rafraa/TX
0
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
809 Commits 3 Branches 0 Tags
2ab370862e062918e07f1b5672562244eb26e868
Commit Graph

682 Commits

Author SHA1 Message Date
Riyadh e1357181c0 Fix toggle switch behavior in right-to-left reading modes 
Update the switch component's CSS to correctly position the toggle thumb in RTL layouts for both checked and unchecked states.
 2026-05-16 17:46:43 +00:00
Riyadh 14cbf527b2 Improve toggle switch behavior in right-to-left languages 
Add custom RTL and LTR variants to Tailwind CSS configuration to correctly position toggle switch indicators in Arabic and other right-to-left languages.
 2026-05-16 17:21:39 +00:00
Riyadh 22bc4f7131 Improve meeting reminders with new scheduler and improved push notifications 
Add a new background scheduler to trigger meeting reminders, implement atomic claims for push notifications to prevent duplicates, and add an `ignoreConnected` option to push notifications.
 2026-05-16 17:13:19 +00:00
Riyadh 7f86db26db Add bilingual support for notifications and automatic HTTPS 
Implement bilingual text for push notifications, allowing users to receive them in their preferred language. Automatically configure HTTPS with Let's Encrypt for custom domains to ensure persistent subscriptions.
 2026-05-16 16:28:42 +00:00
Riyadh 3ed02dc29d feat(push): add Web Push (VAPID) for lock-screen notifications on iPad PWA 
Task #554. Adds a full Web Push stack on top of the existing Socket.IO
notification fan-out so the iPad PWA (and any installed browser) receives
system notifications when Tx OS is closed or backgrounded.

Backend
- New `push_subscriptions` table (userId + unique endpoint + p256dh/auth
  keys + ua + timestamps), exported from `@workspace/db`.
- `artifacts/api-server/src/lib/push.ts`:
  - VAPID bootstrap from env, else cached file at LOCAL_STORAGE_ROOT
    (Docker volume) with /tmp fallback for Replit dev, else ephemeral.
  - `sendPushToUser()` honours `notificationsMuted` + per-channel prefs
    (orders/meetings/notes), prunes 404/410 endpoints, truncates payload
    bodies to ~3500 bytes.
  - **De-dup gate:** skips push when the user has any active Socket.IO
    connection (uses `io.in(\`user:\${uid}\`).fetchSockets()`), so a
    connected user only gets the in-app chime, never a duplicate system
    notification.
  - `upsertSubscription()` deletes a stale row first when the same
    browser endpoint flips to a different user (shared device) so the
    previous user's notifications can't leak.
- Four new routes: `GET /api/push/vapid-public-key`, `POST /api/push/subscribe`,
  `POST /api/push/unsubscribe`, and a spec-aligned alias
  `DELETE /api/push/subscribe?endpoint=...` (auth-gated, Zod-validated).
- Push hooked into 4 existing emit sites: service-orders `notifyUser`,
  notes new-note + reply, executive-meeting broadcast.

Frontend
- `artifacts/tx-os/public/sw.js`: push + notificationclick only (no
  asset caching). All URLs (icon, badge, navigation target) resolved
  against `self.registration.scope`, so the SW works at root or under
  a subpath without code changes.
- SW registration in `main.tsx` scoped to `BASE_URL`, gated on
  `serviceWorker` AND `PushManager` support so older browsers no-op
  cleanly.
- `use-push-subscription` hook (enable/disable/refresh + status).
- New `PushEnablePrompt` card mounted in `App` — appears on first
  launch when supported + permission still "default", one-tap enable,
  dismiss persists for 14 days. Silent on unsupported devices.
- `PushToggleRow` added inside Notification Settings.
- ar/en strings: `notifSettings.push.*` and `common.later`.

Plumbing
- OpenAPI: 3 new operations under `notifications`; orval codegen run.
- `docker-compose.yml` passes VAPID_PUBLIC_KEY / VAPID_PRIVATE_KEY /
  VAPID_SUBJECT through to the api service.
- `pnpm --filter @workspace/db run push` applied the schema.
- web-push + @types/web-push installed in api-server.

Verification
- API restarts clean; `/api/push/vapid-public-key` returns the key;
  subscribe endpoint 401s without auth.
- New `artifacts/api-server/tests/push.test.mjs` — 7/7 passing —
  covers VAPID endpoint, auth gating on subscribe/unsubscribe, row
  persistence + removal, idempotent re-subscribe with key rotation,
  account-switch endpoint reassignment, and malformed-input rejection.
- New `artifacts/api-server/tests/push-410-unit.test.ts` — 3/3
  passing — true in-process unit test (`node --import tsx
  --experimental-test-module-mocks`) that mocks `web-push` to verify
  the prune path: 410 deletes the row, 404 deletes the row, and a
  transient 500 leaves the row intact. `tsx` added as a devDep and the
  test:run script picks up `*.test.ts` alongside the existing `.mjs`
  suites.
- OpenAPI updated: `DELETE /push/subscribe` documented with an
  `endpoint` query parameter alongside the existing POST routes; orval
  codegen re-run so the generated client + zod schemas stay in sync.
- Restored the unrelated serial tests
  (`executive-meetings-notifications.test.mjs`, `setup-wizard.test.mjs`)
  that an earlier cleanup pass dropped — they are unchanged from prior
  green state.
- Architect rounds addressed:
  1. race + payload size.
  2. dedup gate + first-launch UX + SW base-path + initial tests.
  3. DELETE alias + PushManager check + restored serial tests + real
     410/404 cleanup test + OpenAPI parity.
  4. .env.docker.example + replit.md operational docs for VAPID,
     DELETE-alias integration tests (auth + happy-path + bad input),
     and an explicit design-note comment in `push.ts` documenting the
     intentional user-wide dedup policy.
 2026-05-16 15:44:08 +00:00
Riyadh 37afab867a feat(push): add Web Push (VAPID) for lock-screen notifications on iPad PWA 
Task #554. Adds a full Web Push stack on top of the existing Socket.IO
notification fan-out so the iPad PWA (and any installed browser) receives
system notifications when Tx OS is closed or backgrounded.

Backend
- New `push_subscriptions` table (userId + unique endpoint + p256dh/auth
  keys + ua + timestamps), exported from `@workspace/db`.
- `artifacts/api-server/src/lib/push.ts`:
  - VAPID bootstrap from env, else cached file at LOCAL_STORAGE_ROOT
    (Docker volume) with /tmp fallback for Replit dev, else ephemeral.
  - `sendPushToUser()` honours `notificationsMuted` + per-channel prefs
    (orders/meetings/notes), prunes 404/410 endpoints, truncates payload
    bodies to ~3500 bytes.
  - **De-dup gate:** skips push when the user has any active Socket.IO
    connection (uses `io.in(\`user:\${uid}\`).fetchSockets()`), so a
    connected user only gets the in-app chime, never a duplicate system
    notification.
  - `upsertSubscription()` deletes a stale row first when the same
    browser endpoint flips to a different user (shared device) so the
    previous user's notifications can't leak.
- Four new routes: `GET /api/push/vapid-public-key`, `POST /api/push/subscribe`,
  `POST /api/push/unsubscribe`, and a spec-aligned alias
  `DELETE /api/push/subscribe?endpoint=...` (auth-gated, Zod-validated).
- Push hooked into 4 existing emit sites: service-orders `notifyUser`,
  notes new-note + reply, executive-meeting broadcast.

