net-kingdom/docs/platform-identity-security-architecture.md

# Platform Identity and Security Architecture

Status: implemented architecture baseline for NetKingdom/Railiance/Coulomb
Date: 2026-05-18

## Purpose

This document captures the production-oriented identity, authorization,
MFA, credential, and bootstrap architecture for the platform we are
building. It deliberately treats Coulomb as the first internal tenant and
reference workload, not as the platform itself.

The architecture must be recursive: the same platform that protects
future tenants also protects the services and repositories used to build
and operate the platform. That recursion is useful, but it is also where
many security designs accidentally collapse into self-administering root
power. This document exists to prevent that.

## Core Model

```text
Bootstrap plane
  establishes initial trust before normal platform services exist

Platform control plane
  operates identity, MFA, secrets, policy, audit, and authorization

Tenant planes
  run Coulomb and future customer/project/domain workloads
```

Coulomb is the first internal tenant. It is also the reference tenant that
helps validate the platform. It must not become the platform root of
trust merely because it is first.

## Planes

### Bootstrap Plane

The bootstrap plane exists before the full platform is alive. It owns the
minimal authority needed to create and recover the control plane.

Responsibilities:

- host provisioning and hardening
- root age/SOPS material and emergency bundles
- initial cluster access
- initial identity service deployment
- initial secret injection
- break-glass recovery
- transition to managed runtime authority

Owned primarily by `railiance-infra`, `railiance-cluster`, and the
credential bootstrap work in `net-kingdom`.

### Platform Control Plane

The platform control plane owns shared security services.

Responsibilities:

- NetKingdom IAM Profile
- lightweight identity mode through key-cape
- expanded identity mode through Keycloak
- MFA/token lifecycle through privacyIDEA where applicable
- canonical authorization through flex-auth
- delegated authorization runtime through Topaz first, with other PDPs as
  adapters
- runtime secret authority through OpenBao
- audit and explanation records
- platform service secrets, dynamic credentials, leases, and rotation

Owned conceptually by `net-kingdom`; deployed through the Railiance stack.

### Tenant Plane

Tenant planes are where workloads live. Coulomb is tenant zero/reference
tenant; later tenants may be projects, customers, domains, sandboxes, or
isolated deployments.

Responsibilities:

- protected services and repositories
- tenant-owned resources
- tenant-specific groups, policies, and service accounts
- local enforcement of authorization decisions
- workload audit events and diagnostics

Tenant administrators may manage their tenant resources. They must not be
able to alter platform root trust, global identity configuration,
platform break-glass material, or the policy pipeline that governs the
platform itself.

## Component Responsibilities

| Component | Primary role | Must not become |
| --- | --- | --- |
| `net-kingdom` | canonical security architecture, IAM Profile, SSO/MFA, credential bootstrap decisions | a deployment repo for every stack layer |
| `key-cape` | lightweight IAM implementation of the NetKingdom IAM Profile | a general-purpose IAM platform or authorization engine |
| Keycloak | expanded-mode IAM and optional Keycloak Authorization Services adapter | the canonical model for all platform authorization |
| privacyIDEA | MFA/token authority, especially in lightweight/key-cape mode | a policy decision point for application resources |
| OpenBao | runtime platform secrets service, dynamic credential broker, lease/revocation point, and audit source for secret access | the bootstrap root of trust or an application-specific configuration store |
| `flex-auth` | authorization control plane, CARING descriptors, policy packages, decision envelopes, audit/explain | an identity provider or backend-specific wrapper |
| Topaz | first delegated authorization runtime/PDP for flex-auth | the platform control plane or identity provider |
| Railiance repos | converged infrastructure, cluster, platform services, enablement, and app deployment | the source of security policy semantics |

## Identity Path

```text
Human/service/agent principal
        |
        v
NetKingdom IAM Profile
        |
        +-- lightweight mode: key-cape
        |       Authelia + LLDAP + privacyIDEA
        |
        +-- expanded mode: Keycloak
                Keycloak + LDAP/Entra federation + MFA integration
```

Applications depend on the IAM Profile, not on the concrete provider.
key-cape is the lightweight profile implementation. Keycloak is the
expanded-mode profile implementation. privacyIDEA provides MFA/token
capabilities where the deployment mode uses it.

Identity answers: who is this actor, how was the actor authenticated,
what coarse claims are asserted, and what assurance evidence exists?

Identity does not answer final resource-specific authorization.

## Authorization Path

```text
Identity claims from IAM Profile
        |
        v
flex-auth
  resource registry
  policy packages
  CARING descriptors
  decision/audit/explain envelope
        |
        +-- standalone evaluator
        +-- Topaz delegated PDP
        +-- optional Keycloak AuthZ adapter
        +-- future OpenFGA/SpiceDB/OPA/Cedar adapters
        |
        v
Protected service enforcement
```

Authorization answers: may this actor perform this action on this
resource in this context, and what explanation/audit/CARING metadata
supports that answer?

