Artifact Verifier Spec · Config Workshop

What this command does. cub installer is a released, open-source plugin for the cub CLI. cub installer setup pulls a catalog package and writes its Kubernetes files locally. It does not apply those files to a cluster; use kubectl, Argo CD, or Flux for delivery. The generated scripts stop before doing any work when the plugin or kustomize is missing.

New to cub? Install the cub CLI first. You can pull and render public catalog packages without an account. Commands that save or change ConfigHub data require you to sign in.

Generated at: 2026-07-30T12:38:02.000Z UTC · source: committed helm-expt evidence for this rendered repository document.

This is the executable proof contract for helm-expt.

The verifier checks the current recipe/package artifact chain, the Redis recipe/variant/revision proof, the durable Redis installer package proof, the promoted metrics-server, ingress-nginx, cert-manager, external-secrets, Argo CD, PostgreSQL, RabbitMQ, kube-prometheus-stack, Loki, Longhorn, MySQL, Grafana, Vault, Secrets Store CSI Driver, Prometheus, MongoDB, Nginx, Tempo, and Consul proofs, and the adversarial public-chart harnesses.

Scope

The current artifact-chain verifier checks every recipe/package pair:

recipes/*/*/*/recipe.yaml
recipes/*/*/*/publication/installer-package-receipt.yaml
packages/*/*/*/installer.yaml
packages/*/*/*/bases/*/upstream.yaml
recipes/*/*/*/revisions/*/r001/rendered/release-objects.yaml
recipes/*/*/*/revisions/*/r001/receipts/*.yaml

It proves that current executable fixtures point at packages/, package receipts match package source files by SHA256/byte length, rendered revision digests bind to inventories, equivalence receipts, scan receipts, and install gates, and negative self-tests catch tampering.

The Redis proof verifier checks:

recipes/bitnami/redis/25.5.3/
packages/bitnami/redis/25.5.3/

That includes HelmPlan, ChartDossier, source/dependency locks, control points, effective values, variants, variant revisions, rendered objects, object inventories, Helm equivalence receipts, render receipts, scan receipts, install gates, variant diff evidence, the installer package receipt, and the package source tree. Remote ConfigHub upload/OCI evidence is recorded separately under runs/redis-confighub/latest/ because it depends on hosted ConfigHub auth.

The adversarial public-chart harness verifier checks:

data/adversarial10/corpus.yaml
data/adversarial10/corpus.lock.yaml
data/adversarial10/proof-readiness.csv
data/adversarial10/charts/*/helm-plan.yaml
data/adversarial10/charts/*/render-receipt.yaml
data/adversarial10/charts/*/rendered/default.yaml
data/adversarial10/charts/*/rendered/object-inventory.yaml

This is the first scale-out proof index. It is not a full recipe/variant proof for all 10 charts. It proves that the corpus is pinned, every row has a machine-readable HelmPlan and render receipt, successful render attempts bind stored manifests by SHA256, failed attempts record blocker receipts, and the CSV is generated from those artifacts.

The metrics-server proof verifier checks:

recipes/metrics-server/metrics-server/3.13.0/
packages/metrics-server/metrics-server/3.13.0/

That proof is the first promoted row from the adversarial harness. It checks two variants, default and external-tls-ca, including source/dependency locks, control points, effective values, target fact requirements, rendered object inventories, render receipts, Helm equivalence receipts, scan receipts, install gates, and deterministic cub installer package/setup behavior.

The ingress-nginx proof verifier checks:

recipes/ingress-nginx/ingress-nginx/4.15.1/
packages/ingress-nginx/ingress-nginx/4.15.1/

That proof is the second promoted row from the adversarial harness. It checks two variants, default and admission-disabled, including source/dependency locks, control points, effective values, rendered object inventories, render receipts, Helm equivalence receipts, scan receipts, install gates, and deterministic cub installer package/setup behavior.

The cert-manager proof verifier checks:

recipes/jetstack/cert-manager/v1.20.2/
packages/jetstack/cert-manager/v1.20.2/

That proof is the third promoted row from the adversarial harness. It checks two variants, default and crds-enabled, including source/dependency locks, control points, effective values, rendered object inventories, render receipts, Helm equivalence receipts, scan receipts, install gates, and deterministic cub installer package/setup behavior.