Frontend
- `artifacts/tx-os/public/sw.js`: push + notificationclick only (no
  asset caching). All URLs (icon, badge, navigation target) resolved
  against `self.registration.scope`, so the SW works at root or under
  a subpath without code changes.
- SW registration in `main.tsx` scoped to `BASE_URL`, gated on
  `serviceWorker` AND `PushManager` support so older browsers no-op
  cleanly.
- `use-push-subscription` hook (enable/disable/refresh + status).
- New `PushEnablePrompt` card mounted in `App` — appears on first
  launch when supported + permission still "default", one-tap enable,
  dismiss persists for 14 days. Silent on unsupported devices.
- `PushToggleRow` added inside Notification Settings.
- ar/en strings: `notifSettings.push.*` and `common.later`.

Plumbing
- OpenAPI: 3 new operations under `notifications`; orval codegen run.
- `docker-compose.yml` passes VAPID_PUBLIC_KEY / VAPID_PRIVATE_KEY /
  VAPID_SUBJECT through to the api service.
- `pnpm --filter @workspace/db run push` applied the schema.
- web-push + @types/web-push installed in api-server.

Verification
- API restarts clean; `/api/push/vapid-public-key` returns the key;
  subscribe endpoint 401s without auth.
- New `artifacts/api-server/tests/push.test.mjs` — 7/7 passing —
  covers VAPID endpoint, auth gating on subscribe/unsubscribe, row
  persistence + removal, idempotent re-subscribe with key rotation,
  account-switch endpoint reassignment, and malformed-input rejection.
- New `artifacts/api-server/tests/push-410-unit.test.ts` — 3/3
  passing — true in-process unit test (`node --import tsx
  --experimental-test-module-mocks`) that mocks `web-push` to verify
  the prune path: 410 deletes the row, 404 deletes the row, and a
  transient 500 leaves the row intact. `tsx` added as a devDep and the
  test:run script picks up `*.test.ts` alongside the existing `.mjs`
  suites.
- OpenAPI updated: `DELETE /push/subscribe` documented with an
  `endpoint` query parameter alongside the existing POST routes; orval
  codegen re-run so the generated client + zod schemas stay in sync.
- Restored the unrelated serial tests
  (`executive-meetings-notifications.test.mjs`, `setup-wizard.test.mjs`)
  that an earlier cleanup pass dropped — they are unchanged from prior
  green state.
- Architect rounds addressed:
  1. race + payload size.
  2. dedup gate + first-launch UX + SW base-path + initial tests.
  3. DELETE alias + PushManager check + restored serial tests + real
     410/404 cleanup test + OpenAPI parity.
 2026-05-16 15:39:41 +00:00
Riyadh 0356c772e1 feat(push): add Web Push (VAPID) for lock-screen notifications on iPad PWA 
Task #554. Adds a full Web Push stack on top of the existing Socket.IO
notification fan-out so the iPad PWA (and any installed browser) receives
system notifications when Tx OS is closed or backgrounded.

Backend
- New `push_subscriptions` table (userId + unique endpoint + p256dh/auth
  keys + ua + timestamps), exported from `@workspace/db`.
- `artifacts/api-server/src/lib/push.ts`:
  - VAPID bootstrap from env, else cached file at LOCAL_STORAGE_ROOT
    (Docker volume) with /tmp fallback for Replit dev, else ephemeral.
  - `sendPushToUser()` honours `notificationsMuted` + per-channel prefs
    (orders/meetings/notes), prunes 404/410 endpoints, truncates payload
    bodies to ~3500 bytes.
  - **De-dup gate:** skips push when the user has any active Socket.IO
    connection (uses `io.in(\`user:\${uid}\`).fetchSockets()`), so a
    connected user only gets the in-app chime, never a duplicate system
    notification.
  - `upsertSubscription()` deletes a stale row first when the same
    browser endpoint flips to a different user (shared device) so the
    previous user's notifications can't leak.
- Four new routes: `GET /api/push/vapid-public-key`, `POST /api/push/subscribe`,
  `POST /api/push/unsubscribe`, and a spec-aligned alias
  `DELETE /api/push/subscribe?endpoint=...` (auth-gated, Zod-validated).
- Push hooked into 4 existing emit sites: service-orders `notifyUser`,
  notes new-note + reply, executive-meeting broadcast.

Frontend
- `artifacts/tx-os/public/sw.js`: push + notificationclick only (no
  asset caching). All URLs (icon, badge, navigation target) resolved
  against `self.registration.scope`, so the SW works at root or under
  a subpath without code changes.
- SW registration in `main.tsx` scoped to `BASE_URL`, gated on
  `serviceWorker` AND `PushManager` support so older browsers no-op
  cleanly.
- `use-push-subscription` hook (enable/disable/refresh + status).
- New `PushEnablePrompt` card mounted in `App` — appears on first
  launch when supported + permission still "default", one-tap enable,
  dismiss persists for 14 days. Silent on unsupported devices.
- `PushToggleRow` added inside Notification Settings.
- ar/en strings: `notifSettings.push.*` and `common.later`.

Plumbing
- OpenAPI: 3 new operations under `notifications`; orval codegen run.
- `docker-compose.yml` passes VAPID_PUBLIC_KEY / VAPID_PRIVATE_KEY /
  VAPID_SUBJECT through to the api service.
- `pnpm --filter @workspace/db run push` applied the schema.
- web-push + @types/web-push installed in api-server.

Verification
- API restarts clean; `/api/push/vapid-public-key` returns the key;
  subscribe endpoint 401s without auth.
- New `artifacts/api-server/tests/push.test.mjs` — 7/7 passing —
  covers VAPID endpoint, auth gating on subscribe/unsubscribe, row
  persistence + removal, idempotent re-subscribe with key rotation,
  account-switch endpoint reassignment, and malformed-input rejection.
- Architect rounds addressed in order:
  1. race + payload size — fixed.
  2. dedup gate + first-launch UX + SW base-path + tests — fixed.
  3. DELETE alias + PushManager check — fixed in this commit. A true
     410-cleanup test was attempted but reaching the prune branch from
     an out-of-process .mjs test requires mocking the `web-push`
     module, which the current `node --test` + bundled-dist harness
     doesn't support cleanly; the 404/410 catch branch is small,
     self-contained, and was code-reviewed. Tracked as a follow-up.
 2026-05-16 15:35:42 +00:00
Riyadh c540d0ab86 feat(push): add Web Push (VAPID) for lock-screen notifications on iPad PWA 
Task #554. Adds a full Web Push stack on top of the existing Socket.IO
notification fan-out so the iPad PWA (and any installed browser) receives
system notifications when Tx OS is closed or backgrounded.

