoices or payments, and rates that differ from the contract.
Continuous audit readiness
Planned source-linked field tickets, signed approvals, contracts and rate books, invoices, timesheets, payments, general ledger, exceptions and reviewer decisions. Evidence packets accumulate over time; missing records remain visible.
Qwen-assisted financial intelligence · planned
Extraction and explanation with versioned snapshots and held-out testing—not live inference, live training or financial authority.
One reconciliation and evidence core, with segregated tenant data by design—not an implemented customer portal. Pre-alpha foundations and synthetic research do not establish deployed financial capability.
How it works · proposed workflow
- Dispatch
- Field ticket
- Document / evidence gaps
- Rate / billing reconciliation
- Human-reviewed invoice draft
Oilfield: compare crew and equipment records, field-ticket signoff, PO and approved rates. Towing/recovery: compare dispatch authorization, tow proof, loaded mileage and storage records. Missing documents or approvals stay unresolved—not zero and not safe. Live Railway integration for these industrial workflows is NOT VERIFIED; transport checks are not proof of a connected CRM or audit service.
Evidence → Qualification → Reality → Authority → Effect
Financial authority: HOLD · Human review required. Audit-ready does not mean a certified audit, an audit opinion, or guaranteed regulatory compliance.
FICTIONAL DOSSIER
Oilfield operations
- Invoice
- Signed field ticket
- Approved rate
- Equipment / hour reconciliation
- Invoice
- $28,750
- Documented field ticket
- $25,000
- Documented total
- $25,000
- Provisional variance
- +$3,750
Human review: HOLD
FICTIONAL DOSSIER
Towing operations
- Dispatch authorization
- Tow invoice
- Contracted base fee
- Loaded mileage
- Storage
- Invoice
- $185
- Contracted base fee
- $100
- Loaded mileage · 10 mi × $5
- $50
- Documented total
- $150
- Provisional variance
- +$35
Human review: HOLD
PROPOSED COMMERCIAL DIRECTION · USD
A clear scope. A human decision.
Proposed launch pricing · not yet available to purchase
PROPOSED
Field Starter
$499/month
For small operators and service companies.
- Basic reconciliation
- Source-linked evidence tracking
300 cases/month
TARGET allowance · conditional on cost validation
Inquire about StarterFLAGSHIP · PROPOSED
Field Professional
$1,499/month
The flagship financial evidence workflow.
- Revenue + expense verification
- Continuous audit readiness
- Advanced evidence reports
1,500 cases/month
Proposed monthly allowance
Inquire about ProfessionalPROPOSED
Field Enterprise
from $3,999/month
For multi-location operations.
- Planned integrations
- Volume and governance
- Individually scoped service
Custom allowance
Custom volume · negotiated service
Inquire about EnterpriseFounding customer pilot: proposed $499/month for 60–90 days AFTER closed-beta admission and contract approval.
Indicative terms, subject to scope and cost validation. No charged subscribers or payment processing on this website. No guaranteed savings, recovery, audit results or regulatory compliance. All inquiries go to a person; they do not create a subscription.
CURRENT STATE · SNAPSHOT 2026-10-18 · NOT LIVE TELEMETRY
Where AUREX stands today.
WHAT WORKS
Authenticated status checks
Authenticated read-only status checking is implemented. Exact owner-origin production provenance remains NOT VERIFIED; transport reachability is not intelligence verification.
BEING VALIDATED
Failure handling and local adapters
A draft backend change improves worker-failure accounting, and local model adapters exist as implementation code. Neither is merged, deployed or independently verified.
WHAT COMES NEXT
First real model-backed job
Next gates: a real model-backed job, evidence and result retrieval, then owner acceptance review. Local Qwen integration is in development; successful startup and inference are NOT VERIFIED.
The deployed service is an older read-only bridge; newer backend work exists as draft code that is not merged or deployed. Job submissions from this site remain gated: the canonical API admits only an explicit offline probe, not inference prompts. Engineering SHAs and branch test counts are kept in the owner-only studio diagnostics.
Sources & method · external research as design input
Design input only: enforce boundaries outside the model, evaluate actual outcomes, prefer human-reviewed pilots, and track cost per accepted result. No endorsement and no copied benchmark figures.
