{"schemaVersion":"corex.scalability-capabilities.v1","state":"ACTIVE","counts":{"ACTIVE_FUNCTIONAL":8,"ACTIVE_OPTIMIZED":6,"ADAPTER_READY_PROVIDER_NOT_BOUND":1,"LATENT_NOT_ACTIVATED":2,"EVIDENCE_PENDING":1},"principles":{"providerSubscriptionIsArchitectureCeiling":false,"canonicalWriteAuthoritySeparatedFromReadScale":true,"modeledScaleIsMeasuredClaim":false,"performanceClaimsRemainWorkloadSpecific":true},"capabilities":[{"id":"fluid-horizontal-runtime","title":"Elastic runtime concurrency and scale-out","state":"ACTIVE_FUNCTIONAL","layer":"RUNTIME","valueProposition":"Uses warm in-function concurrency first and allows provider-managed horizontal scale without changing CoreX authority or tenant semantics.","evidence":["vercel.json#fluid=true","src/core/adaptiveCapacity.ts"],"currentUse":"Production runtime elasticity.","optimizationPath":"Increase provider compute class only after measured CPU/memory pressure remains binding.","publicSafe":true},{"id":"bounded-postgresql-pooling","title":"Bounded PostgreSQL connection pooling","state":"ACTIVE_OPTIMIZED","layer":"PERSISTENCE","valueProposition":"Prevents application-instance growth from multiplying database connections without control.","evidence":["src/core/prisma.ts","src/core/adaptiveCapacity.ts"],"currentUse":"All Prisma/PostgreSQL access.","optimizationPath":"Tune pool/database compute/IOPS together from measured pool pressure.","publicSafe":true},{"id":"leased-horizontal-event-workers","title":"Leased horizontally scalable delivery workers","state":"ACTIVE_OPTIMIZED","layer":"QUEUE","valueProposition":"PROCESSING leases allow multiple workers to drain durable delivery backlogs without concurrent duplicate selection.","evidence":["src/core/eventDelivery.ts","scripts/benchmark-corex-queue-pressure.mjs"],"currentUse":"Durable event delivery, retry and replay with automatic bounded scheduled backlog draining.","optimizationPath":"Increase workers only when queue depth/age is binding.","publicSafe":true},{"id":"scheduled-event-backlog-drain","title":"Automatic bounded event backlog draining","state":"ACTIVE_OPTIMIZED","layer":"QUEUE","valueProposition":"A portable scheduled reconciler drains multiple leased delivery batches per invocation, exits quickly when idle, and allows overlapping schedulers without duplicate ownership.","evidence":["src/routes/eventDeliveryReconciliation.ts","src/core/scheduledJobs.ts","vercel.json"],"currentUse":"Production scheduler binding at one-minute cadence with bounded batch/time policy.","optimizationPath":"Tune batch count/time budget from observed queue depth and provider limits rather than keeping always-on workers.","publicSafe":true},{"id":"dead-letter-replay","title":"Dead-letter and replay recovery","state":"ACTIVE_FUNCTIONAL","layer":"QUEUE","valueProposition":"Absorbs downstream outages without forcing synchronous request paths to fail permanently.","evidence":["src/core/eventDelivery.ts","src/routes/eventDelivery.ts"],"currentUse":"Durable integration delivery recovery.","publicSafe":true},{"id":"credential-derived-projections","title":"Canonical credential plus derived projections","state":"ACTIVE_FUNCTIONAL","layer":"READ_PATH","valueProposition":"Separates canonical authority from document/card/W3C/proof representations so read/verification surfaces can scale independently.","evidence":["src/core/credentialProjectionRegistry.ts","src/core/credentialProjection.ts"],"currentUse":"Credential documents, cards, VC and independent-proof representations.","optimizationPath":"Introduce dedicated read replicas/materialized projections only when verification traffic becomes persistence-bound.","publicSafe":true},{"id":"hot-proof-verification-cache","title":"Hot cryptographic proof verification cache","state":"ACTIVE_OPTIMIZED","layer":"CRYPTO","valueProposition":"Reuses verification results for immutable credential payload/proof pairs inside warm instances while always re-reading live lifecycle state.","evidence":["src/core/credentialProofCache.ts","src/core/credentialLifecycle.ts"],"currentUse":"Repeated verification of hot credentials.","optimizationPath":"Keep lifecycle uncached; use distributed/materialized read projection only after measured demand.","publicSafe":true},{"id":"single-read-certificate-projection","title":"Single-read certificate projection","state":"ACTIVE_OPTIMIZED","layer":"READ_PATH","valueProposition":"Avoids a duplicate credential database lookup when rendering human-readable certificate/QR projections.","evidence":["src/core/credentialProjection.ts","src/core/credentialLifecycle.ts"],"currentUse":"Certificate and QR projection paths.","publicSafe":true},{"id":"public-verification-rate-limit-provider-boundary","title":"Public verification rate-limit provider boundary","state":"ADAPTER_READY_PROVIDER_NOT_BOUND","layer":"READ_PATH","valueProposition":"Preserves a low-latency local guard today while allowing a future distributed limiter to provide globally consistent abuse control across autoscaled instances.","evidence":["src/core/publicVerificationGuard.ts"],"currentUse":"Local per-instance verification abuse