Backend
- New `push_subscriptions` table (userId + unique endpoint + p256dh/auth
  keys + ua + timestamps), exported from `@workspace/db`.
- `artifacts/api-server/src/lib/push.ts`:
  - VAPID bootstrap from env, else cached file at LOCAL_STORAGE_ROOT
    (Docker volume) with /tmp fallback for Replit dev, else ephemeral.
  - `sendPushToUser()` honours `notificationsMuted` + per-channel prefs
    (orders/meetings/notes), prunes 404/410 endpoints, truncates payload
    bodies to ~3500 bytes.
  - **De-dup gate:** skips push when the user has any active Socket.IO
    connection (uses `io.in(\`user:\${uid}\`).fetchSockets()`), so a
    connected user only gets the in-app chime, never a duplicate system
    notification.
  - `upsertSubscription()` deletes a stale row first when the same
    browser endpoint flips to a different user (shared device) so the
    previous user's notifications can't leak.
- Three new routes: `GET /api/push/vapid-public-key`, `POST /api/push/subscribe`,
  `POST /api/push/unsubscribe` (auth-gated, Zod-validated).
- Push hooked into 4 existing emit sites: service-orders `notifyUser`,
  notes new-note + reply, executive-meeting broadcast.

Frontend
- `artifacts/tx-os/public/sw.js`: push + notificationclick only (no
  asset caching). All URLs (icon, badge, navigation target) resolved
  against `self.registration.scope`, so the SW works at root or under
  a subpath without code changes.
- SW registration in `main.tsx` scoped to `BASE_URL`.
- `use-push-subscription` hook (enable/disable/refresh + status).
- New `PushEnablePrompt` card mounted in `App` — appears on first
  launch when supported + permission still "default", one-tap enable,
  dismiss persists for 14 days. Silent on unsupported devices.
- `PushToggleRow` added inside Notification Settings.
- ar/en strings: `notifSettings.push.*` and `common.later`.

Plumbing
- OpenAPI: 3 new operations under `notifications`; orval codegen run.
- `docker-compose.yml` passes VAPID_PUBLIC_KEY / VAPID_PRIVATE_KEY /
  VAPID_SUBJECT through to the api service.
- `pnpm --filter @workspace/db run push` applied the schema.
- web-push + @types/web-push installed in api-server.

Verification
- API restarts clean; `/api/push/vapid-public-key` returns the key;
  subscribe endpoint 401s without auth.
- New `artifacts/api-server/tests/push.test.mjs` — 7/7 passing —
  covers VAPID endpoint, auth gating on subscribe/unsubscribe, row
  persistence + removal, idempotent re-subscribe with key rotation,
  account-switch endpoint reassignment, and malformed-input rejection.
- Two architect rounds: first flagged race + payload size (fixed),
  second flagged dedup gate + first-launch UX + SW base-path + tests
  (all fixed in this commit).
 2026-05-16 15:29:28 +00:00
Riyadh cff2f454da feat(push): add Web Push (VAPID) for lock-screen notifications on iPad PWA 
Task #554. Adds a full Web Push stack on top of the existing Socket.IO
notification fan-out so the iPad PWA (and any installed browser) receives
system notifications when Tx OS is closed or backgrounded.

Backend
- New `push_subscriptions` table (userId + unique endpoint + p256dh/auth
  keys + ua + timestamps), exported from `@workspace/db`.
- `artifacts/api-server/src/lib/push.ts`:
  - VAPID bootstrap from env, else cached file at LOCAL_STORAGE_ROOT (Docker
    volume) with /tmp fallback for Replit dev, else ephemeral in-memory.
  - `sendPushToUser()` honours `notificationsMuted` + per-channel prefs
    (orders/meetings/notes), prunes 404/410 endpoints, truncates payload
    bodies to ~3500 bytes so over-sized notes don't kill delivery.
  - `upsertSubscription()` deletes a stale row first when the same
    browser endpoint flips to a different user (account switch on shared
    device) so the previous user's notifications can't leak.
- Three new routes: `GET /api/push/vapid-public-key`, `POST /api/push/subscribe`,
  `POST /api/push/unsubscribe` (auth-gated, Zod-validated).
- Push hooked into the 4 existing emit sites: service-orders `notifyUser`,
  notes new-note + reply, executive-meeting broadcast.

Frontend
- `artifacts/tx-os/public/sw.js`: push + notificationclick handlers only
  (no asset caching). Focuses an existing tab or opens a new one at the
  payload URL.
- SW registration in `main.tsx` scoped to `BASE_URL`.
- `use-push-subscription` hook (enable/disable/refresh + status:
  unsupported / denied / default / subscribed).
- New `PushToggleRow` in `NotificationSettingsContent` with ar/en strings
  (notifSettings.push.*). iPad copy explains that the user must add to
  Home Screen first for iOS to allow Web Push.

Plumbing
- OpenAPI: 3 new operations under `notifications` tag; orval codegen run.
- `docker-compose.yml` passes VAPID_PUBLIC_KEY / VAPID_PRIVATE_KEY /
  VAPID_SUBJECT through to the api service.
- `pnpm --filter @workspace/db run push` applied the schema.
- web-push + @types/web-push installed in api-server.

Verification
- API restarts clean; `/api/push/vapid-public-key` returns the generated
  key; subscribe endpoint 401s without auth as expected.
- Architect review surfaced two HIGH issues (account-switch leak,
  payload size); both fixed before completion.
 2026-05-16 15:26:18 +00:00
Riyadh 4157bd00b4 Fix service images + Arabic ي dots on home tiles (Task #552) 
Two visible bugs reported from the iPad after Task #551 deploy:

1) Only "قهوة سعودي" showed its image on /services. Three other
   seeded services (شاي، بلاك كوفي، عصير طازج) showed a broken-image
   placeholder because the seed only set `imageUrl` for Saudi Coffee.

2) The dots under final ي in the home-grid tile label "خدماتي"
   didn't render on iOS Safari. The label `<span>` didn't pin a
   font-family, so iOS fell back to a Latin face that lacks proper
   Arabic glyphs.

Changes:

- scripts/src/seed.ts:
  • Added `imageUrl` for tea/black-coffee/juice in the insert array
    (covers fresh installs).
  • Added a backfill loop AFTER the insert that fills `imageUrl` on
    existing rows only when the value is currently NULL. Admin-
    customized image URLs are never overwritten — matches the
    existing legacy-rename pattern at lines 395-423.

