@happyvertical/smrt-app-runtime 0.51.27 → 0.51.29

package/AGENTS.md

@@ -48,7 +48,10 @@ Application infrastructure composition for the validated runtime profiles in
   readiness-checked before startup succeeds.
 - `createTaskWorker()` and `createScheduleWorker()` initialize the normal jobs
   package runners against the shared PostgreSQL database. They are intended for
-  separate processes; the web process does not start them automatically.
+  separate processes; the web process does not start them automatically. The
+  worker process must import the app's compiled registration
+  (`.smrt/runtime/register.js`, #3117) first; runners resolve `objectType`
+  through the registry and never load application source themselves.
 - `close()` drains in-flight readiness/session/worker initialization and
   serialized runner start/stop operations, stops every runner returned by the
   runtime, and then closes PostgreSQL. Returned runners must not be restarted

package/README.md

@@ -163,9 +163,14 @@ provider selectors, tenancy posture, and worker topology without secret values.
 
 Job producers keep using `SmrtObject.bg()` / `background().enqueue()` in every
 profile. Deployed worker entry points initialize the ordinary runners against
-the same PostgreSQL database:
+the same PostgreSQL database. A worker is a separate Node process, so it must
+first import the application's compiled object registration —
+`.smrt/runtime/register.js`, produced by a SvelteKit `smrtPlugin()` build — or
+any job targeting an application object fails with `Unknown object type`:
 
 ```ts
+await import('./.smrt/runtime/register.js');
+
 const taskWorker = await initialized.createTaskWorker({ concurrency: 8 });
 await taskWorker.start();
 

package/dist/manifest.json

@@ -2,7 +2,7 @@
   "version": "1.0.0",
   "timestamp": 0,
   "packageName": "@happyvertical/smrt-app-runtime",
-  "packageVersion": "0.51.27",
+  "packageVersion": "0.51.29",
   "objects": {},
   "moduleType": "smrt",
   "smrtDependencies": [

package/dist/smrt-knowledge.json

@@ -3,13 +3,13 @@
   "sensitiveFieldsExcluded": true,
   "generatedAt": "1970-01-01T00:00:00.000Z",
   "packageName": "@happyvertical/smrt-app-runtime",
-  "packageVersion": "0.51.27",
+  "packageVersion": "0.51.29",
   "sourceManifestPath": "dist/manifest.json",
   "agentDocPath": "AGENTS.md",
   "sourceHashes": {
-    "manifest": "b995f0afb582e0636c4a178aeb2a801a77efc7032bf4c2c929be17c9e4555332",
-    "packageJson": "dbf607267b560cd1c6142bf37ed6b228341381b699be7122b93c8bb6d3220e54",
-    "agents": "6231c7cb6a95eb891f2e8d63623c1a6ad47b61fa886368bd45bd3e8057952b13"
+    "manifest": "1b88c0c9d79aaed44b035a026700d3f38c09db106779822c240a6e5fc1f958fd",
+    "packageJson": "f281f9b757a95ae6fedb43edac025a577e2580361a12daf3ff236e41e087a719",
+    "agents": "d8202c6d897a78eaf23c048682bd922899605e7ddad595c1341be65544b7000b"
   },
   "exports": [
     "."
