fix(federation): use node http healthcheck and any-restart for gateway test stack

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

.gitignore

@@ -9,3 +9,6 @@ coverage
 *.tsbuildinfo
 .pnpm-store
 docs/reports/
+
+# Step-CA dev password — real file is gitignored; commit only the .example
+infra/step-ca/dev-password

apps/gateway/src/federation/scope-schema.spec.ts

@@ -0,0 +1,187 @@
+/**
+ * Unit tests for FederationScopeSchema and parseFederationScope.
+ *
+ * Coverage:
+ *  - Valid: minimal scope
+ *  - Valid: full PRD §8.1 example
+ *  - Valid: resources + excluded_resources (no overlap)
+ *  - Invalid: empty resources
+ *  - Invalid: unknown resource value
+ *  - Invalid: resources / excluded_resources intersection
+ *  - Invalid: filter key not in resources
+ *  - Invalid: max_rows_per_query = 0
+ *  - Invalid: max_rows_per_query = 10001
+ *  - Invalid: not an object / null
+ *  - Defaults: include_personal defaults to true; excluded_resources defaults to []
+ *  - Sentinel: console.warn fires for sensitive resources
+ */
+
+import { describe, it, expect, vi, afterEach } from 'vitest';
+import {
+  parseFederationScope,
+  FederationScopeError,
+  FederationScopeSchema,
+} from './scope-schema.js';
+
+afterEach(() => {
+  vi.restoreAllMocks();
+});
+
+describe('parseFederationScope — valid inputs', () => {
+  it('accepts a minimal scope (resources + max_rows_per_query only)', () => {
+    const scope = parseFederationScope({
+      resources: ['tasks'],
+      max_rows_per_query: 100,
+    });
+    expect(scope.resources).toEqual(['tasks']);
+    expect(scope.max_rows_per_query).toBe(100);
+    expect(scope.excluded_resources).toEqual([]);
+    expect(scope.filters).toBeUndefined();
+  });
+
+  it('accepts the full PRD §8.1 example', () => {
+    const scope = parseFederationScope({
+      resources: ['tasks', 'notes', 'memory'],
+      filters: {
+        tasks: { include_teams: ['team_uuid_1', 'team_uuid_2'], include_personal: true },
+        notes: { include_personal: true, include_teams: [] },
+        memory: { include_personal: true },
+      },
+      excluded_resources: ['credentials', 'api_keys'],
+      max_rows_per_query: 500,
+    });
+    expect(scope.resources).toEqual(['tasks', 'notes', 'memory']);
+    expect(scope.excluded_resources).toEqual(['credentials', 'api_keys']);
+    expect(scope.filters?.tasks?.include_teams).toEqual(['team_uuid_1', 'team_uuid_2']);
+    expect(scope.max_rows_per_query).toBe(500);
+  });
+
+  it('accepts a scope with excluded_resources and no filter overlap', () => {
+    const scope = parseFederationScope({
+      resources: ['tasks', 'notes'],
+      excluded_resources: ['memory'],
+      max_rows_per_query: 250,
+    });
+    expect(scope.resources).toEqual(['tasks', 'notes']);
+    expect(scope.excluded_resources).toEqual(['memory']);
+  });
+});
+
+describe('parseFederationScope — defaults', () => {
+  it('defaults excluded_resources to []', () => {
+    const scope = parseFederationScope({ resources: ['tasks'], max_rows_per_query: 1 });
+    expect(scope.excluded_resources).toEqual([]);
+  });
+
+  it('defaults include_personal to true when filter is provided without it', () => {
+    const scope = parseFederationScope({
+      resources: ['tasks'],
+      filters: { tasks: { include_teams: ['t1'] } },
+      max_rows_per_query: 10,
+    });
+    expect(scope.filters?.tasks?.include_personal).toBe(true);
+  });
+});
+
+describe('parseFederationScope — invalid inputs', () => {