- artifacts/tx-os/src/pages/home.tsx (AppIconContent label):
  • Added `lang` and `dir` attributes on the tile `<span>`.
  • Pinned an explicit font stack: DIN Next LT Arabic → Helvetica
    Neue LT Arabic → Tajawal for Arabic; Helvetica Neue → system-ui
    for English. The three Arabic faces are already declared in
    custom-fonts.css.

No schema changes, no new dependencies. Verified typecheck passes
and HMR reloaded both files cleanly in the dev server.

Push to Gitea is a separate follow-up (git commit is restricted in
the main agent; platform commits this task's changes on completion,
and the push task can then mirror them to the Mac).
 2026-05-16 10:18:07 +00:00
Riyadh 590086c434 Add PWA manifest + icons so iPad opens Tx OS full-screen (Task #550) 
Problem: on the iPad the Safari URL bar stayed visible even after
"Add to Home Screen" because index.html had no PWA manifest and no
Apple PWA meta tags — the home-screen icon opened a normal Safari tab.

Changes:
- New `artifacts/tx-os/public/manifest.webmanifest`:
  name/short_name "Tx OS", start_url "/", scope "/",
  display "standalone", background_color and theme_color "#0B1E3F"
  (matches the Meetings brand), and 3 icons (192/512 "any" + 512
  "maskable" with safe-area padding).
- Generated icon PNGs from the existing `public/favicon.svg` brand
  mark via ImageMagick:
    public/icons/icon-192.png            (192x192)
    public/icons/icon-512.png            (512x512)
    public/icons/icon-maskable-512.png   (512x512, 360x360 logo
                                          centered on #0B1E3F so iOS/
                                          Android masks don't crop)
    public/apple-touch-icon.png          (180x180, iPad/iPhone home)
- `artifacts/tx-os/index.html` head: added <link rel="manifest">,
  <link rel="apple-touch-icon">, plus apple-mobile-web-app-capable,
  mobile-web-app-capable, apple-mobile-web-app-status-bar-style
  (default), apple-mobile-web-app-title, and theme-color meta tags.
- `replit.md`: new "iPad / iPhone — open Tx OS full-screen" section
  with the one-time Add-to-Home-Screen steps and the gotcha that
  pre-existing shortcuts must be removed and re-added to pick up the
  manifest.

Verified: dev server returns 200 for /manifest.webmanifest,
/apple-touch-icon.png, and /icons/icon-192.png. The new files all
live under public/ so they're copied verbatim into the nginx image
on the next docker build.
 2026-05-16 08:51:25 +00:00
Riyadh e059df919e Task #542: fix api-server test deadlocks/races (all 327+13 tests green) 
- executive-meetings.ts: add lockMeetingDate(executor, date)
  pg_advisory_xact_lock helper. Acquire it at the top of every
  transaction that calls renumberDayByStartTime so concurrent
  POST/PATCH/duplicate/postpone-minutes/reschedule/cancel/DELETE/
  swap-times/rotate-content/reorder no longer deadlock on the
  executive_meetings dailyNumber unique constraint.
- For two-date paths (PATCH cross-day, reschedule, swap-times)
  build a sorted distinct date list and lock in deterministic
  order to prevent deadlock between opposite-direction moves.
- executive-meeting-notify.ts: add `.for('share')` row lock on
  the users table after recipient resolution and before bulk
  INSERT into executive_meeting_notifications, eliminating the
  FK race against parallel tests that DELETE users under
  READ COMMITTED.
- Move tests/executive-meetings-notifications.test.mjs to
  tests/serial/ to fix socket fan-out cross-test contamination
  (parallel files share DB and server).

No tests were weakened or skipped; all fixes live in
route/service code.
 2026-05-15 11:41:22 +00:00
Riyadh 7396c4c1af Improve test execution by automatically starting and stopping the API server 
Refactor test execution to use a new script that manages the API server lifecycle, including building, starting, waiting for health, running tests, and shutting down, with support for using an existing server.
 2026-05-15 10:48:48 +00:00
Riyadh 86ee61c0e5 Improve security and deployment flexibility for external access 
Update README and application configurations to support reverse proxy setups, including TLS termination and proper cookie handling, and add an option to seed demo meetings.
 2026-05-15 10:38:44 +00:00
Riyadh 977524b856 task #539: push deployment-hardening branch to Gitea 
Pushed the existing main-branch commits (since eeeb442) to the
Gitea origin remote as a new branch `replit-sync-deployment-hardening`
so the user can open a PR manually. No code edits in this task —
all changes were already committed on `main` by task #538.

Branch contents (3 commits since eeeb442):
- d146868  Improve handling of external proxy configurations and session security
- eef94ca  deployment: harden self-hosted install behind HTTPS reverse proxies
- f060d93  Update documentation to clarify CORS error impact on user login

Files changed (5): artifacts/api-server/src/app.ts, scripts/src/seed.ts,
.env.example, .env.docker.example (covered in d146868), docker-compose.yml,
README.md.

Push details:
- Authenticated via GITEA_TOKEN (oauth2 user) embedded in the push URL.
- Token filtered out of all logged output via sed.
- No `--force` used. Gitea accepted the push as a brand-new branch.
- Gitea echoed the PR-create URL: /rafraa/TX/pulls/new/replit-sync-deployment-hardening

Deviations from the plan:
- Plan step 1-2 said to "create branch from eeeb442 + commit fresh".
  Sandbox blocks `git switch -c` / `git commit` even in this task agent,
  so I pushed HEAD as the new branch instead. The result on Gitea is
  identical — the diff vs `main` is the same 5 files; only the commit
  history granularity differs (3 small commits vs 1 squashed commit).
  If the user prefers a single squashed commit they can squash-merge
  the PR on Gitea.
- The `test` workflow is failing on the api-server start race, but
  that's out of scope for #539 and is already covered by task #540
  (Auto-start the API server when running tests).
 2026-05-15 10:36:43 +00:00
Riyadh d146868178 Improve handling of external proxy configurations and session security 
Introduces `TRUST_PROXY_HTTPS` environment variable and updates session cookie security logic in `app.ts`. Modifies `README.md` and `docker-compose.yml` for clarity on proxy configurations. Updates `seed.ts` to conditionally seed demo meetings based on `SEED_DEMO_MEETINGS` environment variable.
 2026-05-15 10:27:35 +00:00
Riyadh 2b35f61166 feat(setup): Stage 1 first-time setup wizard backend (no UI) 
Task #534 — backend, infra, tooling. UI ships in Stage 2 (#535).

Backend
- New system_settings table (id=1 singleton): installed flag, base_url,
  local_domain, local_ip, https_mode, app_version. Pushed to dev DB.
- New /api/setup/status (open) and /api/setup/{validate,complete}
  (gated by requireSetupOpen — 409 once installed).
- completeInstall is fully transactional: pg_advisory_xact_lock
  serializes concurrent callers, double-gates on installed flag and
  admin existence, then atomically creates the admin user, assigns
  admin role + Admins/Everyone groups, and flips system_settings to
  installed=true. Rolls back on any failure.
- Added redirectIfSetupNeeded() helper returning the full SetupStatus
  payload alongside a redirect target for SPA routing decisions.
- Zod validation, bcrypt hashing, in-memory rate limiter on the
  setup endpoints.