Protected services enforce decisions locally. flex-auth is the canonical
policy and decision boundary; delegated PDPs are runtime implementations
behind it.

## Secret And Credential Path

```text
Bootstrap SOPS/age material
        |
        v
OpenBao platform secrets service
  KV v2 platform configuration
  dynamic database credentials
  Kubernetes auth / workload identity
  future object-storage credential brokering
  audit devices and lease/revocation records
        |
        +-- direct OpenBao clients
        +-- External Secrets Operator / synced Kubernetes Secrets
        +-- CSI-mounted secrets where appropriate
        |
        v
Platform and tenant workloads
```

SOPS/age remains the bootstrap and Git-at-rest protection mechanism. It
can create the initial cluster secrets and emergency recovery bundles, but
it should not become the long-lived runtime authority for every workload
secret.

OpenBao is the runtime platform secrets service once the control plane is
alive. It owns secret leases, revocation, audit, dynamic credentials, and
workload-facing secret delivery patterns. Workloads should receive scoped
secrets or short-lived credentials, not platform-root material. Tenant
administrators may manage tenant-scoped secrets through approved policy
paths; they must not gain access to OpenBao root tokens, unseal keys,
platform mounts, or global secret engine configuration.

OpenBao does not replace identity or authorization. NetKingdom IAM
identifies actors and workloads; flex-auth decides whether a credential
or secret request is allowed; OpenBao stores, issues, audits, and revokes
the resulting secret material.

## Recursive Trust Rule

Normal tenant administration must never be sufficient to alter the
platform root of trust.

This applies even when the tenant is Coulomb. Coulomb can be a tenant and
a reference workload, but platform-root actions require platform control
plane authority and appropriate bootstrap/break-glass safeguards.

Examples of platform-root actions:

- changing IAM Profile semantics
- rotating root bootstrap keys
- changing break-glass access
- changing global MFA requirements
- activating authorization policy that governs platform administration
- changing flex-auth/Topaz policy import pipelines
- changing OpenBao root tokens, unseal policy, platform mounts, or global
  auth methods
- changing audit retention or tamper-evidence settings

## Tenant Model

Every protected resource should belong to a tenant or to the platform
control plane.

Suggested identifiers:

```text
tenant:platform          # platform control plane resources
tenant:coulomb           # first internal/reference tenant
tenant:sandbox:<name>    # sandbox tenants
tenant:customer:<name>   # future customer tenants
```

Tenant membership and platform membership are distinct. A subject may be
an administrator in `tenant:coulomb` without being a platform operator.

CARING descriptors should explicitly identify scope and tenant when the
access is tenant-scoped. Platform-scoped descriptors should be rare,
audited, and usually condition-bound.

## Bootstrap To Runtime Transition

Production setup should move through explicit trust states:

1. **Bare host trust** - provisioned and verified by Railiance infra.
2. **Cluster trust** - Kubernetes runtime exists and is verified.
3. **Bootstrap secret trust** - age/SOPS and emergency bundles are
   established.
4. **Bootstrap identity trust** - local/bootstrap identity can operate
   enough to install full identity services.
5. **Runtime secret trust** - OpenBao is deployed, initialized, unsealed,
   audited, backed up, and ready to issue scoped secrets.
6. **Runtime identity trust** - key-cape or Keycloak becomes the normal
   IAM Profile issuer.
7. **Runtime authorization trust** - flex-auth and Topaz are initialized
   with platform and tenant policies.
8. **Tenant onboarding trust** - Coulomb and later tenants register
   resources and receive scoped authority.

Each transition needs a verification check and a rollback/recovery path.

## Production Topology

For an initial production-capable Coulomb deployment:

```text
railiance-infra
  host baseline, SSH, age keys, emergency material

railiance-cluster
  Kubernetes, ingress, cert-manager, network policy

railiance-platform
  OpenBao, PostgreSQL, object storage, platform service secret delivery
  key-cape or Keycloak
  privacyIDEA where used
  flex-auth
  Topaz

railiance-apps
  Coulomb services as tenant:coulomb workloads
```

`net-kingdom` owns the architecture and standards. Railiance owns the
converged deployment layers. Component repos own their implementation
contracts.

## Orchestration Implication

A future orchestration repo may be justified, but only after the state
machine is clear. It should not own resources directly. It should own
safe sequencing across repos.

Possible responsibilities:

- verify Railiance preconditions
- initialize credential bootstrap
- deploy or validate identity services
- deploy or validate flex-auth and Topaz
- run IAM Profile conformance checks
- run authorization conformance checks
- produce a platform security readiness report

This orchestration layer should build on Railiance capabilities rather
than bypassing the Railiance stack boundaries.