@@ -63,5 +63,5 @@
     "polymorphicAssociations": 0,
     "uuidColumns": 0
   },
-  "agentDoc": "# @happyvertical/smrt-app-runtime\n\nApplication infrastructure composition for the validated runtime profiles in\n`@happyvertical/smrt-config`.\n\n## Local profile\n\n- `validateApplicationId()` is the canonical strict validator for explicit app\n  IDs. `encodeApplicationId()` derives a length-bounded, collision-resistant\n  ID from package names; generated operational surfaces must share these APIs.\n- `initializeLocalApplicationRuntime()` owns user data paths, SQLite tuning,\n  local application-secret creation, and the single-use owner bootstrap flow.\n- SQLite is acquired through `@happyvertical/sql`'s explicit\n  `{ driver: 'node:sqlite', custody: 'trusted-parent' }` boundary after the\n  runtime establishes its user-owned mode-0700 data root.\n- Application migrations are explicit through `prepareDatabase`; runtime never\n  creates application model tables implicitly. A rejected `prepareDatabase` is\n  normalized into `LocalRuntimeError` with the stable `migration_failed` code\n  and the fixed recovery message `MIGRATION_FAILED_MESSAGE` (\"run pnpm\n  app:setup and inspect the private migration logs\"). The caller's error text\n  is never surfaced — it is retained only as a non-enumerable `cause` for\n  private logs — and the same application root stays retryable.\n- Standalone migration commands must call `prepareLocalDatabaseStorage()`\n  before opening SQLite so custody is established without creating schema or\n  bootstrap records.\n- Read-only/copy/import operator commands must call\n  `validateLocalDatabaseStorage()` before opening SQLite; it verifies the\n  app-bound marker, source separation, real path chain, ownership, and mode\n  without creating or repairing storage.\n- Owner bootstrap creates normal `Person`, `User`, `Tenant`, owner `Role` /\n  `Membership`, and `Session` records in one transaction.\n- Bootstrap is loopback-only. Only an HMAC of the short-lived token is stored.\n- Background jobs and application-defined paid capabilities are default-off.\n- The embedded runner reuses `TaskRunner`; it is not a second job contract.\n\n## Deployed profiles\n\n- `initializeDeployedApplicationRuntime()` composes only `self-hosted` and\n  `cloud`; it rejects local initialization and provider/binding mismatches.\n- Provider bindings own credentials, vendor clients, and readiness checks.\n  Runtime diagnostics contain selectors and status only.\n- The database binding must expose a provider-owned `close` callback before\n  `connect` can run. The runtime owns that cleanup boundary and runs the\n  application's explicit, idempotent `prepareDatabase` hook.\n- A failed startup cleanup throws `DeployedRuntimeCleanupError`; callers retain\n  and retry its idempotent `retryCleanup()` boundary until it succeeds.\n- Public authentication, asset storage, and secret bindings are mandatory and\n  readiness-checked before startup succeeds.\n- `createTaskWorker()` and `createScheduleWorker()` initialize the normal jobs\n  package runners against the shared PostgreSQL database. They are intended for\n  separate processes; the web process does not start them automatically.\n- `close()` drains in-flight readiness/session/worker initialization and\n  serialized runner start/stop operations, stops every runner returned by the\n  runtime, and then closes PostgreSQL. Returned runners must not be restarted\n  after runtime shutdown begins.\n- A tracked operation may await a re-entrant `close()` request without\n  deadlocking; external `close()` callers still await complete cleanup.\n- `health()` is process liveness. `readiness()` probes database/auth/assets/\n  secrets; it does not claim that an external worker fleet is running.\n- Database-provider readiness is additive; a PostgreSQL-specific server-version\n  probe always gates startup and live readiness.\n- Cloud must keep required tenant context and must never introduce a root or\n  unscoped tenant fallback. RLS remains an explicit deployment/migration choice.\n\n## Public runtime diagnostics\n\n- `projectRuntimeDiagnostics()` is the only public diagnostic projection. It\n  constructs the schema-version-1 allowlist and never serializes the private\n  local or deployed diagnostics objects.