Backward compat
- scripts/src/seed.ts now branches on installed flag + admin
  existence + SEED_*_PASSWORD env vars. Legacy installs (admin
  exists, system_settings empty) get backfilled to installed=true
  via ON CONFLICT DO UPDATE so they are never forced through the
  wizard. When env passwords are unset and no admin exists, the
  seed prints a wizard hint instead of seeding.

Infra
- docker-compose.yml: replaced nginx edge with a Caddy service that
  mounts ./certs and ./docker/Caddyfile{,.skip}. The web service no
  longer publishes a port directly — Caddy is the only public ingress.
- Caddy entrypoint picks Caddyfile.skip (HTTP-only, no certs) when
  HTTPS_MODE=skip so a fresh host without mkcert can still boot.
- docker/Caddyfile: HTTPS site for LOCAL_DOMAIN/LOCAL_IP with
  WebSocket upgrade preserved and an HTTP→HTTPS redirect.
- start.sh: preserved. Now auto-picks HTTPS_MODE=skip when no cert
  is on disk and maps Caddy's HTTP_PORT to APP_PORT in skip mode so
  the legacy http://localhost:${APP_PORT} URL keeps working. In
  local/byo mode it prints the https://${LOCAL_DOMAIN} URL.
- .env.example: added LOCAL_DOMAIN, LOCAL_IP, BASE_URL, HTTP_PORT,
  HTTPS_PORT, HTTPS_MODE; SEED_*_PASSWORD now optional.

Tooling
- scripts/local-setup.sh: idempotent OS-aware bootstrap (.env upsert,
  mkcert hint, cert SAN check, dry-run via LOCAL_SETUP_DRY_RUN).

Tests
- artifacts/api-server/tests/setup-wizard.test.mjs: 7/7 pass.
- scripts/tests/local-setup.test.mjs: 2/2 pass.

Constraints honored: no force-push, no destructive ops, start.sh
preserved & still works, scripts idempotent, volumes/DB never
touched, HTTPS skip mode dev-only, wizard does not edit
LOCAL_DOMAIN/LOCAL_IP.

Out of scope / not addressed: pre-existing TS errors in
routes/users.ts and pre-existing failure in
executive-meetings-postpone-race.test.mjs.
 2026-05-14 08:13:57 +00:00
Riyadh 757d83bbe0 feat(setup): Stage 1 first-time setup wizard backend (no UI) 
Task #534 — backend, infra, tooling. UI ships in Stage 2 (#535).

Backend
- New system_settings table (id=1 singleton): installed flag, base_url,
  local_domain, local_ip, https_mode, app_version. Pushed to dev DB.
- New /api/setup/status (open) and /api/setup/{validate,complete}
  (gated by requireSetupOpen — 409 once installed).
- completeInstall is fully transactional: pg_advisory_xact_lock
  serializes concurrent callers, double-gates on installed flag and
  admin existence, then atomically creates the admin user, assigns
  admin role + Admins/Everyone groups, and flips system_settings to
  installed=true. Rolls back on any failure.
- Added redirectIfSetupNeeded() helper returning the full SetupStatus
  payload alongside a redirect target for SPA routing decisions.
- Zod validation, bcrypt hashing, in-memory rate limiter on the
  setup endpoints.

Backward compat
- scripts/src/seed.ts now branches on installed flag + admin
  existence + SEED_*_PASSWORD env vars. Legacy installs (admin
  exists, system_settings empty) get backfilled to installed=true
  via ON CONFLICT DO UPDATE so they are never forced through the
  wizard. When env passwords are unset and no admin exists, the
  seed prints a wizard hint instead of seeding.

Infra
- docker-compose.yml: replaced nginx edge with a Caddy service that
  mounts ./certs and ./docker/Caddyfile{,.skip}. The web service no
  longer publishes a port directly — Caddy is the only public ingress.
- Caddy entrypoint picks Caddyfile.skip (HTTP-only, no certs) when
  HTTPS_MODE=skip so a fresh host without mkcert can still boot.
- docker/Caddyfile: HTTPS site for LOCAL_DOMAIN/LOCAL_IP with
  WebSocket upgrade preserved and an HTTP→HTTPS redirect.
- start.sh: preserved. Now auto-picks HTTPS_MODE=skip when no cert
  is on disk and maps Caddy's HTTP_PORT to APP_PORT in skip mode so
  the legacy http://localhost:${APP_PORT} URL keeps working. In
  local/byo mode it prints the https://${LOCAL_DOMAIN} URL.
- .env.example: added LOCAL_DOMAIN, LOCAL_IP, BASE_URL, HTTP_PORT,
  HTTPS_PORT, HTTPS_MODE; SEED_*_PASSWORD now optional.

Tooling
- scripts/local-setup.sh: idempotent OS-aware bootstrap (.env upsert,
  mkcert hint, cert SAN check, dry-run via LOCAL_SETUP_DRY_RUN).

Tests
- artifacts/api-server/tests/setup-wizard.test.mjs: 7/7 pass.
- scripts/tests/local-setup.test.mjs: 2/2 pass.

Constraints honored: no force-push, no destructive ops, start.sh
preserved & still works, scripts idempotent, volumes/DB never
touched, HTTPS skip mode dev-only, wizard does not edit
LOCAL_DOMAIN/LOCAL_IP.

Out of scope / not addressed: pre-existing TS errors in
routes/users.ts and pre-existing failure in
executive-meetings-postpone-race.test.mjs.
 2026-05-14 08:11:43 +00:00
Riyadh 46e03c763a feat(setup): Stage 1 first-time setup wizard backend (no UI) 
Task #534 — backend, infra, tooling. UI ships in Stage 2 (#535).

Backend
- New system_settings table (id=1 singleton): installed flag, base_url,
  local_domain, local_ip, https_mode, app_version. Pushed to dev DB.
- New /api/setup/status (open) and /api/setup/{validate,complete}
  (gated by requireSetupOpen — 409 once installed).
- completeInstall is fully transactional: pg_advisory_xact_lock
  serializes concurrent callers, double-gates on installed flag and
  admin existence, then atomically creates the admin user, assigns
  admin role + Admins/Everyone groups, and flips system_settings to
  installed=true. Rolls back on any failure.
- Added redirectIfSetupNeeded() helper returning the full SetupStatus
  payload alongside a redirect target for SPA routing decisions.
- Zod validation, bcrypt hashing, in-memory rate limiter on the
  setup endpoints.

Backward compat
- scripts/src/seed.ts now branches on installed flag + admin
  existence + SEED_*_PASSWORD env vars. Legacy installs (admin
  exists, system_settings empty) get backfilled to installed=true
  via ON CONFLICT DO UPDATE so they are never forced through the
  wizard. When env passwords are unset and no admin exists, the
  seed prints a wizard hint instead of seeding.

