You are working on the existing production website:
MISSION:
Build a major GLOBAL authority cluster around:
KUBERNETES + AGENTIC AI + AI INFRASTRUCTURE + LLM / GENAI INFRASTRUCTURE + GPU INFRASTRUCTURE + CLOUD AI + HYBRID / ON-PREMISES AI + AI SRE + AI SECURITY + AI FINOPS + PRODUCTION INCIDENTS + INDUSTRY AI + GLOBAL JOB SUPPORT / PRODUCTION SUPPORT.
This must become a CONNECTED SITE-WIDE KNOWLEDGE GRAPH.
This is NOT a collection of disconnected Kubernetes SEO pages.
ZERO ORPHAN PAGES.
No page may be created unless Claude can determine:
- which parent/hub links to it;
- which existing pages link to it;
- which related pages it links to;
- which country/industry/use-case pages connect to it;
- how a visitor can eventually reach a commercial CTA.
Every page must participate in a deliberate internal-link graph.
DO NOT create a page first and try to connect it later.
Internal linking architecture must be planned BEFORE generation.
Build a clear navigation/conversion path from the HOME PAGE all the way to deep technical content and conversion.
Target architecture:
HOME ↓ TECHNOLOGIES / SERVICES ↓ KUBERNETES + AI HUB ↓ Choose path:
CLOUD PROVIDER COUNTRY TECHNOLOGY AGENTIC AI GPU / INFERENCE INDUSTRY USE CASE PRODUCTION ISSUE ROLE
↓
DETAILED TECHNICAL PAGE ↓
RELATED ARCHITECTURE / INCIDENT / TROUBLESHOOTING PAGE ↓
JOB SUPPORT / PRODUCTION SUPPORT / INTERVIEW SUPPORT ↓
WHATSAPP / CONTACT / LEAD CTA.
Users must never reach a dead end.
Every deep page should answer:
"What should this person read next?"
and
"What service can help them if they have this problem?"
Before creating anything:
- Crawl/audit the existing project.
- Read sitemap and existing routes.
- Identify existing Kubernetes pages.
- Identify AWS AI pages.
- Identify Azure AI pages.
- Identify GCP AI pages.
- Identify Agentic AI pages.
- Identify MLOps pages.
- Identify AI/ML pages.
- Identify SRE/DevOps pages.
- Identify production-support pages.
- Identify existing country pages.
- Identify existing industry pages.
- Identify existing city pages where relevant.
- Analyze current internal linking.
- Detect orphan or weakly linked existing pages.
- Research September 2026 technologies and incidents.
DO NOT start mass-generating pages before this analysis.
Use CURRENT primary sources.
Research technology status through September 2026.
Prioritize:
kubernetes.io CNCF AWS Microsoft / Azure Google Cloud Red Hat NVIDIA KServe vLLM llm-d Ray Kubeflow MLflow Hugging Face OpenTelemetry Prometheus Grafana Datadog Istio Kubernetes Gateway API.
Research Kubernetes 1.37 and specifically verify relevant status for:
Workload-Aware Scheduling PodGroup gang scheduling Workload-Aware Preemption CompositePodGroup Dynamic Resource Allocation accelerator scheduling GPU scheduling HPA scale-to-zero Memory QoS Metrics API rootless Kubernetes security/isolation large AI workloads.
DO NOT assume preview/GA status.
Verify everything.
The central topic is:
"Operating production AI and autonomous agents on Kubernetes and modern cloud infrastructure in 2026."
Not:
"What is Kubernetes?"
Content should emphasize:
Agentic AI LLM serving RAG MCP GPU inference agent runtimes agent sandboxes multi-agent systems AI observability AI security AI SRE production incidents cost scaling hybrid cloud private AI.
Create or identify a strong main hub such as:
/kubernetes-ai-agentic-job-support/
or a better URL after auditing existing routes.
This becomes the central authority node.
Home must link to this hub through an appropriate existing navigation/content mechanism.
The hub must link to:
AWS Azure GCP OpenShift On-Premises Agentic AI MCP LLM inference GPU MLOps AI SRE AI Security AI FinOps Production Support Industries Countries Use Cases.
Kubernetes, platform engineering, cloud-native infrastructure and AI infrastructure are GLOBAL skills.
Therefore this cluster MUST have intentional country coverage.
But do NOT clone one template hundreds of times.