guard with distributed-provider adapter contract.","optimizationPath":"Bind a distributed edge/Redis/WAF-backed provider when traffic or abuse patterns justify global coordination.","publicSafe":true},{"id":"workload-identity-portability","title":"Keyless workload identity portability","state":"ACTIVE_FUNCTIONAL","layer":"IDENTITY","valueProposition":"Supports Vercel OIDC plus generic/AWS/Azure/GCP/Kubernetes/GitHub workload identity without making secrets or one runtime provider architectural dependencies.","evidence":["src/core/workloadIdentity.ts","scripts/verify-runtime-portability.mjs"],"currentUse":"Service-to-service and signer transport composition.","publicSafe":true},{"id":"oci-portability","title":"OCI/container runtime portability","state":"ACTIVE_FUNCTIONAL","layer":"PORTABILITY","valueProposition":"Node 22 standard entrypoint allows CoreX to move from serverless to containers/Kubernetes/private cloud when workload economics or sovereignty require it.","evidence":["Dockerfile","scripts/verify-oci-portability.mjs"],"currentUse":"Portability and future dedicated-node deployment.","publicSafe":true},{"id":"signer-abstraction","title":"Independent signer capacity boundary","state":"ACTIVE_FUNCTIONAL","layer":"CRYPTO","valueProposition":"Separates CoreX software throughput from KMS/provider signer quotas so signing capacity can scale independently.","evidence":["src/core/productionSigner.ts","src/core/awsKmsCredentialSigner.ts"],"currentUse":"Production credential signing.","optimizationPath":"Benchmark local and KMS signing separately; increase signer quota only when it becomes binding.","publicSafe":true},{"id":"integrity-anchor-abstraction","title":"Asynchronous integrity-anchor provider boundary","state":"ACTIVE_FUNCTIONAL","layer":"INTEGRATION","valueProposition":"Keeps ledger/transparency confirmation outside the CoreX synchronous authority path where policy allows.","evidence":["src/core/integrityAnchor.ts","src/core/ledgerAnchor.ts"],"currentUse":"Independent integrity evidence.","optimizationPath":"Trust Epoch batching is active when the checkpoint provider is bound; tune epoch size/age from measured anchor cost and confirmation throughput.","publicSafe":true},{"id":"usage-and-quota-observation","title":"Usage metering and entitlement quota primitives","state":"ACTIVE_FUNCTIONAL","layer":"EVIDENCE","valueProposition":"Provides idempotent tenant/workspace usage observations and quota enforcement foundations for commercial control and capacity economics.","evidence":["src/core/usageMetering.ts","src/core/quotaEnforcement.ts"],"currentUse":"Commercial metrics and entitlement/quota control.","publicSafe":true},{"id":"distributed-verification-read-plane","title":"Distributed read/verification plane","state":"LATENT_NOT_ACTIVATED","layer":"READ_PATH","valueProposition":"The projection architecture can support read replicas, regional verification nodes and safe projection caches without moving canonical write authority.","evidence":["src/core/credentialProjectionRegistry.ts","docs/performance/COREX_ADAPTIVE_CAPACITY_AUTOSCALE_BASELINE_2026-09-23.md"],"currentUse":"Architecture-ready; dedicated replica topology not currently bound.","optimizationPath":"Activate only after read traffic is measured as persistence-bound.","publicSafe":true},{"id":"multi-core-crypto-workers","title":"Multi-core cryptographic worker execution","state":"LATENT_NOT_ACTIVATED","layer":"CRYPTO","valueProposition":"CPU-bound hashing/sign/verify workloads can use dedicated worker processes/threads on multi-vCPU nodes or container clusters.","evidence":["scripts/benchmark-corex-primitive-scale.mjs"],"currentUse":"Benchmark-only multi-process evidence.","optimizationPath":"Activate on dedicated multi-vCPU/container runtime if CPU-bound cryptographic work becomes material.","publicSafe":true},{"id":"batch-merkle-anchor","title":"Batch/Merkle Trust Epoch anchoring","state":"ACTIVE_OPTIMIZED","layer":"INTEGRATION","valueProposition":"Amortizes external integrity checkpoints across many historical-integrity records while preserving per-record Merkle inclusion proofs.","evidence":["src/core/trustEpoch.ts","src/core/trustEpochReconciliation.ts","src/core/polygonPosTrustEpochCheckpoint.ts"],"currentUse":"Trust Epoch sealing and external checkpoint reconciliation through COREX_POLYGON_POS_TRUST_EPOCH_V1 on polygon-pos.","optimizationPath":"Tune minimum leaves, maximum leaves and maximum age from measured anchor cost, latency and provider rate limits.","publicSafe":true},{"id":"independent-multihost-certification","title":"Independent multi-host distributed certification","state":"EVIDENCE_PENDING","layer":"EVIDENCE","valueProposition":"Separates shared-host composition tests from true cross-node scale-efficiency claims.","evidence":["docs/performance/COREX_AUTOMATED_CAPACITY_TRANCHE_BASELINE_2026-09-23.md"],"currentUse":"Current 1/2/4/8 logical runtime scale tests share one benchmark host.","optimizationPath":"Run on independent hosts/provider-grade distributed infrastructure when a public distributed throughput claim is required.","publicSafe":true}]}