The external-secrets proof verifier checks:

recipes/external-secrets/external-secrets/2.5.0/
packages/external-secrets/external-secrets/2.5.0/

That proof is the fourth promoted row from the adversarial harness. It checks two variants, default and no-crds, including source/dependency locks, control points, effective values, rendered object inventories, render receipts, Helm equivalence receipts, scan receipts, install gates, separated Secret handling, and deterministic cub installer package/setup behavior.

The Longhorn proof verifier checks:

recipes/longhorn/longhorn/1.11.2/
packages/longhorn/longhorn/1.11.2/

That proof is the tenth promoted row from the adversarial harness. It checks two variants, default and ui-ingress, including source/dependency locks, 22 Longhorn CRDs, pre-upgrade hook policy, admission/recovery observation, cluster RBAC, privileged storage workload policy, StorageClass/default-setting policy, UI ingress exposure, rendered object inventories, render receipts, Helm equivalence receipts, scan receipts, install gates, and deterministic cub installer package/setup behavior.

The MySQL proof verifier checks:

recipes/bitnami/mysql/14.0.3/
packages/bitnami/mysql/14.0.3/

That proof is the eleventh promoted row from the adversarial harness. It checks two variants, static-passwords and existing-secret, including source/dependency locks, generated fact binding for root, user, and replication passwords, target Secret binding, rendered object inventories, render receipts, Helm equivalence receipts, scan receipts, install gates, separated Secret handling, and deterministic cub installer package/setup behavior.

The Grafana proof verifier checks:

recipes/grafana/grafana/10.5.15/
packages/grafana/grafana/10.5.15/

That proof is the twelfth promoted row from the adversarial harness. It checks two variants, static-passwords and existing-secret-ingress, including source/dependency locks, upstream deprecation status, generated admin password binding, target Secret binding, UI ingress exposure, rendered object inventories, render receipts, Helm equivalence receipts, scan receipts, install gates, separated Secret handling, and deterministic cub installer package/setup behavior.

The Vault proof verifier checks:

recipes/hashicorp/vault/0.32.0/
packages/hashicorp/vault/0.32.0/

That proof is the thirteenth full public-chart proof row. It checks two variants, default and ha-raft-ui, including source/dependency locks, TLS posture, injector admission webhook, StatefulSet storage, init/unseal operating policy, HA Raft and UI service exposure, rendered object inventories, render receipts, Helm equivalence receipts, scan receipts, install gates, and deterministic cub installer package/setup behavior.

The Secrets Store CSI Driver proof verifier checks:

recipes/secrets-store-csi-driver/secrets-store-csi-driver/1.6.0/
packages/secrets-store-csi-driver/secrets-store-csi-driver/1.6.0/

That proof is the fourteenth full public-chart proof row. It checks two variants, default and sync-secret-rotation, including source/dependency locks, SecretProviderClass CRDs, CSIDriver kubelet integration, Linux DaemonSet workload policy, cluster RBAC, synced Secret ownership, rotation/provider health settings, rendered object inventories, render receipts, Helm equivalence receipts, scan receipts, install gates, and deterministic cub installer package/setup behavior.

The Prometheus proof verifier checks:

recipes/prometheus-community/prometheus/29.8.0/
packages/prometheus-community/prometheus/29.8.0/

That proof is the fifteenth full public-chart proof row. It checks two variants, default and server-only-ephemeral, including source/dependency locks for the bundled monitoring subcharts, Prometheus server scrape config, Alertmanager/exporter/pushgateway component selection, server PVC/storage policy, cluster RBAC, rendered object inventories, render receipts, Helm equivalence receipts, scan receipts, install gates, and deterministic cub installer package/setup behavior.

The MongoDB proof verifier checks:

recipes/bitnami/mongodb/19.0.7/
packages/bitnami/mongodb/19.0.7/

That proof is the sixteenth full public-chart proof row. It checks two variants, static-passwords and existing-secret-replicaset, including source/dependency locks, generated root password binding, target Secret binding, replica-set and arbiter StatefulSets, persistent storage, NetworkPolicy/PDB policy, Helm hook lifecycle review, rendered object inventories, render receipts, Helm equivalence receipts, scan receipts, install gates, and deterministic cub installer package/setup behavior.