Country content must have unique market context, architecture relevance and internal linking.
PRIORITY GLOBAL MARKETS:
UNITED STATES CANADA UNITED KINGDOM IRELAND AUSTRALIA NEW ZEALAND SINGAPORE HONG KONG UAE SAUDI ARABIA.
Do NOT treat "Europe" as one page only.
Create a European hub plus country-level coverage.
Priority Europe:
Germany Netherlands Ireland France Sweden Switzerland Denmark Finland Norway Belgium Austria Spain Portugal.
Also research demand/justification for:
Italy Poland Czech Republic Romania Hungary Luxembourg Estonia Lithuania Latvia Greece Croatia Slovenia.
Do not create weak pages purely for geographic SEO.
Country inclusion criteria should consider:
cloud adoption Kubernetes/platform-engineering ecosystem AI adoption technology employment enterprise demand regional cloud infrastructure regulated industries data sovereignty relevance.
European content must be more technically differentiated than simply changing the country name.
Research topics such as:
EU AI Act implications where applicable GDPR data residency sovereign AI European cloud regions private AI hybrid AI regulated AI financial services healthcare telecommunications manufacturing automotive government/public sector.
DO NOT provide unsupported legal advice.
Explain engineering/architecture implications rather than making legal conclusions.
Examples:
data residency architecture private endpoints regional inference on-prem/private LLM deployment audit logging workload identity AI governance model/data isolation multi-region design.
Create a structure such as:
Global Kubernetes AI Hub ↓ USA Kubernetes AI Canada Kubernetes AI UK Kubernetes AI Europe Kubernetes AI Australia Kubernetes AI New Zealand Kubernetes AI Singapore Kubernetes AI UAE Kubernetes AI etc.
European hub:
Europe Kubernetes AI ↓
Germany Netherlands Ireland France Sweden Switzerland Denmark Finland Norway Belgium Austria Spain Portugal and validated additional markets.
Possible URL patterns AFTER checking existing naming:
/kubernetes-ai-job-support-usa/ /kubernetes-ai-job-support-canada/ /kubernetes-ai-job-support-uk/ /kubernetes-ai-job-support-germany/ /kubernetes-ai-job-support-netherlands/ /kubernetes-ai-job-support-ireland/ /kubernetes-ai-job-support-france/ /kubernetes-ai-job-support-sweden/ /kubernetes-ai-job-support-switzerland/ /kubernetes-ai-job-support-denmark/ /kubernetes-ai-job-support-finland/ /kubernetes-ai-job-support-norway/ /kubernetes-ai-job-support-belgium/ /kubernetes-ai-job-support-austria/ /kubernetes-ai-job-support-spain/ /kubernetes-ai-job-support-portugal/ /kubernetes-ai-job-support-australia/ /kubernetes-ai-job-support-new-zealand/ /kubernetes-ai-job-support-singapore/ /kubernetes-ai-job-support-hong-kong/ /kubernetes-ai-job-support-uae/ /kubernetes-ai-job-support-saudi-arabia/
These are examples.
FIRST inspect existing naming conventions.
Avoid creating competing URLs.
A country page should connect:
local market + Kubernetes + Cloud Providers + AI workloads + Agentic AI + production support + industry demand.
For example:
Germany:
Kubernetes OpenShift AKS EKS GKE private AI automotive AI manufacturing AI financial services EU data-residency architectures on-prem/hybrid AI.
Netherlands:
cloud-native platforms fintech enterprise SaaS multi-cloud Kubernetes AI agent infrastructure.
Ireland:
AWS Azure GCP multinational cloud operations SRE platform engineering AI infrastructure.
Switzerland:
financial services insurance pharma regulated/private AI hybrid/on-prem AI.
Sweden/Finland/Norway/Denmark:
cloud-native telecommunications industrial systems energy platform engineering AI infrastructure.
UK:
financial services fintech insurance retail healthcare AI platforms Kubernetes.
Australia:
financial services government healthcare mining telecommunications cloud-native AI.
New Zealand:
cloud/platform engineering public sector financial services AI infrastructure.
Singapore:
financial services regional cloud architecture AI platforms multi-region systems regulated enterprise AI.
Country pages MUST link to cloud/provider architecture.
Example:
Germany Kubernetes AI → EKS AI → AKS AI → GKE AI → OpenShift AI → Private AI.