Engineering doctrine
KUSANAGI 333 — Sovereign Engineering Standard
“333 is the name of the standard, not the number of lines. Build only what is necessary. Verify everything that matters. Never confuse passing tests with proven production readiness.”
KUSANAGI 333 · Revision 3.0
Structural
Schemas, types and interfaces must agree before anything runs. A build that fails structure never reaches review.
Operational
Behavior is checked against the actual service and endpoints, not mocks alone. A green unit test is not a live system.
Adversarial
Deliberate failures, hostile inputs and a separate witness are required before any claim survives. The system may never be its own sole witness.
Evidence → Qualification → Reality → Authority → Effect
Every outcome follows the same path: gather evidence, check its quality, establish what it supports, confirm authority, then record the bounded effect.
The system may never be its own sole witness.
Status · Local integration tested · Production backend ↔ website connection NOT VERIFIED. This card states a development standard, not a certification of production.
AUREX / OX — Live System Lab
REAL CONNECTION CHECKS · DEMO IS SEPARATE
CONNECTION
UNKNOWN · HOLD
Last checked: INVALID / FUTURE TIME · INVALID TIME
Intelligence: NOT VERIFIED · Spending: UNKNOWN / DISABLED
CAPABILITIES (REPORTED BY SERVICE)
- Memory graphUNKNOWN / HOLD · NOT VERIFIED
- Evidence vaultUNKNOWN / HOLD · NOT VERIFIED
- Model benchmarkingUNKNOWN / HOLD · NOT VERIFIED
- Hypothesis testingUNKNOWN / HOLD · NOT VERIFIED
- Computer useUNKNOWN / HOLD · NOT VERIFIED
BACKEND CONTRACT TEST (OWNER ONLY)
Read-only checks of health, capabilities, budgets, evidence and jobs. Job execution: NOT AVAILABLE unless the backend reports the offline capability.
Owner sign-in requiredEvidence → Qualification → Reality → Authority → Effect
Engineering milestones
- IMPLEMENTED
REST API foundation
Read-only /v1 health, capabilities, evidence and budget endpoints, checked by the separate current transport panel.
- IMPLEMENTED
Server-side bearer token
The service token lives only on the server; the browser never receives it.
- IMPLEMENTED · NOT INDEPENDENTLY VERIFIED
Durable job storage
SQLite-backed job records in the backend repository.
- IMPLEMENTED · NOT INDEPENDENTLY VERIFIED
Evidence ledger
Source design for hashed evidence records; current evidence is determined only by transport checks.
- IMPLEMENTED · NOT INDEPENDENTLY VERIFIED
Ed25519 signature verification
Signed evidence checks in the backend code.
- IN DEVELOPMENT
Structured provenance evidence
Reproducible-workflow and supply-chain attestation formats.
- IMPLEMENTED
$0 budget containment
External inference is OFF by policy; observed budgets are separate and never authorize billing.
- CONCEPT · PLANNED
Adversarial verification
Independent challenge of results before anything is accepted.
- IMPLEMENTED · NOT INDEPENDENTLY VERIFIED
Fail closed: worker unavailable
Source-level refusal handling exists; current worker operation and database persistence remain UNKNOWN.
- CANDIDATE
Exact-SHA release gate
Draft pull request #31 — not merged, not CI-verified.
Limitations: no claimed test counts, website end-to-end tests, production deployment or live worker are presented as proven without CI evidence. Release gate draft PR #31 (not merged)
DEMO · Fixed synthetic sample data · No live task execution
Review a request. Inspect the evidence.
Available: browser-only sample review. No request is sent to a model or execution service.
Live execution: UNAVAILABLE
Transport connectivity does not enable tasks.
- 1. Request
Fixed sample
- 2. Evidence
Ready to review
- 3. Verified outcome
NOT VERIFIED · HOLD
Question: SYNTHETIC SAMPLE: Two sensor logs disagree about when a valve closed. Which timestamp should be trusted?
Generated candidate: Candidate (sample text): prefer log B, because its clock was synchronized more recently.