\n- The allowlist is limited to active profile/coarse health, schema and\n  migration readiness, six fixed capability statuses, a sorted/capped public\n  tool inventory plus digest, explicit operational topology differences,\n  coarse worker heartbeat state, and eight stable code/timestamp errors.\n- Callers supply an explicit clock and worker-heartbeat seam. A missing\n  heartbeat is `unknown`; web-process liveness never proves worker liveness.\n- Error ingestion maps unknown codes to `runtime_error`, truncates timestamps\n  to the minute, and ignores messages, stacks, logs, records, and nested state.\n- Authentication and diagnostics-read authorization belong before every call\n  to the projector or its runtime probes.\n\n## Invariants\n\n- Never expose application-secret bytes or bootstrap token hashes in diagnostics.\n- Never include migration or provider error text in a local or deployed startup\n  error message; migration driver text is a likely credential carrier.\n- Never include provider error text, URLs, tokens, or credentials in deployed\n  errors, diagnostics, health, or readiness payloads.\n- Keep local data, assets, database, and secrets outside the source tree.\n- Do not reach into collection/database private fields. Add an upstream public\n  API if composition cannot be expressed through exported surfaces.\n- Keep profile selection out of domain models, generated APIs, MCP/WebMCP\n  surfaces, effects, approvals, and authorization records.\n\n## Validation\n\n```bash\npnpm --filter @happyvertical/smrt-app-runtime test\npnpm --filter @happyvertical/smrt-app-runtime typecheck\npnpm --filter @happyvertical/smrt-app-runtime build\n```\n"
+  "agentDoc": "# @happyvertical/smrt-app-runtime\n\nApplication infrastructure composition for the validated runtime profiles in\n`@happyvertical/smrt-config`.\n\n## Local profile\n\n- `validateApplicationId()` is the canonical strict validator for explicit app\n  IDs. `encodeApplicationId()` derives a length-bounded, collision-resistant\n  ID from package names; generated operational surfaces must share these APIs.\n- `initializeLocalApplicationRuntime()` owns user data paths, SQLite tuning,\n  local application-secret creation, and the single-use owner bootstrap flow.\n- SQLite is acquired through `@happyvertical/sql`'s explicit\n  `{ driver: 'node:sqlite', custody: 'trusted-parent' }` boundary after the\n  runtime establishes its user-owned mode-0700 data root.\n- Application migrations are explicit through `prepareDatabase`; runtime never\n  creates application model tables implicitly. A rejected `prepareDatabase` is\n  normalized into `LocalRuntimeError` with the stable `migration_failed` code\n  and the fixed recovery message `MIGRATION_FAILED_MESSAGE` (\"run pnpm\n  app:setup and inspect the private migration logs\"). The caller's error text\n  is never surfaced — it is retained only as a non-enumerable `cause` for\n  private logs — and the same application root stays retryable.\n- Standalone migration commands must call `prepareLocalDatabaseStorage()`\n  before opening SQLite so custody is established without creating schema or\n  bootstrap records.\n- Read-only/copy/import operator commands must call\n  `validateLocalDatabaseStorage()` before opening SQLite; it verifies the\n  app-bound marker, source separation, real path chain, ownership, and mode\n  without creating or repairing storage.\n- Owner bootstrap creates normal `Person`, `User`, `Tenant`, owner `Role` /\n  `Membership`, and `Session` records in one transaction.\n- Bootstrap is loopback-only. Only an HMAC of the short-lived token is stored.\n- Background jobs and application-defined paid capabilities are default-off.\n- The embedded runner reuses `TaskRunner`; it is not a second job contract.\n\n## Deployed profiles\n\n- `initializeDeployedApplicationRuntime()` composes only `self-hosted` and\n  `cloud`; it rejects local initialization and provider/binding mismatches.\n- Provider bindings own credentials, vendor clients, and readiness checks.\n  Runtime diagnostics contain selectors and status only.\n- The database binding must expose a provider-owned `close` callback before\n  `connect` can run. The runtime owns that cleanup boundary and runs the\n  application's explicit, idempotent `prepareDatabase` hook.\n- A failed startup cleanup throws `DeployedRuntimeCleanupError`; callers retain\n  and retry its idempotent `retryCleanup()` boundary until it succeeds.\n- Public authentication, asset storage, and secret bindings are mandatory and\n  readiness-checked before startup succeeds.