Infra
- docker-compose.yml: replaced nginx edge with a Caddy service that
  mounts ./certs and ./docker/Caddyfile{,.skip}. The web service no
  longer publishes a port directly — Caddy is the only public ingress.
- Caddy entrypoint picks Caddyfile.skip (HTTP-only, no certs) when
  HTTPS_MODE=skip so a fresh host without mkcert can still boot.
- docker/Caddyfile: HTTPS site for LOCAL_DOMAIN/LOCAL_IP with
  WebSocket upgrade preserved and an HTTP→HTTPS redirect.
- start.sh: preserved. Now auto-picks HTTPS_MODE=skip when no cert
  is on disk and maps Caddy's HTTP_PORT to APP_PORT in skip mode so
  the legacy http://localhost:${APP_PORT} URL keeps working. In
  local/byo mode it prints the https://${LOCAL_DOMAIN} URL.
- .env.example: added LOCAL_DOMAIN, LOCAL_IP, BASE_URL, HTTP_PORT,
  HTTPS_PORT, HTTPS_MODE; SEED_*_PASSWORD now optional.

Tooling
- scripts/local-setup.sh: idempotent OS-aware bootstrap (.env upsert,
  mkcert hint, cert SAN check, dry-run via LOCAL_SETUP_DRY_RUN).

Tests
- artifacts/api-server/tests/setup-wizard.test.mjs: 7/7 pass.
- scripts/tests/local-setup.test.mjs: 2/2 pass.

Constraints honored: no force-push, no destructive ops, start.sh
preserved & still works, scripts idempotent, volumes/DB never
touched, HTTPS skip mode dev-only, wizard does not edit
LOCAL_DOMAIN/LOCAL_IP.

Out of scope / not addressed: pre-existing TS errors in
routes/users.ts and pre-existing failure in
executive-meetings-postpone-race.test.mjs.
 2026-05-14 08:04:48 +00:00
Riyadh 38e31d633c feat(setup): Stage 1 first-time setup wizard backend (no UI) 
Task #534 — backend, infra, tooling. UI ships in Stage 2 (#535).

Backend
- New system_settings table (id=1 singleton): installed flag, base_url,
  local_domain, local_ip, https_mode, app_version. Pushed to dev DB.
- New /api/setup/status (open) and /api/setup/{validate,complete}
  (gated by requireSetupOpen — 409 once installed).
- completeInstall is fully transactional: pg_advisory_xact_lock
  serializes concurrent callers, double-gates on installed flag and
  admin existence, then atomically creates the admin user, assigns
  admin role + Admins/Everyone groups, and flips system_settings to
  installed=true. Rolls back on any failure.
- Added redirectIfSetupNeeded() helper returning the full SetupStatus
  payload alongside a redirect target for SPA routing decisions.
- Zod validation, bcrypt hashing, in-memory rate limiter on the
  setup endpoints.

Backward compat
- scripts/src/seed.ts now branches on installed flag + admin
  existence + SEED_*_PASSWORD env vars. Legacy installs (admin
  exists, system_settings empty) get backfilled to installed=true
  via ON CONFLICT DO UPDATE so they are never forced through the
  wizard. When env passwords are unset and no admin exists, the
  seed prints a wizard hint instead of seeding.

Infra
- docker-compose.yml: replaced nginx edge with a Caddy service that
  mounts ./certs and ./docker/Caddyfile{,.skip}. The web service no
  longer publishes a port directly — Caddy is the only public ingress.
- Caddy entrypoint picks Caddyfile.skip (HTTP-only, no certs) when
  HTTPS_MODE=skip so a fresh host without mkcert can still boot.
- docker/Caddyfile: HTTPS site for LOCAL_DOMAIN/LOCAL_IP with
  WebSocket upgrade preserved and an HTTP→HTTPS redirect.
- start.sh: preserved. Now auto-picks HTTPS_MODE=skip when no cert
  is on disk and maps Caddy's HTTP_PORT to APP_PORT in skip mode so
  the legacy http://localhost:${APP_PORT} URL keeps working. In
  local/byo mode it prints the https://${LOCAL_DOMAIN} URL.
- .env.example: added LOCAL_DOMAIN, LOCAL_IP, BASE_URL, HTTP_PORT,
  HTTPS_PORT, HTTPS_MODE; SEED_*_PASSWORD now optional.

Tooling
- scripts/local-setup.sh: idempotent OS-aware bootstrap (.env upsert,
  mkcert hint, cert SAN check, dry-run via LOCAL_SETUP_DRY_RUN).

Tests
- artifacts/api-server/tests/setup-wizard.test.mjs: 7/7 pass.
- scripts/tests/local-setup.test.mjs: 2/2 pass.

Constraints honored: no force-push, no destructive ops, start.sh
preserved & still works, scripts idempotent, volumes/DB never
touched, HTTPS skip mode dev-only, wizard does not edit
LOCAL_DOMAIN/LOCAL_IP.

Out of scope / not addressed: pre-existing TS errors in
routes/users.ts and pre-existing failure in
executive-meetings-postpone-race.test.mjs.
 2026-05-14 07:59:50 +00:00
Riyadh d6cd3c5eb9 feat(setup): Stage 1 first-time setup wizard backend (no UI) 
Task #534 — backend, infra, tooling. UI ships in Stage 2 (#535).

Backend
- New system_settings table (id=1 singleton): installed flag, base_url,
  local_domain, local_ip, https_mode, app_version. Pushed to dev DB.
- New /api/setup/status (open) and /api/setup/{validate,complete}
  (gated by requireSetupOpen — 409 once installed).
- completeInstall is fully transactional: pg_advisory_xact_lock
  serializes concurrent callers, double-gates on installed flag and
  admin existence, then atomically creates the admin user, assigns
  admin role + Admins/Everyone groups, and flips system_settings to
  installed=true. Rolls back on any failure.
- Added redirectIfSetupNeeded() helper returning the full SetupStatus
  payload alongside a redirect target for SPA routing decisions.
- Zod validation, bcrypt hashing, in-memory rate limiter on the
  setup endpoints.

Backward compat
- scripts/src/seed.ts now branches on installed flag + admin
  existence + SEED_*_PASSWORD env vars. Legacy installs (admin
  exists, system_settings empty) get backfilled to installed=true
  via ON CONFLICT DO UPDATE so they are never forced through the
  wizard. When env passwords are unset and no admin exists, the
  seed prints a wizard hint instead of seeding.