Sample log A: 10:02:14, clock sync age unknown · Sample log B: 10:02:31, clock sync age 4 minutes (sample value)
- Verification
- NOT VERIFIED
- Authority
- HOLD
- Effect
- NONE
- Scope
- LOCAL_PRACTICE_ONLY
Sample scope and limitations
LOCAL_PRACTICE_ONLY is not an independently verified result. This demo runs entirely in your browser and makes no network requests. Demo execution status: DISABLED / UNCONFIGURED. Live service transport is reported separately.
SERVICE · read-only authenticated checks
Connection is not verification.
UNKNOWN · HOLD
Operations: UNKNOWN / NOT INDEPENDENTLY VERIFIED · release HOLD
Connection evidence and limitations
Intelligence: NOT VERIFIED. These authenticated server checks establish transport and response format only—not working intelligence, independent evidence or security authority. The system may never be its own sole witness. UNKNOWN ≠ SAFE.
- health
- UNKNOWN / unavailable
- capabilities
- UNKNOWN / unavailable
- evidence
- UNKNOWN / unavailable
- budgets
- UNKNOWN / unavailable
- jobs
- UNKNOWN / unavailable
Service version: Unknown · Evidence: UNKNOWN · Spending: UNKNOWN / DISABLED. No inference or jobs are requested by this page.
Last server check: INVALID / FUTURE TIME · INVALID TIME · INVALID TIME · Refreshes while this page is open.
Memory graph
UNKNOWN · NOT VERIFIED
Evidence vault
UNKNOWN · NOT VERIFIED
Model benchmarking
UNKNOWN · NOT VERIFIED
Hypothesis testing
UNKNOWN · NOT VERIFIED
Computer use
UNKNOWN · NOT VERIFIED
Current connectivity could not be established. Owner diagnostics remain private; no previous successful check counts as current clearance.
Evidence-first research
A persuasive answer is not enough. AUREX is our experimental evidence-first research and reasoning evaluation lab: a place to test sources, challenge explanations and learn when not to conclude.
The read-only connection checks below are separate from intelligence verification. Model execution, tool connections and security scans remain unverified here; the educational demo is synthetic.
PROPOSED · DEVELOPMENTAL · NOT OPERATIONAL
A proposed security lab in three colors.
This describes design intent only. No scans, penetration tests or attack simulations run on this website, and no result here is evidence that anything is secure. UNKNOWN ≠ SAFE.
RED · challenge
Authorized, scoped adversarial testing of owner-controlled systems only—planned, not running.
BLUE · defense
Reviewing configuration, dependencies and logs to find what needs fixing—planned, not running.
PURPLE · verification
Comparing challenge and defense findings with independent evidence; the system may never be its own sole witness.
Evidence-governed design / In development
What would prove the design wrong?
AUREX is the developmental AI intelligence system. Its research tests evidence handling, competing model interpretations, reproducible questions and when to withhold a conclusion.
Its role is to support independent adversarial evaluation, not arbitrate BCXMET product security. An internally produced test is not an independent witness. Protocols, item-level records and external replication are required before promoting a performance claim.
The aim is simple: know the source, test alternative explanations and keep permission separate from persuasion. Independent checks must support consequential claims. Approval must be current, limited in scope and withdrawable before an action—not after it.
EVIDENCE → QUALIFICATION → REALITY → AUTHORITY → EFFECT
- 01
Evidence
What was observed?
- 02
Qualification
Can we rely on it here?
- 03
Reality
What is supported?
- 04
Authority
What is permitted now?
- 05
Effect
What happened within bounds?
History is not permission.
A past observation stays in the record. A present action needs current, scoped authorization. Missing, stale or unverified evidence never counts as PASS. UNKNOWN ≠ SAFE.
A system must never be its own sole witness.
Independent support must be demonstrated, not assumed from agreement between models. Unresolved checks mean HOLD → human review. ORB MOTION AUTHORITY = NONE; colors, wheels and zoom provide presentation only.
DESIGN INTENT / DEVELOPMENTAL · This public explanation does not establish operational gates, independently verified product performance or autonomous authority.