+  it('throws FederationScopeError for empty resources array', () => {
+    expect(() => parseFederationScope({ resources: [], max_rows_per_query: 100 })).toThrow(
+      FederationScopeError,
+    );
+  });
+
+  it('throws for unknown resource value in resources', () => {
+    expect(() =>
+      parseFederationScope({ resources: ['unknown_resource'], max_rows_per_query: 100 }),
+    ).toThrow(FederationScopeError);
+  });
+
+  it('throws when resources and excluded_resources intersect', () => {
+    expect(() =>
+      parseFederationScope({
+        resources: ['tasks', 'memory'],
+        excluded_resources: ['memory'],
+        max_rows_per_query: 100,
+      }),
+    ).toThrow(FederationScopeError);
+  });
+
+  it('throws when filters references a resource not in resources', () => {
+    expect(() =>
+      parseFederationScope({
+        resources: ['tasks'],
+        filters: { notes: { include_personal: true } },
+        max_rows_per_query: 100,
+      }),
+    ).toThrow(FederationScopeError);
+  });
+
+  it('throws for max_rows_per_query = 0', () => {
+    expect(() => parseFederationScope({ resources: ['tasks'], max_rows_per_query: 0 })).toThrow(
+      FederationScopeError,
+    );
+  });
+
+  it('throws for max_rows_per_query = 10001', () => {
+    expect(() => parseFederationScope({ resources: ['tasks'], max_rows_per_query: 10001 })).toThrow(
+      FederationScopeError,
+    );
+  });
+
+  it('throws for null input', () => {
+    expect(() => parseFederationScope(null)).toThrow(FederationScopeError);
+  });
+
+  it('throws for non-object input (string)', () => {
+    expect(() => parseFederationScope('not-an-object')).toThrow(FederationScopeError);
+  });
+});
+
+describe('parseFederationScope — sentinel warning', () => {
+  it('emits console.warn when resources includes "credentials"', () => {
+    const warnSpy = vi.spyOn(console, 'warn').mockImplementation(() => {});
+    parseFederationScope({
+      resources: ['tasks', 'credentials'],
+      max_rows_per_query: 100,
+    });
+    expect(warnSpy).toHaveBeenCalledWith(
+      expect.stringContaining(
+        '[FederationScope] WARNING: scope grants sensitive resource "credentials"',
+      ),
+    );
+  });
+
+  it('emits console.warn when resources includes "api_keys"', () => {
+    const warnSpy = vi.spyOn(console, 'warn').mockImplementation(() => {});
+    parseFederationScope({
+      resources: ['tasks', 'api_keys'],
+      max_rows_per_query: 100,
+    });
+    expect(warnSpy).toHaveBeenCalledWith(
+      expect.stringContaining(
+        '[FederationScope] WARNING: scope grants sensitive resource "api_keys"',
+      ),
+    );
+  });
+
+  it('does NOT emit console.warn for non-sensitive resources', () => {
+    const warnSpy = vi.spyOn(console, 'warn').mockImplementation(() => {});
+    parseFederationScope({ resources: ['tasks', 'notes', 'memory'], max_rows_per_query: 100 });
+    expect(warnSpy).not.toHaveBeenCalled();
+  });
+});
+
+describe('FederationScopeSchema — boundary values', () => {
+  it('accepts max_rows_per_query = 1 (lower bound)', () => {
+    const result = FederationScopeSchema.safeParse({ resources: ['tasks'], max_rows_per_query: 1 });
+    expect(result.success).toBe(true);
+  });
+
+  it('accepts max_rows_per_query = 10000 (upper bound)', () => {
+    const result = FederationScopeSchema.safeParse({
+      resources: ['tasks'],
+      max_rows_per_query: 10000,
+    });
+    expect(result.success).toBe(true);
+  });
+});