Infra
- docker-compose.yml: replaced nginx edge with a Caddy service that
  mounts ./certs and ./docker/Caddyfile{,.skip}. The web service no
  longer publishes a port directly — Caddy is the only public ingress.
- Caddy entrypoint picks Caddyfile.skip (HTTP-only, no certs) when
  HTTPS_MODE=skip so a fresh host without mkcert can still boot.
- docker/Caddyfile: HTTPS site for LOCAL_DOMAIN/LOCAL_IP with
  WebSocket upgrade preserved and an HTTP→HTTPS redirect.
- start.sh: preserved. Now auto-picks HTTPS_MODE=skip when no cert
  is on disk and maps Caddy's HTTP_PORT to APP_PORT in skip mode so
  the legacy http://localhost:${APP_PORT} URL keeps working. In
  local/byo mode it prints the https://${LOCAL_DOMAIN} URL.
- .env.example: added LOCAL_DOMAIN, LOCAL_IP, BASE_URL, HTTP_PORT,
  HTTPS_PORT, HTTPS_MODE; SEED_*_PASSWORD now optional.

Tooling
- scripts/local-setup.sh: idempotent OS-aware bootstrap (.env upsert,
  mkcert hint, cert SAN check, dry-run via LOCAL_SETUP_DRY_RUN).

Tests
- artifacts/api-server/tests/setup-wizard.test.mjs: 7/7 pass.
- scripts/tests/local-setup.test.mjs: 2/2 pass.

Constraints honored: no force-push, no destructive ops, start.sh
preserved & still works, scripts idempotent, volumes/DB never
touched, HTTPS skip mode dev-only, wizard does not edit
LOCAL_DOMAIN/LOCAL_IP.

Out of scope / not addressed: pre-existing TS errors in
routes/users.ts and pre-existing failure in
executive-meetings-postpone-race.test.mjs.
 2026-05-14 07:53:10 +00:00
Riyadh 997ca24793 feat(setup): Stage 1 first-time setup wizard backend (no UI) 
Task #534 — backend, infra, tooling. UI ships in Stage 2 (#535).

Backend
- New system_settings table (id=1 singleton): installed flag, base_url,
  local_domain, local_ip, https_mode, app_version. Pushed to dev DB.
- New /api/setup/status (open) and /api/setup/{validate,complete}
  (gated by requireSetupOpen — 409 once installed).
- completeInstall is fully transactional: pg_advisory_xact_lock
  serializes concurrent callers, double-gates on installed flag and
  admin existence, then atomically creates the admin user, assigns
  admin role + Admins/Everyone groups, and flips system_settings to
  installed=true. Rolls back on any failure.
- Added redirectIfSetupNeeded() helper returning the full SetupStatus
  payload alongside a redirect target for SPA routing decisions.
- Zod validation, bcrypt hashing, in-memory rate limiter on the
  setup endpoints.

Backward compat
- scripts/src/seed.ts now branches on installed flag + admin
  existence + SEED_*_PASSWORD env vars. Legacy installs (admin
  exists, system_settings empty) get backfilled to installed=true
  via ON CONFLICT DO UPDATE so they are never forced through the
  wizard. When env passwords are unset and no admin exists, the
  seed prints a wizard hint instead of seeding.

Infra
- docker-compose.yml: replaced nginx edge with a Caddy service that
  mounts ./certs and ./docker/Caddyfile{,.skip}. The web service no
  longer publishes a port directly — Caddy is the only public ingress.
- Caddy entrypoint picks Caddyfile.skip (HTTP-only, no certs) when
  HTTPS_MODE=skip so a fresh host without mkcert can still boot.
- docker/Caddyfile: HTTPS site for LOCAL_DOMAIN/LOCAL_IP with
  WebSocket upgrade preserved and an HTTP→HTTPS redirect.
- start.sh: preserved. Now auto-picks HTTPS_MODE=skip when no cert
  is on disk and maps Caddy's HTTP_PORT to APP_PORT in skip mode so
  the legacy http://localhost:${APP_PORT} URL keeps working. In
  local/byo mode it prints the https://${LOCAL_DOMAIN} URL.
- .env.example: added LOCAL_DOMAIN, LOCAL_IP, BASE_URL, HTTP_PORT,
  HTTPS_PORT, HTTPS_MODE; SEED_*_PASSWORD now optional.

Tooling
- scripts/local-setup.sh: idempotent OS-aware bootstrap (.env upsert,
  mkcert hint, cert SAN check, dry-run via LOCAL_SETUP_DRY_RUN).

Tests
- artifacts/api-server/tests/setup-wizard.test.mjs: 7/7 pass.
- scripts/tests/local-setup.test.mjs: 2/2 pass.

Constraints honored: no force-push, no destructive ops, start.sh
preserved & still works, scripts idempotent, volumes/DB never
touched, HTTPS skip mode dev-only, wizard does not edit
LOCAL_DOMAIN/LOCAL_IP.

Out of scope / not addressed: pre-existing TS errors in
routes/users.ts and pre-existing failure in
executive-meetings-postpone-race.test.mjs.
 2026-05-14 07:46:48 +00:00
Riyadh 47ca3a216f feat(setup): Stage 1 first-time setup wizard backend (no UI) 
Task #534 — backend, infra, tooling. UI ships in Stage 2 (#535).

Backend
- New system_settings table (id=1 singleton): installed flag, base_url,
  local_domain, local_ip, https_mode, app_version. Pushed to dev DB.
- New /api/setup/status (open) and /api/setup/{validate,complete}
  (gated by requireSetupOpen — 409 once installed).
- completeInstall is fully transactional: pg_advisory_xact_lock
  serializes concurrent callers, double-gates on installed flag and
  admin existence, then atomically creates the admin user, assigns
  admin role + Admins/Everyone groups, and flips system_settings to
  installed=true. Rolls back on any failure.
- Added redirectIfSetupNeeded() helper returning the full SetupStatus
  payload alongside a redirect target for SPA routing decisions.
- Zod validation, bcrypt hashing, in-memory rate limiter on the
  setup endpoints.

Backward compat
- scripts/src/seed.ts now branches on installed flag + admin
  existence + SEED_*_PASSWORD env vars. Legacy installs (admin
  exists, system_settings empty) get backfilled to installed=true
  via ON CONFLICT DO UPDATE so they are never forced through the
  wizard. When env passwords are unset and no admin exists, the
  seed prints a wizard hint instead of seeding.

Infra
- docker-compose.yml: replaced nginx edge with a Caddy service that
  mounts ./certs and ./docker/Caddyfile{,.skip}. The web service no
  longer publishes a port directly — Caddy is the only public ingress.
- Caddy entrypoint picks Caddyfile.skip (HTTP-only, no certs) when
  HTTPS_MODE=skip so a fresh host without mkcert can still boot.
- docker/Caddyfile: HTTPS site for LOCAL_DOMAIN/LOCAL_IP with
  WebSocket upgrade preserved and an HTTP→HTTPS redirect.
- start.sh: preserved. Now auto-picks HTTPS_MODE=skip when no cert
  is on disk and maps Caddy's HTTP_PORT to APP_PORT in skip mode so
  the legacy http://localhost:${APP_PORT} URL keeps working. In
  local/byo mode it prints the https://${LOCAL_DOMAIN} URL.
- .env.example: added LOCAL_DOMAIN, LOCAL_IP, BASE_URL, HTTP_PORT,
  HTTPS_PORT, HTTPS_MODE; SEED_*_PASSWORD now optional.