PUBLIC · Real public source metadata · Read-only from GitHub
AUREX source repository
This panel shows live public repository facts, kept separate from the synthetic demonstration above. Repository and CI status describe code activity only; they do not verify the AUREX research engine, its evidence or any result.
- Research engine
- NOT_CONNECTED — cannot run research from this website
- Repository
- Loading…
AUREX OX · connection status
Research scope. Separate from transport.
The live checks above report connection only. Local shell materials and independent intelligence verification remain separate; the educational demo is not service output.
NOT AN EXECUTION CONSOLE
AUREX OX public interface (Phase 6)
The public interface exists here. Synthetic examples and read-only checks never establish worker or financial authority.
NOT INCLUDED
Public/private shells (Phase 7)
Private shell files are never bundled into this public site.
NOT VERIFIED
Independent evidence
Capabilities become REAL only with fresh evidence from a healthy hosted service.
- Architecture
- Design intent and contracts. Describes, never proves.
- Synthetic demo
- Fixed, labeled sample text. Not produced by any AUREX engine.
- Verified live state
- Operational verification UNKNOWN / HOLD. Fresh authenticated transport is separate; deployment records alone never prove execution.
Development integration
Authored status · 2026-10-27- API source / testssource · SOURCE REPORTED · CI UNKNOWN
Historical branch-local tests and drafts are source evidence only. Exact deployed-head CI artifacts are not supplied; CI remains UNKNOWN / HOLD.
- Railway deploymentdeployed · DEPLOYMENT SUCCESS REPORTED
Owner supplied direct account inspection: staging October 24 and production October 22; owner reports staging aurex-api online October 27. Deployment success proves neither live endpoint health nor end-to-end operation.
- Staging Qwen inferencedeployed · OWNER-REPORTED · 2026-10-27
Owner reports qwen-private-staging deployed with Qwen2.5-0.5B-Instruct loaded; Railway logs show a processed prompt and an internal authenticated /v1/xaeli-chat/ask check returned text. Internal checks are not independent verification or general-chat proof.
- Website ↔ XAELI chatend-to-end · USER-OBSERVED WORKING · STAGING · 2026-10-27
User/owner-reported mobile website screenshot · 2026-10-27. User/owner reports a mobile website screenshot showing an open-ended audit-preparedness cost question and a contextual XAELI response. This supports a user-observed browser chat interaction in staging, not independently verified transport, inference, model provenance or answer accuracy. No cryptographic verification, generalized availability, audit approval, storage isolation or production-security proof; production readiness HOLD, financial authority NONE.
- Website ↔ backend authenticationend-to-end · UNKNOWN / HOLD
Only the current complete, fresh authenticated five-GET projection can establish read-only transport. Authored deployment facts never substitute for this check.
- Worker / SQLite ↔ websiteend-to-end · UNKNOWN / HOLD
A storage attachment is not persistence proof. Current worker operation, job/evidence read-back, restart durability and independent operational verification are UNKNOWN. Storage isolation and recovery remain blocked.
Current authenticated transport · last checked: NOT REPORTED / INVALID TIME · INVALID TIME · INVALID TIME
- staging · deployed 2026-10-24
- SUCCESS · one replica · storage attached
- Branch main · SHA f2db438f0020dbd455bcb5c7b21b7352d133bf2c
- production · deployed 2026-10-22
- SUCCESS · one replica · storage attached
- Branch integration/kusanagi-xvi-railway · SHA 57a079f93b947594f1a6f854aa46a6f606f86f76
Source: owner-supplied direct Railway account inspection, authored 2026-10-27; deployment dates have no supplied time-of-day. Scope: deployment records only, not website-observed health or live telemetry.
CRITICAL deployment blocker: storage isolation, full recovery and rollback NOT VERIFIED. Storage mount ≠ persistence. No writes or deployment authorized.
Independent operational verification UNKNOWN · staging browser chat user-observed, answers unverified · financial authority NONE · production release HOLD · paid external inference OFF. Evidence → Qualification → Reality → Authority → Effect. No self-witness.