AWS EKS Agentic AI should also link BACK where appropriate to:
USA Canada UK Germany Ireland Australia Singapore etc.
This creates a two-dimensional graph:
TECHNOLOGY ↔ GEOGRAPHY.
Country pages must also connect to industries.
Example:
Germany Kubernetes AI → Automotive AI → Manufacturing AI → Financial Services AI.
Switzerland → Banking AI → Insurance AI → Pharma AI.
Singapore → Banking AI → Fintech AI → Telecom AI.
Australia → Mining/industrial AI if researched → Financial Services AI → Healthcare AI.
This creates:
GEOGRAPHY ↔ INDUSTRY ↔ TECHNOLOGY.
Someone entering from:
"kubernetes job support germany"
should eventually reach pages such as:
GPU Pod Pending vLLM latency MCP authorization agent loop OOMKilled model-loading failures EKS/AKS/GKE production issues.
And those pages should link back to the relevant commercial support paths.
Country pages must NOT be:
same page + country-name substitution.
Each must contain useful differentiated information.
If insufficient differentiated information exists:
combine the country appropriately under a regional hub rather than publishing a thin page.
Quality > URL count.
Research and cover:
long-running agents agent execution runtimes agent session isolation agent sandboxes secure code execution suspend/resume snapshotting state memory persistent sessions tool execution MCP human checkpoints policy enforcement agent identity workload identity multi-agent systems asynchronous workers event-driven agents queues sandbox lifecycle scale-to-zero tenant isolation filesystem isolation network isolation secrets isolation runtime security agent observability agent action auditing.
Explain when:
managed agent runtimes
are preferable to Kubernetes
and when:
Kubernetes-hosted agent infrastructure
is useful.
No forced Kubernetes bias.
Research current September 2026:
Amazon Bedrock AgentCore AgentCore Runtime Gateway Memory Identity observability MCP Strands EKS EKS Auto Mode if relevant SageMaker AI HyperPod if relevant Inferentia Trainium KServe Ray vLLM NVIDIA Dynamo GPU infrastructure IAM workload identity Secrets Manager KMS CloudWatch OpenTelemetry.
Existing AWS pages MUST be reused and connected rather than duplicated.
Potential missing Kubernetes-specific pages only:
EKS Agentic AI EKS Agent Runtime EKS MCP EKS Agent Sandbox EKS vLLM EKS NVIDIA Dynamo EKS GPU Scheduling EKS AI Observability EKS AI Production Incidents.
Research current:
Microsoft Foundry Hosted Agents Microsoft Agent Framework Semantic Kernel MCP Azure OpenAI AKS GPU workloads vLLM KServe Ray Azure Arc Foundry Local hybrid AI Container Apps API Management Entra Managed Identity Key Vault Private Link OpenTelemetry.
Existing Azure/Foundry pages MUST be incorporated into the graph.
Potential Kubernetes-specific gaps:
AKS Agentic AI AKS Agent Runtime AKS MCP AKS vLLM AKS AI Observability AKS GPU Infrastructure AKS Private AI AKS Production Issues.
Research:
GKE Vertex AI current Google agent runtime/platform GKE Agent Substrate Agent Sandbox Inference Gateway GPU/TPU routing multi-cluster inference Gemini ADK KServe vLLM Ray GPU TPU autoscaling security observability.
Potential pages:
GKE Agentic AI GKE Agent Substrate GKE Agent Sandbox GKE MCP GKE vLLM GKE Inference Gateway GKE GPU AI GKE AI Observability GKE Production Incidents.
Build a major authority area around:
Red Hat OpenShift OpenShift AI OpenShift AI Self-Managed Llama Stack llm-d vLLM KServe Ray GPU Operator MCP OpenShift MCP server MCP Gateway RBAC Ansible hybrid private AI air-gapped AI disconnected AI regulated AI sovereign AI.
Also research:
Rancher RKE2 SUSE AI VMware/Tanzu Nutanix where applicable bare-metal Kubernetes OpenStack DGX private GPU clusters.
Research September 2026:
NVIDIA Dynamo current Dynamo release Dynamo Kubernetes Operator Dynamo Router KV-aware routing disaggregated serving prefill/decode separation vLLM SGLang TensorRT-LLM NIM GPU Operator MIG multi-GPU multi-node inference NVLink RDMA NCCL topology-aware scheduling.
Create only missing high-value pages.