The Nginx proof verifier checks:

recipes/bitnami/nginx/24.0.2/
packages/bitnami/nginx/24.0.2/

That proof is the seventeenth full public-chart proof row. It checks two variants, http-clusterip and existing-tls-ingress, including source and dependency locks, generated TLS mitigation, declared backend and ingress TLS target Secrets, ingress exposure, NetworkPolicy/PDB policy, service exposure, static-site supply-chain slots, metrics add-ons, raw/template extension-slot review, rendered object inventories, render receipts, Helm equivalence receipts, scan receipts, install gates, and deterministic cub installer package/setup behavior.

The Tempo proof verifier checks:

recipes/grafana/tempo/1.24.4/
packages/grafana/tempo/1.24.4/

That proof is the eighteenth full public-chart proof row. It checks two variants, local-persistent and s3-query-observability, including source and empty dependency locks, deprecated chart marker, local/S3 storage backend policy, target S3 credential Secret, query ingress, ServiceMonitor target capability, NetworkPolicy, StatefulSet/headless-Service runtime risk, raw/template extension-slot review, rendered object inventories, render receipts, Helm equivalence receipts, scan receipts, install gates, and deterministic cub installer package/setup behavior.

The Consul proof verifier checks:

recipes/hashicorp/consul/2.0.0/
packages/hashicorp/consul/2.0.0/

That proof is the nineteenth full public-chart proof row and twentieth target chart. It checks two variants, default-control-plane and secure-mesh-existing-secrets, including source and empty dependency locks, 28 rendered CRDs, cluster RBAC, injector admission webhooks, TLS/ACL/gossip target Secret binding, gateway topology, UI ingress, lifecycle Job policy, rendered Secret ownership, StatefulSet/storage policy, raw/template extension-slot review, rendered object inventories, render receipts, Helm equivalence receipts, scan receipts, install gates, and deterministic cub installer package/setup behavior.

The Argo CD proof verifier checks:

recipes/argo-cd/argo-cd/9.5.15/
packages/argo-cd/argo-cd/9.5.15/

That proof is the fifth promoted row from the adversarial harness. It checks two variants, default and no-crds, including source/dependency locks, control points, effective values, rendered object inventories, render receipts, Helm equivalence receipts, scan receipts, install gates, separated Secret handling, and deterministic cub installer package/setup behavior.

The PostgreSQL proof verifier checks:

recipes/bitnami/postgresql/18.6.7/
packages/bitnami/postgresql/18.6.7/

That proof is the sixth promoted row from the adversarial harness. It checks two variants, static-passwords and existing-secret, including source/dependency locks, generated fact binding, target fact binding, rendered object inventories, render receipts, Helm equivalence receipts, scan receipts, install gates, separated Secret handling, and deterministic cub installer package/setup behavior.

The RabbitMQ proof verifier checks:

recipes/bitnami/rabbitmq/16.0.14/
packages/bitnami/rabbitmq/16.0.14/

That proof is the seventh promoted row from the adversarial harness. It checks two variants, static-passwords and existing-secret, including source/dependency locks, generated fact binding for password and Erlang cookie, target fact binding for both Secrets, rendered object inventories, render receipts, Helm equivalence receipts, scan receipts, install gates, separated Secret handling, and deterministic cub installer package/setup behavior.

The kube-prometheus-stack proof verifier checks:

recipes/prometheus-community/kube-prometheus-stack/85.3.3/
packages/prometheus-community/kube-prometheus-stack/85.3.3/

That proof is the eighth promoted row from the adversarial harness. It checks two variants, default and no-crds, including umbrella dependency locks, generated fact binding for Grafana admin password, rendered object inventories, render receipts, Helm equivalence receipts, scan receipts, install gates, separated Secret handling, dashboard ConfigMap normalization, and deterministic cub installer package/setup behavior.