Historical October 23 operator evidence · not current telemetry
Historical October 23 progress · operator report
Owner-verified Chromebook / Oct 23, 2026 · 2026-10-23 · not live telemetry
- NVIDIA model catalog
- CATALOG LISTED
- nvidia/nemotron-3.5-lightning-30b-a3b appeared in an authenticated catalog GET. Zero inference calls; entitlement and accuracy NOT VERIFIED.
- Railway staging access
- STAGING AUTH PASS
- Agent status returned authenticated HTTP 200 externally and inside the container. Running provider: OPENAI; chat configured: false. This does not prove NVIDIA deployment.
- NVIDIA backend configuration
- SAVED · NOT DEPLOYED
- Owner reports staging configuration saved. Inference and billing flags remain false; website-side staging availability NOT VERIFIED.
- SQLite local snapshot
- INTEGRITY CHECK PASSED
- The identified database snapshot passed its integrity check. Full volume recovery, storage isolation and rollback remain NOT VERIFIED.
- XAELI website model link · historical October 23
- NOT LIVE THEN · SUPERSEDED OCTOBER 27
- This October 23 observation predates the user-observed working staging website chat on October 27. The NVIDIA owner-only bridge remains a draft candidate; neither observation grants financial authority or independent model verification.
AUTHORITY NONE · RELEASE HOLD · Unknown ≠ verified. Catalog, transport and local integrity do not prove model accuracy or recovery readiness.
Evidence → Qualification → Reality → Authority → Effect. Current transport results stay separate; this report grants no execution or spending permission.
PUBLIC · AUREX capability registry · Wave 1
What exists, what is waiting, what is planned
REAL / CONNECTED describes website implementation only—not verified intelligence. NOT CONFIGURED means a capability is not established here, even when service transport is reachable. PLANNED means no implementation exists yet. The live read-only checks above are separate from this design registry.
REAL / CONNECTED · owner only
Private Command Center
Owner-only console with server-verified access.
REAL / CONNECTED · owner only
Website health & claims auditor
Owner-run allowlisted page health audit; claim review remains manual.
REAL / CONNECTED
Public source metadata
Read-only public GitHub branch, commit and CI facts.
NOT CONFIGURED
Sovereign computing
Run research on owner-controlled hardware.
NOT CONFIGURED
Memory graph
Linked records of sources, claims and decisions.
NOT CONFIGURED
Semantic memory
Similarity search over approved research records.
NOT CONFIGURED
Evidence vault
Provenance-tracked storage of evidence artifacts.
NOT CONFIGURED
Model router
Select models per task within budget limits.
NOT CONFIGURED
Model benchmarking
Repeatable model comparisons with item-level records.
NOT CONFIGURED
Hypothesis testing
Pre-registered tests with held-out evaluation.
NOT CONFIGURED
Adversarial council
Separate challenge perspectives; not proof of independence.
NOT CONFIGURED
Research swarm
Coordinated bounded research tasks.
NOT CONFIGURED
Regression hunter
Detect performance or behavior regressions between runs.
NOT CONFIGURED
Code archaeology
Trace history and intent across repositories.
NOT CONFIGURED · owner only
Unix workers
Sandboxed allowlisted jobs on the AUREX service; never a public shell.
NOT CONFIGURED · owner only
Python lab
Sandboxed vetted notebooks on the AUREX service; no arbitrary execution from the website.
NOT CONFIGURED · owner only
Private research API
Authenticated server-to-server AUREX API.
NOT CONFIGURED
Offline intelligence
Local-only operation without network dependence.
PLANNED
Digital twin
Modeled system replicas for scenario testing.
PLANNED
Scientific discovery
Structured hypothesis generation from open data.
PLANNED
Model distillation
Smaller task models trained from approved outputs.
PLANNED
Temporal intelligence
Freshness, ordering and time-aware evidence reasoning.
PLANNED · owner only
Patent evidence mapping
Map evidence to claims for counsel review; not legal advice.
PLANNED · owner only
Reverse engineering (authorized only)
Only on owned or explicitly authorized targets.
PLANNED
Controlled self-improvement
Proposed changes gated by tests and human approval.
REAL / CONNECTED
Cost governor
Server-enforced spend limit; default budget is zero (no spend).