Potential topics:
kubernetes-agentic-ai-job-support kubernetes-ai-agents-production-support kubernetes-agent-runtime-support kubernetes-agent-sandbox-support kubernetes-coding-agent-infrastructure-support kubernetes-multi-agent-system-support kubernetes-agent-state-management-support kubernetes-agent-scale-to-zero-support kubernetes-agent-security-support kubernetes-agent-observability-support kubernetes-agent-cost-optimization.
Cover:
MCP servers MCP gateways tool servers authentication authorization OAuth/OIDC where appropriate RBAC workload identity audit logging rate limits network policies multi-tenancy agent-to-tool security.
Potential pages:
kubernetes-mcp-server-job-support kubernetes-mcp-gateway-support kubernetes-mcp-production-support kubernetes-mcp-security-support kubernetes-mcp-observability-support.
Cover:
vLLM KServe Ray Serve NVIDIA Dynamo SGLang TensorRT-LLM NIM distributed inference multi-model serving KV cache prefix caching batching model routing autoscaling disaggregated inference.
Operational metrics:
TTFT TPOT tokens/sec queue latency throughput GPU utilization GPU memory KV-cache utilization cache hit ratio batch size concurrency cold starts cost/request cost/token.
Connect:
User → Agent → Workflow → LLM → MCP → Tool → Retrieval → Vector DB → API → Pod → Node → GPU → Network → Storage.
Cover:
OpenTelemetry Prometheus Grafana Datadog CloudWatch Azure Monitor Google Cloud Monitoring.
Explain why:
healthy Pods ≠ healthy AI application.
Cover:
prompt injection tool abuse agent privilege escalation overprivileged service accounts credential leakage code execution container breakout dependency risk secret exfiltration network egress model supply chain cross-tenant leakage RBAC OPA/Gatekeeper Kyverno Seccomp AppArmor gVisor Kata Containers rootless Kubernetes.
Do not claim Kubernetes itself solves application-layer AI risks.
Cover:
GPU utilization idle GPU agent idle time scale-to-zero suspend/resume batching quantization model routing Spot/Preemptible capacity cold-start economics storage network transfer token economics.
Research REAL incidents from:
official provider status official postmortems CVE databases Kubernetes security announcements GitHub project issues maintainer reports.
Never fabricate incidents.
For real incidents record:
DATE PLATFORM COMPONENT WHAT HAPPENED IMPACT ROOT CAUSE IF CONFIRMED MITIGATION FIX OPERATIONAL LESSON SOURCE.
Also create evergreen troubleshooting pages for:
GPU Pod Pending GPU OOM CUDA problems device plugin failures CrashLoopBackOff vLLM OOM high TTFT inference timeout model loading GPU underutilization autoscaling failures agent loops MCP failures agent permission errors RAG latency vector DB connectivity.
Connect Kubernetes + Agentic AI to existing industry AI pages.
Industries:
Healthcare Pharma Banking Financial Services Fintech Insurance Retail E-commerce Telecom Manufacturing Automotive Energy Utilities Logistics Supply Chain Cybersecurity Government SaaS Travel Hospitality Education.
Do not duplicate existing industry AI content.
Create Kubernetes-specific industry pages only if they have meaningful infrastructure/operational depth.
Potential architecture pages:
customer-support-agent-kubernetes coding-agent-kubernetes enterprise-rag-kubernetes document-processing-agent-kubernetes incident-response-agent-kubernetes devops-agent-kubernetes sre-agent-kubernetes security-agent-kubernetes data-agent-kubernetes analytics-agent-kubernetes research-agent-kubernetes workflow-agent-kubernetes multi-agent-platform-kubernetes computer-vision-kubernetes edge-ai-kubernetes.
Before implementation create an internal matrix:
PAGE PARENT CHILDREN EXISTING PAGES LINKING IN NEW PAGES LINKING IN RELATED TECHNOLOGIES RELATED CLOUDS RELATED COUNTRIES RELATED INDUSTRIES RELATED USE CASES RELATED INCIDENTS COMMERCIAL CTA BREADCRUMB PATH.
No page may have zero inbound contextual links.
No page may have zero useful outbound links.
After implementation run a crawl.
Detect:
pages with zero inbound links pages linked only from sitemap pages linked only from footer pages with no parent pages with no related-content links pages with no commercial path.
Fix all legitimate orphan pages.