apps/gateway/src/federation/scope-schema.ts

@@ -0,0 +1,147 @@
+/**
+ * Federation grant scope schema and validator.
+ *
+ * Source of truth: docs/federation/PRD.md §8.1
+ *
+ * This module is intentionally pure — no DB, no NestJS, no CA wiring.
+ * It is reusable from grant CRUD (M2-06) and scope enforcement (M3+).
+ */
+
+import { z } from 'zod';
+
+// ---------------------------------------------------------------------------
+// Allowlist of federation resources (canonical — M3+ will extend this list)
+// ---------------------------------------------------------------------------
+export const FEDERATION_RESOURCE_VALUES = [
+  'tasks',
+  'notes',
+  'memory',
+  'credentials',
+  'api_keys',
+] as const;
+
+export type FederationResource = (typeof FEDERATION_RESOURCE_VALUES)[number];
+
+/**
+ * Sensitive resources require explicit admin approval (PRD §8.4).
+ * The parser warns when these appear in `resources`; M2-06 grant CRUD
+ * will add a hard gate on top of this warning.
+ */
+const SENSITIVE_RESOURCES: ReadonlySet<FederationResource> = new Set(['credentials', 'api_keys']);
+
+// ---------------------------------------------------------------------------
+// Sub-schemas
+// ---------------------------------------------------------------------------
+
+const ResourceArraySchema = z
+  .array(z.enum(FEDERATION_RESOURCE_VALUES))
+  .nonempty({ message: 'resources must contain at least one value' })
+  .refine((arr) => new Set(arr).size === arr.length, {
+    message: 'resources must not contain duplicate values',
+  });
+
+const ResourceFilterSchema = z.object({
+  include_teams: z.array(z.string()).optional(),
+  include_personal: z.boolean().default(true),
+});
+
+// ---------------------------------------------------------------------------
+// Top-level schema
+// ---------------------------------------------------------------------------
+
+export const FederationScopeSchema = z
+  .object({
+    resources: ResourceArraySchema,
+
+    excluded_resources: z
+      .array(z.enum(FEDERATION_RESOURCE_VALUES))
+      .default([])
+      .refine((arr) => new Set(arr).size === arr.length, {
+        message: 'excluded_resources must not contain duplicate values',
+      }),
+
+    filters: z.record(z.string(), ResourceFilterSchema).optional(),
+
+    max_rows_per_query: z
+      .number()
+      .int({ message: 'max_rows_per_query must be an integer' })
+      .min(1, { message: 'max_rows_per_query must be at least 1' })
+      .max(10000, { message: 'max_rows_per_query must be at most 10000' }),
+  })
+  .superRefine((data, ctx) => {
+    const resourceSet = new Set(data.resources);
+
+    // Intersection guard: a resource cannot be both granted and excluded
+    for (const r of data.excluded_resources) {
+      if (resourceSet.has(r)) {
+        ctx.addIssue({
+          code: z.ZodIssueCode.custom,
+          message: `Resource "${r}" appears in both resources and excluded_resources`,
+          path: ['excluded_resources'],
+        });
+      }
+    }
+
+    // Filter keys must be a subset of resources
+    if (data.filters) {
+      for (const key of Object.keys(data.filters)) {
+        if (!resourceSet.has(key as FederationResource)) {
+          ctx.addIssue({
+            code: z.ZodIssueCode.custom,
+            message: `filters key "${key}" references a resource not present in resources`,
+            path: ['filters', key],
+          });
+        }
+      }
+    }
+  });
+
+export type FederationScope = z.infer<typeof FederationScopeSchema>;
+
+// ---------------------------------------------------------------------------
+// Error class
+// ---------------------------------------------------------------------------
+
+export class FederationScopeError extends Error {
+  constructor(message: string) {
+    super(message);
+    this.name = 'FederationScopeError';
+  }
+}
+
+// ---------------------------------------------------------------------------
+// Typed parser
+// ---------------------------------------------------------------------------
+
+/**
+ * Parse and validate an unknown value as a FederationScope.
+ *
+ * Throws `FederationScopeError` with aggregated Zod issues on failure.
+ *
+ * Emits `console.warn` when sensitive resources (`credentials`, `api_keys`)
+ * are present in `resources` — per PRD §8.4, these require explicit admin
+ * approval. M2-06 grant CRUD will add a hard gate on top of this warning.
+ */
+export function parseFederationScope(input: unknown): FederationScope {
+  const result = FederationScopeSchema.safeParse(input);
+
+  if (!result.success) {
+    const issues = result.error.issues
+      .map((e) => `  - [${e.path.join('.') || 'root'}] ${e.message}`)
+      .join('\n');
+    throw new FederationScopeError(`Invalid federation scope:\n${issues}`);
+  }
+
+  const scope = result.data;
+
+  // Sentinel warning for sensitive resources (PRD §8.4)
+  for (const resource of scope.resources) {
+    if (SENSITIVE_RESOURCES.has(resource)) {
+      console.warn(
+        `[FederationScope] WARNING: scope grants sensitive resource "${resource}". Per PRD §8.4 this requires explicit admin approval and is logged.`,
+      );
+    }
+  }
+
+  return scope;
+}