The Loki proof verifier checks:

recipes/grafana/loki/7.0.0/
packages/grafana/loki/7.0.0/

That proof is the ninth promoted row from the adversarial harness. It checks two variants, single-binary-filesystem and simple-scalable-minio, including the blocked default render receipt, source/dependency locks, required storage/schema values, bundled MinIO object-store fixture, rendered object inventories, render receipts, Helm equivalence receipts, scan receipts, install gates, classified Loki ConfigMap normalization, separated Secret handling, and deterministic cub installer package/setup behavior.

Required Invariants

For every current recipe/package directory:

Required recipe files exist: HelmPlan, ChartDossier, source/dependency locks, control points, value model, recipe, and installer package receipt.
recipe.spec.currentExecutableFixture.installerPackage points at the current packages/ tree.
The installer package receipt points at the same package path.
Every package source file listed in the receipt exists and matches its SHA256 and byte count.
Every recipe variant path exists.
Every variant revision has rendered objects, object inventory, Helm equivalence receipt, render receipt, scan receipt, and install gate.
Rendered object SHA256 is identical across the variant revision digest inputs, object inventory, render receipt, scan receipt, and install gate.
Helm equivalence receipt result is pass and points at the same rendered object SHA256.

For the adversarial public-chart harness:

The corpus lock has one row for every chart in corpus.yaml.
Every chart row has a HelmPlan, render receipt, and object inventory.
Successful render receipts point at a stored rendered manifest.
Stored rendered manifest SHA, byte count, and resource count match the receipt.
Failed render receipts do not claim a rendered manifest and record an errorSHA256.
CSV status, readiness, feature flags, primary control point, paths, counts, and rendered manifest SHA match the receipt.
Chart identity, version, render context, and package SHA are present in the receipt.

For the promoted metrics-server proof:

Both variants render deterministically with Helm under the pinned capability profile.
external-tls-ca binds tls.existingSecret.lookup=false, an explicit apiService.caBundle, and the target Secret requirement kube-system/metrics-server-tls.
Each variant has exactly 9 Helm objects and a rendered APIService.
cub installer package produces byte-identical bundles across two local runs.
cub installer setup --base default and cub installer setup --base external-tls-ca match Helm semantically, plus only v1|Namespace||kube-system.
Semantic comparison prunes null fields, because cub installer/kustomize drops metadata.annotations: null from the APIService.

For the promoted ingress-nginx proof:

Both variants render deterministically with Helm under the pinned capability profile.
default renders exactly 11 Helm objects, including the admission Service and ValidatingWebhookConfiguration.
admission-disabled renders exactly 9 Helm objects and deliberately omits the admission Service and ValidatingWebhookConfiguration.
Both variants render the controller Deployment and nginx IngressClass.
cub installer package produces byte-identical bundles across two local runs.
cub installer setup --base default and cub installer setup --base admission-disabled match Helm semantically, plus only v1|Namespace||ingress-nginx.
Scan/gate receipts flag admission webhook observation, Helm hook lifecycle policy, and cluster RBAC where applicable.

For the promoted cert-manager proof:

Both variants render deterministically with Helm under the pinned capability profile.
default renders exactly 42 Helm objects and zero CRDs.
crds-enabled renders exactly 48 Helm objects, including the six cert-manager CRDs.
Both variants render the controller, cainjector, webhook Deployment, webhook Service, MutatingWebhookConfiguration, and ValidatingWebhookConfiguration.
cub installer package produces byte-identical bundles across two local runs.
cub installer setup --base default and cub installer setup --base crds-enabled match Helm semantically, plus only v1|Namespace||cert-manager.
Scan/gate receipts flag CRD lifecycle, admission webhook observation, Helm startup hook lifecycle policy, and cluster RBAC.

For the promoted external-secrets proof:

Both variants render deterministically with Helm under the pinned capability profile.
default renders exactly 42 Helm objects, including 23 CRDs and one webhook Secret.
no-crds renders exactly 19 Helm objects, including zero CRDs and one webhook Secret.
Both variants render the controller, cert-controller, webhook Deployment, webhook Service, and both ValidatingWebhookConfigurations.
The disabled bitwarden-sdk-server dependency is recorded in dependency-lock.yaml.
cub installer package produces byte-identical bundles across two local runs.
cub installer setup --base default and cub installer setup --base no-crds match Helm semantically, plus only v1|Namespace||external-secrets, while preserving the separated webhook Secret.
Scan/gate receipts flag CRD lifecycle, admission webhook observation, webhook Secret/cert-controller observation, dependency lock review, and cluster RBAC.