A sitemap entry DOES NOT count as sufficient internal linking.
llms.txt DOES NOT count as sufficient internal linking.
Footer-only linking DOES NOT count as a satisfactory content relationship.
Do NOT turn the homepage into a giant Kubernetes page.
Instead add an appropriate high-level entry into the existing technology/service discovery system.
Example:
HOME → AI / Cloud / Platform Engineering → Kubernetes + AI Infrastructure → Agentic / GPU / Cloud / Industry / Country.
Use the existing visual/design system.
Do not break current homepage intent.
Each hub needs:
intro major problems architecture subtopics cloud choices use cases industries countries production issues related guides job-support CTA.
A user should move naturally:
GENERAL → SPECIFIC → PROBLEM → SOLUTION → SUPPORT.
Implement logical breadcrumbs even with flat URLs.
Example:
Home
Technologies
Kubernetes AI
Agentic AI
Agent Security
or:
Home
Locations
Europe
Germany
Kubernetes AI Job Support.
Use BreadcrumbList schema where valid.
Optimize naturally for:
Google Bing ChatGPT Gemini Claude Perplexity Copilot AI Overviews agent-based search.
Every page should have unique:
title description H1 canonical OG title OG description social metadata.
Use appropriate schema:
WebPage Article TechArticle Service FAQPage when truly appropriate BreadcrumbList Organization.
Do not create spam schema.
Integrate important:
hub pages cloud pages country hubs technical authorities production guides industry nodes.
Do not dump thousands of thin URLs.
Prioritize canonical authority nodes and useful relationships.
Add legitimate new indexable pages to the existing sitemap generation mechanism.
Do NOT create a second sitemap mechanism unless existing architecture requires it.
Before adding any page calculate:
EXISTING URL EXISTING SEARCH INTENT PROPOSED URL PROPOSED INTENT OVERLAP LEVEL ACTION.
ACTION must be:
KEEP UPDATE EXPAND MERGE LINK CREATE DO NOT CREATE.
Do not cannibalize existing ranking pages.
The new cluster must link to existing:
AI/ML Agentic AI RAG MLOps LLM AWS Azure GCP SRE DevOps production support industry country role interview support pages.
Existing strong pages should also receive contextual inbound/outbound links to the new cluster where relevant.
This is not merely:
NEW → OLD.
It must become:
NEW ↔ OLD.
Every major country page should link to:
global Kubernetes AI hub regional hub if relevant AWS/EKS Azure/AKS GCP/GKE OpenShift/on-prem where relevant Agentic AI LLM inference GPU AI SRE AI Security production issue hub relevant industries commercial support.
And those hubs must link back contextually.
Example:
HOME → Kubernetes AI → Europe Kubernetes AI → Germany → OpenShift / AKS / EKS / GKE → Manufacturing AI → Agentic Production Architecture → Production Issue Guide → Job Support CTA.
Another path:
HOME → Locations → Europe → Switzerland → Private AI → Kubernetes GPU Infrastructure → Banking AI → Production Support CTA.
Another:
HOME → GKE AI → Europe → Netherlands → Agent Infrastructure → MCP → Security → Support.
These markets must NOT be an afterthought.
Explicitly include:
Australia New Zealand Singapore.
Research local demand and cloud/platform context first.
Connect them across:
country cloud AI Kubernetes industry job support production support.
Where relevant connect to roles such as:
Kubernetes Engineer Platform Engineer Cloud Engineer DevOps Engineer SRE MLOps Engineer LLMOps Engineer AI Platform Engineer AI Infrastructure Engineer GPU Infrastructure Engineer AI Solutions Architect Cloud AI Engineer Agentic AI Engineer.
Do not generate role pages if equivalent ones already exist.
Technical content should remain genuinely useful.
Do NOT force CTAs every few paragraphs.
Use conversion moments where a user has a natural problem.
Examples:
"Need help debugging this in production?"
"Working with this architecture on a live project?"
"Facing GPU, inference, Kubernetes or agent-runtime issues?"
Then link to relevant support.
Every commercial page should clearly connect to one or more of:
Proxy Job Support Production Issue Support Project Onboarding Cloud/DevOps Support AI/ML Job Support MLOps Support Agentic AI Support SRE Support Interview Support where appropriate.
Do not transform the site into a course/tutorial marketplace.
No generic statements such as:
"AI is transforming every industry."