deploy/portainer/federated-test.stack.yml

@@ -30,9 +30,18 @@
 #   DNS A record        ${HOST_FQDN} → Swarm ingress IP (or Cloudflare proxy).
 #
 # IMAGE
-#   Pinned to digest fed-v0.1.0-m1 (DEPLOY-01 verified).
+#   Pinned to sha-9f1a081 (main HEAD post-#488 Dockerfile fix). The previous
+#   pin (fed-v0.1.0-m1, sha256:9b72e2...) had a broken pnpm copy and could
+#   not resolve @mosaicstack/storage at runtime. The new digest was smoke-
+#   tested locally — gateway boots, imports resolve, tier-detector runs.
 #   Update digest here when promoting a new build.
 #
+# HEALTHCHECK NOTE (2026-04-21)
+#   Switched from busybox wget to node http.get on 127.0.0.1 (not localhost) to
+#   avoid IPv6 resolution issues on Alpine. Retries increased to 5 and
+#   start_period to 60s to cover the NestJS/GC cold-start window (~40-50s).
+#   restart_policy set to `any` so SIGTERM/clean-exit also triggers restart.
+#
 # NOTE: This is a TEST template — production deployments use a separate
 #       parameterised template with stricter resource limits and secrets.
 
@@ -40,8 +49,8 @@ version: '3.9'
 
 services:
   gateway:
-    image: git.mosaicstack.dev/mosaicstack/stack/gateway@sha256:9b72e202a9eecc27d31920b87b475b9e96e483c0323acc57856be4b1355db1ec
+    image: git.mosaicstack.dev/mosaicstack/stack/gateway@sha256:1069117740e00ccfeba357cae38c43f3729fe5ae702740ce474f6512414d7c02
-    # Tag for human reference: fed-v0.1.0-m1
+    # Tag for human reference: sha-9f1a081 (post-#488 Dockerfile fix; smoke-tested locally)
     environment:
       # ── Tier ───────────────────────────────────────────────────────────────
       MOSAIC_TIER: federated
@@ -73,7 +82,7 @@ services:
     deploy:
       replicas: 1
       restart_policy:
-        condition: on-failure
+        condition: any
         delay: 5s
         max_attempts: 3
       labels:
@@ -85,11 +94,15 @@ services:
         - 'traefik.http.routers.${STACK_NAME}.tls.certresolver=letsencrypt'
         - 'traefik.http.services.${STACK_NAME}.loadbalancer.server.port=3000'
     healthcheck:
-      test: ['CMD', 'wget', '-qO-', 'http://localhost:3000/health']
+      test:
+        - 'CMD'
+        - 'node'
+        - '-e'
+        - "require('http').get('http://127.0.0.1:3000/health',r=>process.exit(r.statusCode===200?0:1)).on('error',()=>process.exit(1))"
       interval: 30s
       timeout: 5s
-      retries: 3
+      retries: 5
-      start_period: 20s
+      start_period: 60s
     depends_on:
       - postgres
       - valkey

docker-compose.federated.yml

@@ -27,6 +27,7 @@ services:
   postgres-federated:
     image: pgvector/pgvector:pg17
     profiles: [federated]
+    restart: unless-stopped
     ports:
       - '${PG_FEDERATED_HOST_PORT:-5433}:5432'
     environment:
@@ -45,6 +46,7 @@ services:
   valkey-federated:
     image: valkey/valkey:8-alpine
     profiles: [federated]
+    restart: unless-stopped
     ports:
       - '${VALKEY_FEDERATED_HOST_PORT:-6380}:6379'
     volumes:
@@ -55,6 +57,64 @@ services:
       timeout: 3s
       retries: 5
 