ADR-0007 records the current decision: keep orchestration in Railiance
playbooks for now, with NetKingdom defining the trust-state model,
readiness checks, OpenBao boundaries, and security semantics.

## flex-auth And Topaz Implications

flex-auth work must preserve the recursive boundary between platform
control-plane resources and tenant resources.

Required implications:

- CARING descriptors must include scope and tenant metadata for
  tenant-scoped access, and must mark rare platform-scoped access
  explicitly.
- Policy packages must distinguish `tenant:platform` policy from
  tenant-local packages such as `tenant:coulomb`.
- Decision envelopes must carry subject, issuer, audience, tenant,
  protected-system id, resource, action, requested TTL where relevant,
  assurance evidence, obligations, deny reasons, and audit correlation
  ids.
- Topaz is a delegated PDP runtime behind flex-auth. It must not become
  the canonical policy model, identity provider, or platform control
  plane.
- Audit and explain records must be durable enough to reconstruct why a
  platform-root, secret, credential, or tenant-administration decision was
  allowed or denied.
- Platform-root guardrails must deny tenant administrators the ability to
  alter IAM Profile semantics, OpenBao platform mounts/auth methods,
  flex-auth policy import pipelines, Topaz runtime configuration, or
  platform audit retention.

OpenBao secret access and dynamic credential requests follow the same
authorization rule: identity proves the actor or workload, flex-auth
decides whether the request is permitted, and OpenBao stores, issues,
leases, audits, and revokes the secret material.

## Coulomb Tenant Onboarding Path

The first Coulomb tenant onboarding path should be repeatable before it
becomes automated:

1. Register `tenant:coulomb` as a tenant distinct from
   `tenant:platform`.
2. Map Coulomb human, service, and agent principals to IAM Profile claims
   with issuer, audience, subject, group, tenant, and assurance evidence.
3. Register Coulomb protected systems and resources in flex-auth with
   stable protected-system ids.
4. Import tenant-scoped policy packages and CARING descriptors for
   Coulomb resources.
5. Initialize the delegated PDP runtime, starting with Topaz, using only
   the policy packages approved for the tenant and platform boundary.
6. Provision Coulomb workload secret paths, Kubernetes auth roles, or
   delivery mechanisms in OpenBao without granting access to platform
   mounts, unseal/recovery material, or global auth configuration.
7. Run audit readiness checks before admitting production traffic:
   identity issuance, flex-auth decision envelope, Topaz health,
   OpenBao audit event, workload enforcement event, and correlation id.

The onboarding path is complete when a Coulomb workload can authenticate,
receive a scoped authorization decision, obtain only the allowed secret or
short-lived credential, enforce the decision locally, and produce an
auditable record without receiving platform-root authority.

## Production Readiness Checks

Before the security platform is production-ready, each trust state needs
an explicit check:

| Area | Readiness check |
| --- | --- |
| MFA and identity | key-cape or Keycloak issues IAM Profile-compatible tokens; privacyIDEA or the selected MFA provider enforces required assurance for privileged actions |
| Bootstrap and recovery | age/SOPS material, emergency bundle, and break-glass credentials are present, tested, and separated from tenant administration |
| OpenBao runtime secrets | OpenBao is initialized, unsealed or auto-unsealed by the approved mechanism, backed up, audited, and using scoped auth methods and mounts |
| Secret rotation | service, database, OpenBao-issued, and break-glass rotation paths have documented blast radius and verification steps |
| flex-auth policy state | platform and tenant policy packages are versioned, reviewable, imported, and explainable |
| Topaz runtime | delegated PDP health, data freshness, policy load status, and fail-closed behavior are verified |
| Tenant onboarding | `tenant:coulomb` resources, claims, policies, OpenBao paths, and audit correlation are registered and tested |
| Audit sink | identity, flex-auth, Topaz, OpenBao, Kubernetes, and workload audit records land in durable storage with restore/drill coverage |
| Break-glass | emergency access works when normal identity is unavailable and produces a post-event review record |

## Open Questions

- Where is the durable audit log stored for platform-root decisions?
- Where are OpenBao audit logs durably shipped, and how are they included
  in tamper-evidence and restore drills?
- Which actions require dual control or human confirmation?
- How is break-glass use recorded when normal identity is unavailable?
- Which workloads consume OpenBao directly, via External Secrets Operator,
  or via CSI-mounted secrets?
- Which tenant metadata is required before a service can register
  resources with flex-auth?
- When does the platform switch from key-cape lightweight mode to
  Keycloak expanded mode?
- Does Topaz run centrally for the platform, per tenant, or per service
  for the first production deployment?