\n- `createTaskWorker()` and `createScheduleWorker()` initialize the normal jobs\n  package runners against the shared PostgreSQL database. They are intended for\n  separate processes; the web process does not start them automatically. The\n  worker process must import the app's compiled registration\n  (`.smrt/runtime/register.js`, #3117) first; runners resolve `objectType`\n  through the registry and never load application source themselves.\n- `close()` drains in-flight readiness/session/worker initialization and\n  serialized runner start/stop operations, stops every runner returned by the\n  runtime, and then closes PostgreSQL. Returned runners must not be restarted\n  after runtime shutdown begins.\n- A tracked operation may await a re-entrant `close()` request without\n  deadlocking; external `close()` callers still await complete cleanup.\n- `health()` is process liveness. `readiness()` probes database/auth/assets/\n  secrets; it does not claim that an external worker fleet is running.\n- Database-provider readiness is additive; a PostgreSQL-specific server-version\n  probe always gates startup and live readiness.\n- Cloud must keep required tenant context and must never introduce a root or\n  unscoped tenant fallback. RLS remains an explicit deployment/migration choice.\n\n## Public runtime diagnostics\n\n- `projectRuntimeDiagnostics()` is the only public diagnostic projection. It\n  constructs the schema-version-1 allowlist and never serializes the private\n  local or deployed diagnostics objects.\n- The allowlist is limited to active profile/coarse health, schema and\n  migration readiness, six fixed capability statuses, a sorted/capped public\n  tool inventory plus digest, explicit operational topology differences,\n  coarse worker heartbeat state, and eight stable code/timestamp errors.\n- Callers supply an explicit clock and worker-heartbeat seam. A missing\n  heartbeat is `unknown`; web-process liveness never proves worker liveness.\n- Error ingestion maps unknown codes to `runtime_error`, truncates timestamps\n  to the minute, and ignores messages, stacks, logs, records, and nested state.\n- Authentication and diagnostics-read authorization belong before every call\n  to the projector or its runtime probes.\n\n## Invariants\n\n- Never expose application-secret bytes or bootstrap token hashes in diagnostics.\n- Never include migration or provider error text in a local or deployed startup\n  error message; migration driver text is a likely credential carrier.\n- Never include provider error text, URLs, tokens, or credentials in deployed\n  errors, diagnostics, health, or readiness payloads.\n- Keep local data, assets, database, and secrets outside the source tree.\n- Do not reach into collection/database private fields. Add an upstream public\n  API if composition cannot be expressed through exported surfaces.\n- Keep profile selection out of domain models, generated APIs, MCP/WebMCP\n  surfaces, effects, approvals, and authorization records.\n\n## Validation\n\n```bash\npnpm --filter @happyvertical/smrt-app-runtime test\npnpm --filter @happyvertical/smrt-app-runtime typecheck\npnpm --filter @happyvertical/smrt-app-runtime build\n```\n"
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@happyvertical/smrt-app-runtime",
-  "version": "0.51.27",
+  "version": "0.51.29",
   "description": "Validated local, self-hosted, and cloud runtime composition for s-m-r-t applications",
   "type": "module",
   "main": "./dist/index.js",
@@ -17,16 +17,16 @@
     "AGENTS.md"
   ],
   "dependencies": {
-    "@happyvertical/smrt-config": "0.51.27",
-    "@happyvertical/smrt-core": "0.51.27",
-    "@happyvertical/smrt-jobs": "0.51.27",
-    "@happyvertical/smrt-profiles": "0.51.27",
-    "@happyvertical/smrt-tenancy": "0.51.27",
-    "@happyvertical/smrt-users": "0.51.27",
+    "@happyvertical/smrt-config": "0.51.29",
+    "@happyvertical/smrt-core": "0.51.29",
+    "@happyvertical/smrt-jobs": "0.51.29",
+    "@happyvertical/smrt-profiles": "0.51.29",
+    "@happyvertical/smrt-tenancy": "0.51.29",
+    "@happyvertical/smrt-users": "0.51.29",
     "@happyvertical/sql": "^0.90.0"
   },
   "devDependencies": {
-    "@happyvertical/smrt-vitest": "0.51.27",
+    "@happyvertical/smrt-vitest": "0.51.29",
     "@types/node": "24.13.6",
     "typescript": "5.9.3",
     "vite": "8.1.4",