+  # ---------------------------------------------------------------------------
+  # Step-CA — Mosaic Federation internal certificate authority
+  #
+  # Image: pinned to 0.27.4 (latest stable as of late 2025).
+  # `latest` is forbidden per Mosaic image policy (immutable tag required for
+  # reproducible deployments and digest-first promotion in CI).
+  #
+  # Profile: `federated` — this service must not start in non-federated dev.
+  #
+  # Password:
+  #   Dev:  bind-mount ./infra/step-ca/dev-password (gitignored; copy from
+  #         ./infra/step-ca/dev-password.example and customise locally).
+  #   Prod: replace the bind-mount with a Docker secret:
+  #           secrets:
+  #             ca_password:
+  #               external: true
+  #         and reference it as `/run/secrets/ca_password` (same path the
+  #         init script already uses).
+  #
+  # Provisioner: "mosaic-fed" (consumed by apps/gateway/src/federation/ca.service.ts)
+  # ---------------------------------------------------------------------------
+  step-ca:
+    image: smallstep/step-ca:0.27.4
+    profiles: [federated]
+    restart: unless-stopped
+    ports:
+      - '${STEP_CA_HOST_PORT:-9000}:9000'
+    volumes:
+      - step_ca_data:/home/step
+      # init script — executed as the container entrypoint
+      - ./infra/step-ca/init.sh:/usr/local/bin/mosaic-step-ca-init.sh:ro
+      # X.509 template skeleton (wired in M2-04)
+      - ./infra/step-ca/templates:/etc/step-ca-templates:ro
+      # Dev password file — GITIGNORED; copy from dev-password.example
+      # In production, replace this with a Docker secret (see comment above).
+      - ./infra/step-ca/dev-password:/run/secrets/ca_password:ro
+    entrypoint: ['/bin/sh', '/usr/local/bin/mosaic-step-ca-init.sh']
+    healthcheck:
+      # The healthcheck requires the root cert to exist, which is only true
+      # after init.sh has completed on first boot. start_period gives init
+      # time to finish before Docker starts counting retries.
+      test:
+        [
+          'CMD',
+          'step',
+          'ca',
+          'health',
+          '--ca-url',
+          'https://localhost:9000',
+          '--root',
+          '/home/step/certs/root_ca.crt',
+        ]
+      interval: 10s
+      timeout: 5s
+      retries: 5
+      start_period: 30s
+
 volumes:
   pg_federated_data:
   valkey_federated_data:
+  step_ca_data:

docs/federation/TASKS.md

@@ -64,11 +64,11 @@ Goal: Two federated-tier gateways stood up on Portainer at `mos-test-1.woltje.co
 Goal: An admin can create a federation grant; counterparty enrolls; cert is signed by Step-CA with SAN OIDs for `grantId` + `subjectUserId`. No runtime federation traffic flows yet (that's M3).
 