For the promoted Argo CD proof:

Both variants render deterministically with Helm under the pinned capability profile.
default renders exactly 49 Helm objects, including three Argo CD CRDs and two Secrets.
no-crds renders exactly 46 Helm objects, including zero CRDs and two Secrets.
Both variants render the server, repo-server, applicationset-controller, Redis Deployment, application-controller StatefulSet, server Service, Redis Service, argocd-secret, and argocd-notifications-secret.
The disabled redis-ha dependency is recorded in dependency-lock.yaml.
cub installer package produces byte-identical bundles across two local runs.
cub installer setup --base default and cub installer setup --base no-crds match Helm semantically, plus only v1|Namespace||argocd, while preserving separated Secrets.
Scan/gate receipts flag CRD lifecycle, Helm hook lifecycle, generated Secret ownership, dependency lock review, StatefulSet policy, GitOps handoff, and cluster RBAC.

For the promoted PostgreSQL proof:

Both promoted variants render deterministically with Helm under the pinned inputs.
static-passwords renders exactly 7 Helm objects, including one Secret and one StatefulSet.
existing-secret renders exactly 6 Helm objects, including zero Secrets and one StatefulSet.
static-passwords binds auth.postgresPassword before render.
existing-secret declares target Secret postgresql/postgresql-auth.
The Bitnami common dependency is recorded in dependency-lock.yaml.
cub installer package produces byte-identical bundles across two local runs.
cub installer setup matches Helm semantically, plus only v1|Namespace||postgresql, while preserving separated Secret behavior.
Scan/gate receipts flag generated facts, target facts, Helm hook lifecycle, dependency lock review, StatefulSet/PVC policy, and extension-slot review.

For the promoted RabbitMQ proof:

Both promoted variants render deterministically with Helm under the pinned inputs.
static-passwords renders exactly 10 Helm objects, including the credential Secret, the config Secret, and one StatefulSet.
existing-secret renders exactly 9 Helm objects, including only the config Secret and one StatefulSet.
static-passwords binds auth.password and auth.erlangCookie before render.
existing-secret declares target Secrets rabbitmq/rabbitmq-auth and rabbitmq/rabbitmq-erlang-cookie.
The Bitnami common dependency is recorded in dependency-lock.yaml.
cub installer package produces byte-identical bundles across two local runs.
cub installer setup matches Helm semantically, plus only v1|Namespace||rabbitmq, while preserving separated Secret behavior.
Scan/gate receipts flag generated facts, target facts, dependency lock review, StatefulSet/PVC policy, clustering policy, and extension-slot review.

For the promoted kube-prometheus-stack proof:

Both promoted variants render deterministically with Helm under the pinned inputs.
default renders exactly 124 Helm objects, including 10 Prometheus Operator CRDs, Grafana, kube-state-metrics, node-exporter, and admission webhooks.
no-crds renders exactly 114 Helm objects and zero CRDs.
Both variants bind grafana.adminPassword before render.
The CRD, kube-state-metrics, node-exporter, Grafana, and windows-exporter dependencies are recorded in dependency-lock.yaml.
cub installer package produces byte-identical bundles across two local runs.
cub installer setup matches Helm semantically, plus only v1|Namespace||monitoring, while preserving separated Secret behavior.
Semantic comparison prunes null fields and empty metadata maps, because cub installer/kustomize drops empty annotations on Grafana dashboard ConfigMaps.
Scan/gate receipts flag CRD lifecycle, admission webhook observation, generated Grafana credential ownership, umbrella dependency review, cluster RBAC, and extension-slot review.