Write useful senior-engineering content around:
architecture failure modes debugging observability security capacity cost performance deployment scaling operations trade-offs.
A country page must have:
meaningful local context relevant cloud platforms common industries architecture considerations AI infrastructure relevance links to technical content links to production issues commercial intent.
If we cannot create meaningful differentiation:
do not publish the page yet.
Continue existing site convention.
Prefer flat URLs.
Example:
/kubernetes-ai-job-support-germany/
instead of:
/locations/europe/germany/kubernetes/ai/job-support/
Hierarchy is expressed through:
internal links breadcrumbs schema hub relationships.
PHASE 1 Existing site crawl.
PHASE 2 Current URL inventory.
PHASE 3 Existing internal-link analysis.
PHASE 4 September 2026 research.
PHASE 5 Gap analysis.
PHASE 6 Global information architecture.
PHASE 7 Home-to-hub funnel.
PHASE 8 Global Kubernetes AI hub.
PHASE 9 Agentic AI / MCP cluster.
PHASE 10 GPU / inference cluster.
PHASE 11 AWS / Azure / GCP / OpenShift.
PHASE 12 Hybrid / on-premises.
PHASE 13 Production incidents.
PHASE 14 Industry integration.
PHASE 15 Global country hubs.
PHASE 16 Europe country cluster.
PHASE 17 Australia / New Zealand / Singapore / Gulf.
PHASE 18 Role/use-case integration.
PHASE 19 Bidirectional links to existing pages.
PHASE 20 Sitemap / schema / llms updates.
PHASE 21 Full-site orphan crawl.
PHASE 22 Build and SEO validation.
Before saying the task is complete:
crawl every generated URL.
For every URL confirm:
at least 2 useful internal inbound links where reasonable at least 3 useful outbound contextual links where reasonable parent hub relationship exists breadcrumb exists commercial funnel exists country/industry relationships exist where relevant canonical exists metadata unique schema valid sitemap contains URL no duplicate intent exists.
Then explicitly report:
TOTAL NEW PAGES TOTAL UPDATED PAGES TOTAL INTERNAL LINKS ADDED ORPHAN PAGES BEFORE ORPHAN PAGES AFTER DUPLICATE/CANNIBALIZATION RISKS FOUND PAGES NOT CREATED DUE TO OVERLAP.
Target:
ORPHAN PAGES AFTER = 0 for the new cluster.
Report:
- September 2026 research completed
- Kubernetes updates incorporated
- Agentic AI updates
- AWS coverage
- Azure coverage
- GCP coverage
- OpenShift/on-prem coverage
- NVIDIA/GPU coverage
- MCP coverage
- AI SRE/security/FinOps
- incidents researched
- industry cluster
- countries covered
- European countries covered individually
- Australia/NZ/Singapore coverage
- existing pages updated
- new pages created
- duplicate pages avoided
- internal-link graph changes
- homepage funnel changes
- sitemap changes
- llms.txt changes
- llms-full.txt changes
- structured-data changes
- orphan-page crawl results
- build validation
- remaining opportunities.
Do NOT build:
hundreds of Kubernetes keyword URLs.
Build:
A GLOBAL PRODUCTION AI INFRASTRUCTURE KNOWLEDGE GRAPH.
The graph should connect:
HOME ↕ KUBERNETES ↕ AGENTIC AI ↕ CLOUD PROVIDERS ↕ COUNTRIES ↕ EUROPEAN MARKETS ↕ INDUSTRIES ↕ USE CASES ↕ GPU / INFERENCE ↕ MCP ↕ AI SRE ↕ SECURITY ↕ PRODUCTION INCIDENTS ↕ JOB SUPPORT.
A visitor entering through:
Google ChatGPT Gemini Claude Perplexity a country page a Kubernetes error an AWS page an Azure page a GCP page an OpenShift page an industry an agent problem an MCP problem a GPU problem or the homepage
must be able to navigate naturally through the ecosystem without ever hitting an orphan or dead-end page.
START WITH:
- FULL EXISTING SITE AUDIT
- SEPTEMBER 2026 EXTERNAL RESEARCH
- INTERNAL-LINK GRAPH
- COUNTRY/GLOBAL ARCHITECTURE
- DUPLICATION MATRIX
ONLY AFTER THESE ARE COMPLETE SHOULD PAGE GENERATION BEGIN.