 | id        | status      | description                                                                                                                                                                                      | issue | agent  | branch                             | depends_on       | estimate | notes                                                                                                                                                                                                                                                                                                                                                                                                 |
-| --------- | ----------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | ----- | ------ | ---------------------------------- | ---------------- | -------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| --------- | ----------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | ----- | ------ | ---------------------------------- | ---------------- | -------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
 | FED-M2-01 | needs-qa    | DB migration: `federation_grants`, `federation_peers`, `federation_audit_log` tables + enum types (`grant_status`, `peer_state`). Drizzle schema + migration generation; migration tests.        | #461  | sonnet | feat/federation-m2-schema          | —                | 5K       | PR #486 open. First review NEEDS CHANGES (missing DESC indexes + reserved cols). Remediation subagent `a673dd9355dc26f82` in flight in worktree `agent-a4404ac1`.                                                                                                                                                                                                                                     |
 | FED-M2-02 | not-started | Add Step-CA sidecar to `docker-compose.federated.yml`: official `smallstep/step-ca` image, persistent CA volume, JWK provisioner config baked into init script.                                  | #461  | sonnet | feat/federation-m2-stepca          | DEPLOY-02        | 4K       | Profile-gated under `federated`. CA password from secret; dev compose uses dev-only password file.                                                                                                                                                                                                                                                                                                    |
 | FED-M2-03 | not-started | Scope JSON schema + validator: `resources` allowlist, `excluded_resources`, `include_teams`, `include_personal`, `max_rows_per_query`. Vitest unit tests for valid + invalid scopes.             | #461  | sonnet | feat/federation-m2-scope-schema    | —                | 4K       | Validator independent of CA — reusable from grant CRUD + (later) M3 scope enforcement.                                                                                                                                                                                                                                                                                                                |
-| FED-M2-04 | not-started | `apps/gateway/src/federation/ca.service.ts`: Step-CA client (CSR submission, OID-bearing cert retrieval). Mocked + integration tests against real Step-CA container.                             | #461  | sonnet | feat/federation-m2-ca-service      | M2-02            | 6K       | SAN OIDs: `grantId` (custom OID 1.3.6.1.4.1.99999.1) + `subjectUserId` (1.3.6.1.4.1.99999.2). Document OID assignments in PRD/SETUP.                              |
+| FED-M2-04 | not-started | `apps/gateway/src/federation/ca.service.ts`: Step-CA client (CSR submission, OID-bearing cert retrieval). Mocked + integration tests against real Step-CA container.                             | #461  | sonnet | feat/federation-m2-ca-service      | M2-02            | 6K       | SAN OIDs: `grantId` (custom OID 1.3.6.1.4.1.99999.1) + `subjectUserId` (1.3.6.1.4.1.99999.2). Document OID assignments in PRD/SETUP. **Acceptance**: must (a) wire `federation.tpl` template into `mosaic-fed` provisioner config and (b) include a unit/integration test asserting issued certs contain BOTH OIDs — fails-loud guard against silent OID stripping (carry-forward from M2-02 review). |
 | FED-M2-05 | not-started | Sealed storage for `client_key_pem` reusing existing `provider_credentials` sealing key. Tests prove DB-at-rest is ciphertext, not PEM. Key rotation path documented (deferred impl).            | #461  | sonnet | feat/federation-m2-key-sealing     | M2-01            | 5K       | Separate from M2-06 to keep crypto seam isolated; reviewer focus is sealing only.                                                                                                                                                                                                                                                                                                                     |
 | FED-M2-06 | not-started | `grants.service.ts`: CRUD + status transitions (`pending` → `active` → `revoked`); integrates M2-03 (scope) + M2-05 (sealing). Unit tests cover all transitions including invalid ones.          | #461  | sonnet | feat/federation-m2-grants-service  | M2-03, M2-05     | 6K       | Business logic only — CSR + cert work delegated to M2-04. Revocation handler is M6.                                                                                                                                                                                                                                                                                                                   |
 | FED-M2-07 | not-started | `enrollment.controller.ts`: short-lived single-use token endpoint; CSR signing; updates grant `pending` → `active`; emits enrollment audit (table-only write, M4 tightens).                      | #461  | sonnet | feat/federation-m2-enrollment      | M2-04, M2-06     | 6K       | Tokens single-use with 410 on replay; tokens TTL'd at 15min; rate-limited at request layer (M4 introduces guard, M2 uses simple lock).                                                                                                                                                                                                                                                                |