For the promoted Loki proof:

Both promoted variants render deterministically with Helm under the pinned inputs.
The default chart render is recorded as blocked because loki.storage.bucketNames.chunks and storage/schema values are missing.
single-binary-filesystem renders exactly 19 Helm objects, including one Loki StatefulSet, cache StatefulSets, and zero Secrets.
simple-scalable-minio renders exactly 36 Helm objects, including read, write, backend, cache, and MinIO workloads plus one rendered Secret.
The MinIO, grafana-agent-operator, and rollout-operator dependencies are recorded in dependency-lock.yaml.
cub installer package produces byte-identical bundles across two local runs.
cub installer setup matches Helm semantically, plus only v1|Namespace||loki, while preserving separated Secret behavior.
Semantic comparison classifies the single Loki ConfigMap leading blank-line normalization introduced by cub installer/kustomize.
Scan/gate receipts flag blocked default render, storage/schema selection, object-store Secret ownership, dependency review, cluster RBAC, StatefulSet policy, lifecycle policy, and extension-slot review.

For the promoted Longhorn proof:

Both promoted variants render deterministically with Helm under the pinned inputs.
default renders exactly 41 Helm objects, including 22 Longhorn CRDs, the manager DaemonSet, driver deployer, UI Deployment, cluster RBAC, and no Secrets.
ui-ingress renders exactly 42 Helm objects by adding networking.k8s.io/v1|Ingress|longhorn-system|longhorn-ingress.
Both variants render the Longhorn CRDs as ordinary digest-bound objects.
cub installer package produces byte-identical bundles across two local runs.
cub installer setup matches Helm semantically, plus only v1|Namespace||longhorn-system.
Scan/gate receipts flag CRD lifecycle, pre-upgrade hook lifecycle, admission/recovery observation, cluster RBAC, privileged storage workload policy, StorageClass/default-setting policy, and UI ingress policy.

For the promoted MySQL proof:

Both promoted variants render deterministically with Helm under the pinned inputs.
static-passwords renders exactly 8 Helm objects, including one Secret and one StatefulSet.
existing-secret renders exactly 7 Helm objects, including zero Secrets and one StatefulSet.
static-passwords binds auth.rootPassword, auth.password, and auth.replicationPassword before render.
existing-secret declares target Secret mysql/mysql-auth.
The Bitnami common dependency is recorded in dependency-lock.yaml.
cub installer package produces byte-identical bundles across two local runs.
cub installer setup matches Helm semantically, plus only v1|Namespace||mysql, while preserving separated Secret behavior.
Scan/gate receipts flag generated facts, target facts, dependency lock review, hook lifecycle, StatefulSet/PVC policy, and extension-slot review.

For the promoted Grafana proof:

Both promoted variants render deterministically with Helm under the pinned inputs.
static-passwords renders exactly 9 Helm objects, including one Secret and one Deployment.
existing-secret-ingress renders exactly 9 Helm objects, including zero Secrets, one Deployment, and one Ingress.
static-passwords binds adminPassword before render.
existing-secret-ingress declares target Secret grafana/grafana-admin.
The chart has no subchart dependencies and records an empty dependency closure.
cub installer package produces byte-identical bundles across two local runs.
cub installer setup matches Helm semantically, plus only v1|Namespace||grafana, while preserving separated Secret behavior.
Scan/gate receipts flag generated facts, target facts, deprecated chart status, RBAC review, UI ingress policy, deployment policy, and provisioning/sidecar/Secret extension slots.

For the promoted Vault proof:

Both promoted variants render deterministically with Helm under the pinned inputs.
default renders exactly 12 Helm objects, including the Vault StatefulSet, injector Deployment, injector MutatingWebhookConfiguration, and zero Secrets.
ha-raft-ui renders exactly 18 Helm objects, including HA discovery RBAC, PodDisruptionBudget, active/standby services, UI Service, and zero Secrets.
Source lock records hashicorp/vault@0.32.0, app 1.21.2, package SHA, and non-deprecated upstream status.
The chart has no subchart dependencies and records an empty dependency closure.
cub installer package produces byte-identical bundles across two local runs.
cub installer setup matches Helm semantically, plus only v1|Namespace||vault, while preserving separated Secret behavior.
Scan/gate receipts flag TLS posture, injector webhook, RBAC, service exposure, Vault StatefulSet storage/init/unseal operations, and Secret/env extension slots.