Owner controls are private. Owner sign-in required.
PRIVATE · Owner only
Private workspace
Private research details, service status, budgets and audit records are never sent to this page. The cards below are illustrative placeholders, not blurred real data.
- PRIVATE
Owner research notebook
Placeholder — no data shown
- PRIVATE
Service health & budget
Placeholder — no data shown
- PRIVATE
Audit trail
Placeholder — no data shown
Owner sign-in required — access is checked on the server.
AUREX Command Center / Local research workstation
Organize the work. Inspect the evidence.
These modules describe owner-reported work and design intent. A working connection needs a verifiable run record—not a reassuring status label. Public repository metadata is separate from production readiness.
Project Manager
Organize research projects, tasks and milestones; completion still requires supporting records.
Transfer Hub
Explore moving research materials between workspaces with explicit source, destination and access checks.
GitHub Hub
Coordinate repository-related research work. This website does not connect to or expose private repositories.
AI Model Manager / Local models
Explore model selection and local model workflows. Local Qwen integration is in development; successful startup and inference are NOT VERIFIED here.
Scientific Research
Structure questions, sources, competing hypotheses and reproducible experiments.
CERN / Open-data research tools
Explore research using public scientific datasets. No CERN affiliation, tool connection or completed analysis is established here.
Security Lab
MobSF integration is within the reported scope; installation, connectivity and successful scans are unverified. No device protection or scan result is claimed.
Evaluation Engine
Explore repeatable tests, adversarial cases, failure analysis and abstention—not official benchmark certification.
Evidence Ledger / Reports
Organize source references, run records, limitations and review summaries. Report presence is not proof of a claim.
CUSTOS PRIME Ω / Independent verification
A proposed verification layer for separate checks and challenge. Independence must be demonstrated through witnesses and evidence; the module name does not establish it.
Implementation scope / Connections and readiness unverified. The original AUREX Ω Command Center V6 screenshot is not available here; no dashboard reproduction or newer-version status is presented.
Purpose / Research, not protection
Let the difficult cases change the answer.
AUREX explores how conclusions are supported, challenged and withheld. It is separate from BCXMET phone security authority; research activity cannot establish that a device is safe.
Experimental directions
Reproducible questions, competing explanations, adversarial challenge, failure analysis and abstention are areas of exploration. Abstention means declining to conclude when the evidence is insufficient.
Independent verification
Independent qualification is not established here. No independently verified benchmark result is presented. This is not a completed official Humanity’s Last Exam evaluation, a live XAELI benchmark or evidence that any manuscript has been solved.
HOLD · Supporting artifacts and independent replication required
Methodology / Proposed review requirements
Make the question repeatable.
These are review requirements and research directions, not a claim that every gate has already been completed.
01 / REVIEW REQUIREMENT
Define the question
Record scope, sources, dataset version and exclusions before interpreting a result.
02 / REVIEW REQUIREMENT
Challenge the answer
Compare alternatives and adversarial cases. Preserve contradictions and wrong answers.
03 / REVIEW REQUIREMENT
Record the limits
Separate observations from inferences; disclose freshness, uncertainty and abstentions.
04 / REVIEW REQUIREMENT
Invite independent review
Provide protocols, scoring rules and item-level records for scoped replication before promoting a claim.
Evidence → Qualification → Reality → Authority → Effect
A good explanation cannot authorize itself.
First collect the evidence. Check its quality and what it says about the present situation. Then check permission before any action. A research result, confident answer or luminous display cannot supply that permission.
- 01
Evidence
- 02
Qualification
- 03
Reality
- 04
Authority
- 05
Effect
Forward evidence
Observation → proposed consequenceConverge
or HOLD
or HOLD
Independent reverse reconstruction
Proposed consequence → supporting evidenceWitness ≠ Truth · Witness ≠ Authority. No effect without revalidation. The system may never be its own sole witness.
The system must never be its own sole witness. UNKNOWN ≠ SAFE. Missing, stale, skipped, failed or unresolved evidence never counts as PASS. Historical findings do not establish current clearance. No authority comes from visualization: ORB MOTION AUTHORITY = NONE.