infra/step-ca/dev-password.example

@@ -0,0 +1 @@
+dev-only-step-ca-password-do-not-use-in-production

infra/step-ca/init.sh

@@ -0,0 +1,60 @@
+#!/bin/sh
+# infra/step-ca/init.sh
+#
+# Idempotent first-boot initialiser for the Mosaic Federation CA.
+#
+# On the first run (no /home/step/config/ca.json present) this script:
+#   1. Initialises Step-CA with a JWK provisioner named "mosaic-fed".
+#   2. Writes the CA configuration to the persistent volume at /home/step.
+#
+# On subsequent runs (config already exists) this script skips init and
+# starts the CA directly.
+#
+# The provisioner name "mosaic-fed" is consumed by:
+#   apps/gateway/src/federation/ca.service.ts  (added in M2-04)
+#
+# Password source:
+#   Dev:  mounted from ./infra/step-ca/dev-password via bind mount.
+#   Prod: mounted from a Docker secret at /run/secrets/ca_password.
+#
+# OID template:
+#   infra/step-ca/templates/federation.tpl is copied into the CA config
+#   directory so the JWK provisioner can reference it.  The template
+#   skeleton is wired in M2-04 when the CA service lands the SAN-bearing
+#   CSR work.
+
+set -e
+
+CA_CONFIG="/home/step/config/ca.json"
+PASSWORD_FILE="/run/secrets/ca_password"
+
+if [ ! -f "${CA_CONFIG}" ]; then
+  echo "[step-ca init] First boot detected — initialising Mosaic Federation CA..."
+
+  step ca init \
+    --name "Mosaic Federation CA" \
+    --dns "localhost" \
+    --dns "step-ca" \
+    --address ":9000" \
+    --provisioner "mosaic-fed" \
+    --password-file "${PASSWORD_FILE}" \
+    --provisioner-password-file "${PASSWORD_FILE}" \
+    --no-db
+
+  echo "[step-ca init] CA initialised."
+
+  # Copy the X.509 template into the Step-CA config directory so the
+  # provisioner can reference it in M2-04.
+  if [ -f "/etc/step-ca-templates/federation.tpl" ]; then
+    mkdir -p /home/step/templates
+    cp /etc/step-ca-templates/federation.tpl /home/step/templates/federation.tpl
+    echo "[step-ca init] Federation X.509 template copied to /home/step/templates/."
+  fi
+
+  echo "[step-ca init] Startup complete."
+else
+  echo "[step-ca init] Config already exists — skipping init."
+fi
+
+echo "[step-ca init] Starting Step-CA on :9000..."
+exec step-ca /home/step/config/ca.json --password-file "${PASSWORD_FILE}"

infra/step-ca/templates/federation.tpl

@@ -0,0 +1,48 @@
+{
+  "subject": {{ toJson .Subject }},
+  "sans": {{ toJson .SANs }},
+
+  {{- /*
+    Mosaic Federation X.509 Certificate Template
+    ============================================
+    This template is used by the "mosaic-fed" JWK provisioner to sign
+    federation client certificates.
+
+    Custom OID extensions (per PRD §6):
+      1.3.6.1.4.1.99999.1  — mosaic.federation.grantId   (UUID string)
+      1.3.6.1.4.1.99999.2  — mosaic.federation.subjectUserId (UUID string)
+
+    TODO (M2-04): Wire actual OID extensions below once the CA service
+    (apps/gateway/src/federation/ca.service.ts) lands the SAN-bearing CSR
+    work and the template can be exercised end-to-end.
+
+    Step-CA template reference:
+      https://smallstep.com/docs/step-ca/templates
+
+    Expected final shape of the extensions block (placeholder — not yet
+    activated):
+
+    "extensions": [
+      {
+        "id": "1.3.6.1.4.1.99999.1",
+        "critical": false,
+        "value": {{ toJson (first .Token.mosaic_grant_id) }}
+      },
+      {
+        "id": "1.3.6.1.4.1.99999.2",
+        "critical": false,
+        "value": {{ toJson (first .Token.mosaic_subject_user_id) }}
+      }
+    ],
+
+    The provisioner must pass these values in the ACME/JWK token payload
+    (token claims `mosaic_grant_id` and `mosaic_subject_user_id`) when
+    submitting the CSR.  M2-04 owns that work.
+  */ -}}
+
+  "keyUsage": ["digitalSignature"],
+  "extKeyUsage": ["clientAuth"],
+  "basicConstraints": {
+    "isCA": false
+  }
+}