For the promoted Secrets Store CSI Driver proof:

Both promoted variants render deterministically with Helm under the pinned inputs.
default renders exactly 10 Helm objects, including two CRDs, one DaemonSet, one CSIDriver object, cluster RBAC, and zero Secrets.
sync-secret-rotation renders exactly 12 Helm objects by adding synced Secret RBAC and driver args for rotation and provider health checks.
Source lock records secrets-store-csi-driver/secrets-store-csi-driver@1.6.0, app 1.6.0, package SHA, and non-deprecated upstream status.
The chart has no subchart dependencies and records an empty dependency closure.
cub installer package produces byte-identical bundles across two local runs.
cub installer setup matches Helm semantically, plus only v1|Namespace||kube-system, while preserving separated Secret behavior.
Scan/gate receipts flag CRD lifecycle, CSIDriver lifecycle, privileged-node DaemonSet behavior, RBAC, synced Secret ownership, rotation, provider health checks, and provider identity inputs.

For the promoted Prometheus proof:

Both promoted variants render deterministically with Helm under the pinned inputs.
default renders exactly 23 Helm objects, including Prometheus server, Alertmanager, kube-state-metrics, node-exporter, pushgateway, server PVC, services, and cluster RBAC.
server-only-ephemeral renders exactly 6 Helm objects by disabling bundled components and the server PVC.
Source lock records prometheus-community/prometheus@29.8.0, app v3.11.3, package SHA, and non-deprecated upstream status.
Dependency lock records the four bundled subcharts: alertmanager, kube-state-metrics, prometheus-node-exporter, and prometheus-pushgateway.
cub installer package produces byte-identical bundles across two local runs.
cub installer setup matches Helm semantically, plus only v1|Namespace||monitoring, while preserving separated Secret behavior.
Scan/gate receipts flag bundled component ownership, scrape config, storage retention, workload rollout, cluster RBAC, remote read/write, ingress, network policy, PDB, and extra-manifest extension slots.

For the promoted MongoDB proof:

Both promoted variants render deterministically with Helm under the pinned inputs.
static-passwords renders exactly 8 Helm objects, including one Secret, Deployment, PVC, NetworkPolicy, and PDB.
existing-secret-replicaset renders exactly 10 Helm objects, including no Secret, a primary StatefulSet, an arbiter StatefulSet, and headless services.
Dependency lock records the Bitnami common subchart.
cub installer package produces byte-identical bundles across two local runs.
cub installer setup matches Helm semantically, plus only v1|Namespace||mongodb, while preserving separated Secret behavior.
Scan/gate receipts flag generated facts, target facts, replica-set topology, storage, NetworkPolicy/PDB, hook lifecycle, dependency lock, and extension slots.

For the promoted Nginx proof:

Both promoted variants render deterministically with Helm under the pinned inputs.
http-clusterip renders exactly 5 Helm objects, including Deployment, Service, ServiceAccount, NetworkPolicy, and PDB, with no rendered Secret.
existing-tls-ingress renders exactly 6 Helm objects, adding Ingress while still rendering no Secret because TLS material is a target fact.
Dependency lock records the Bitnami common subchart.
cub installer package produces byte-identical bundles across two local runs.
cub installer setup matches Helm semantically, plus only v1|Namespace||nginx.
Scan/gate receipts flag generated TLS handling, target TLS Secrets, ingress exposure, NetworkPolicy/PDB, service exposure, static-site supply-chain, metrics add-ons, and raw/template extension slots.

For the promoted Tempo proof:

Both promoted variants render deterministically with Helm under the pinned inputs.
local-persistent renders exactly 4 Helm objects: StatefulSet, ConfigMap, ServiceAccount, and Service.
s3-query-observability renders exactly 8 Helm objects, adding NetworkPolicy, query ConfigMap, Ingress, and ServiceMonitor while rendering no Secret because S3 credentials are a target fact.
Source lock records grafana/tempo@1.24.4, app 2.9.0, package SHA, and deprecated upstream status.
Dependency lock records an empty dependency closure.
cub installer package produces byte-identical bundles across two local runs.
cub installer setup matches Helm semantically, plus only v1|Namespace||tempo.
Scan/gate receipts flag chart deprecation, storage backend policy, target S3 credentials, query ingress, ServiceMonitor capability, NetworkPolicy, StatefulSet/headless-Service runtime risk, and raw/template extension slots.

For the promoted Consul proof:

Both promoted variants render deterministically with Helm under the pinned inputs.
default-control-plane renders exactly 70 Helm objects, including 28 CRDs, server StatefulSet, injector/webhook-cert-manager Deployments, injector webhooks, services, PDBs, and cluster RBAC.
secure-mesh-existing-secrets renders exactly 99 Helm objects, including 28 CRDs, one rendered auth-method Secret, ACL init Job, mesh/ingress/ terminating gateway Deployments, and UI Ingress.
Source lock records hashicorp/consul@2.0.0, app 2.0.0, package SHA, and non-deprecated upstream status.
Dependency lock records an empty dependency closure.
cub installer package produces byte-identical bundles across two local runs.
cub installer setup matches Helm semantically, plus only v1|Namespace||consul, with the explicit scalar normalization trim-leading-command-block-newline.
Scan/gate receipts flag CRD ownership, cluster RBAC, admission webhooks, TLS/ACL/gossip target Secrets, gateway topology, UI ingress, lifecycle Job, rendered Secret ownership, StatefulSet storage, and raw/template extension slots.

Negative Golden Check

The verifier must include self-tests that corrupt known-good fixtures and prove verification fails for the expected reason.

Current self-tests include:

corrupt the Redis installer package source and require a package source SHA mismatch;
corrupt the Redis rendered object digest and require rejection;
remove the namespace support classification and require rejection;
claim false scan success and require rejection;
reintroduce the old standalone variant shape and require rejection;
tamper with the reuse-existing-secret target fact and require rejection;
lie about the variant diff and require rejection;
corrupt the metrics-server rendered object set and require rejection;
corrupt the ingress-nginx rendered object set and require rejection;
corrupt the cert-manager rendered object set and require rejection;
corrupt the external-secrets rendered object set and require rejection;
corrupt the Argo CD rendered object set and require rejection;
corrupt the PostgreSQL rendered object set and require rejection;
corrupt the RabbitMQ rendered object set and require rejection;
corrupt the kube-prometheus-stack rendered object set and require rejection;
corrupt the Loki rendered object set and require rejection;
corrupt the Longhorn rendered object set and require rejection;
corrupt the MySQL rendered object set and require rejection;
corrupt the Grafana rendered object set and require rejection;
corrupt the Vault rendered object set and require rejection;
corrupt the Secrets Store CSI Driver rendered object set and require rejection;
corrupt the Prometheus rendered object set and require rejection;
corrupt the MongoDB rendered object set and require rejection;
corrupt the Nginx rendered object set and require rejection;
corrupt the Tempo rendered object set and require rejection;
corrupt the Consul rendered object set and require rejection;
corrupt an adversarial harness rendered manifest and require a rendered manifest SHA mismatch.

The Redis installer package verifier must also prove:

every package source file SHA and byte count matches the package receipt;
cub installer package produces byte-identical bundles across two local runs;
the package bundle SHA matches the receipt;
cub installer setup --base default matches the default Helm-equivalent variant, plus only v1|Namespace||redis;
cub installer setup --base reuse-existing-secret matches the existing-secret Helm-equivalent variant, plus only v1|Namespace||redis.

Non-Scope For V0

The current verifier does not prove:

source chart archive bytes beyond the recorded source locks and package receipts;
hosted ConfigHub upload/OCI state in the default local npm run verify, even though the latest remote receipt is recorded under runs/redis-confighub/latest/upload-oci-receipt.yaml;
upload receipts, upgrade simulation receipts, or rollback simulation receipts;
full JSON Schema enforcement, because the immediate gate is hash/reference integrity over existing artifacts.
full recipe/variant/revision proofs for every chart in data/adversarial10/; that harness is the first generated readiness and blocker map, not the final product proof for those charts.

Extension Rule

Every new proof artifact family must add:

schema or contract file
positive fixture or generated corpus
negative golden check
receipt/reference/digest verification
npm script entry point

Do not add a new chart proof stage unless its verifier contract lands first.