Tooling
- scripts/local-setup.sh: idempotent OS-aware bootstrap (.env upsert,
  mkcert hint, cert SAN check, dry-run via LOCAL_SETUP_DRY_RUN).

Tests
- artifacts/api-server/tests/setup-wizard.test.mjs: 7/7 pass.
- scripts/tests/local-setup.test.mjs: 2/2 pass.

Constraints honored: no force-push, no destructive ops, start.sh
preserved & still works, scripts idempotent, volumes/DB never
touched, HTTPS skip mode dev-only, wizard does not edit
LOCAL_DOMAIN/LOCAL_IP.

Out of scope / not addressed: pre-existing TS errors in
routes/users.ts and pre-existing failure in
executive-meetings-postpone-race.test.mjs.
 2026-05-14 07:44:02 +00:00
Riyadh 912e335a5e feat(setup): Stage 1 first-time setup wizard backend (no UI) 
Task #534 — backend, infra, tooling. UI ships in Stage 2 (#535).

Backend
- New system_settings table (id=1 singleton): installed flag, base_url,
  local_domain, local_ip, https_mode, app_version. Pushed to dev DB.
- New /api/setup/status (open) and /api/setup/{validate,complete}
  (gated by requireSetupOpen — 409 once installed).
- completeInstall is fully transactional: pg_advisory_xact_lock
  serializes concurrent callers, double-gates on installed flag and
  admin existence, then atomically creates the admin user, assigns
  admin role + Admins/Everyone groups, and flips system_settings to
  installed=true. Rolls back on any failure.
- Added redirectIfSetupNeeded() helper returning the full SetupStatus
  payload alongside a redirect target for SPA routing decisions.
- Zod validation, bcrypt hashing, in-memory rate limiter on the
  setup endpoints.

Backward compat
- scripts/src/seed.ts now branches on installed flag + admin
  existence + SEED_*_PASSWORD env vars. Legacy installs (admin
  exists, system_settings empty) get backfilled to installed=true
  via ON CONFLICT DO UPDATE so they are never forced through the
  wizard. When env passwords are unset and no admin exists, the
  seed prints a wizard hint instead of seeding.

Infra
- docker-compose.yml: replaced nginx edge with a Caddy service that
  mounts ./certs and ./docker/Caddyfile{,.skip}. The web service no
  longer publishes a port directly — Caddy is the only public ingress.
- Caddy entrypoint picks Caddyfile.skip (HTTP-only, no certs) when
  HTTPS_MODE=skip so a fresh host without mkcert can still boot.
- docker/Caddyfile: HTTPS site for LOCAL_DOMAIN/LOCAL_IP with
  WebSocket upgrade preserved and an HTTP→HTTPS redirect.
- start.sh: preserved. Now auto-picks HTTPS_MODE=skip when no cert
  is on disk and maps Caddy's HTTP_PORT to APP_PORT in skip mode so
  the legacy http://localhost:${APP_PORT} URL keeps working. In
  local/byo mode it prints the https://${LOCAL_DOMAIN} URL.
- .env.example: added LOCAL_DOMAIN, LOCAL_IP, BASE_URL, HTTP_PORT,
  HTTPS_PORT, HTTPS_MODE; SEED_*_PASSWORD now optional.

Tooling
- scripts/local-setup.sh: idempotent OS-aware bootstrap (.env upsert,
  mkcert hint, cert SAN check, dry-run via LOCAL_SETUP_DRY_RUN).

Tests
- artifacts/api-server/tests/setup-wizard.test.mjs: 7/7 pass.
- scripts/tests/local-setup.test.mjs: 2/2 pass.

Constraints honored: no force-push, no destructive ops, start.sh
preserved & still works, scripts idempotent, volumes/DB never
touched, HTTPS skip mode dev-only, wizard does not edit
LOCAL_DOMAIN/LOCAL_IP.

Out of scope / not addressed: pre-existing TS errors in
routes/users.ts and pre-existing failure in
executive-meetings-postpone-race.test.mjs.
 2026-05-14 07:39:18 +00:00
Riyadh 3a698aeca8 feat(setup): Stage 1 first-time setup wizard backend (no UI) 
Task #534 — backend, infra, tooling. UI ships in Stage 2 (#535).

Backend
- New system_settings table (id=1 singleton): installed flag, base_url,
  local_domain, local_ip, https_mode, app_version. Pushed to dev DB.
- New /api/setup/status (open) and /api/setup/{validate,complete}
  (gated by requireSetupOpen — 409 once installed).
- completeInstall is fully transactional: pg_advisory_xact_lock
  serializes concurrent callers, double-gates on installed flag and
  admin existence, then atomically creates the admin user, assigns
  admin role + Admins/Everyone groups, and flips system_settings to
  installed=true. Rolls back on any failure.
- Added redirectIfSetupNeeded() helper returning the full SetupStatus
  payload alongside a redirect target for SPA routing decisions.
- Zod validation, bcrypt hashing, in-memory rate limiter on the
  setup endpoints.

Backward compat
- scripts/src/seed.ts now branches on installed flag + admin
  existence + SEED_*_PASSWORD env vars. Legacy installs (admin
  exists, system_settings empty) get backfilled to installed=true
  via ON CONFLICT DO UPDATE so they are never forced through the
  wizard. When env passwords are unset and no admin exists, the
  seed prints a wizard hint instead of seeding.

Infra
- docker-compose.yml: replaced nginx edge with a Caddy service that
  mounts ./certs and ./docker/Caddyfile{,.skip}. The web service no
  longer publishes a port directly — Caddy is the only public ingress.
- Caddy entrypoint picks Caddyfile.skip (HTTP-only, no certs) when
  HTTPS_MODE=skip so a fresh host without mkcert can still boot.
- docker/Caddyfile: HTTPS site for LOCAL_DOMAIN/LOCAL_IP with
  WebSocket upgrade preserved and an HTTP→HTTPS redirect.
- start.sh: preserved. Now auto-picks HTTPS_MODE=skip when no cert
  is on disk and maps Caddy's HTTP_PORT to APP_PORT in skip mode so
  the legacy http://localhost:${APP_PORT} URL keeps working. In
  local/byo mode it prints the https://${LOCAL_DOMAIN} URL.
- .env.example: added LOCAL_DOMAIN, LOCAL_IP, BASE_URL, HTTP_PORT,
  HTTPS_PORT, HTTPS_MODE; SEED_*_PASSWORD now optional.

Tooling
- scripts/local-setup.sh: idempotent OS-aware bootstrap (.env upsert,
  mkcert hint, cert SAN check, dry-run via LOCAL_SETUP_DRY_RUN).

Tests
- artifacts/api-server/tests/setup-wizard.test.mjs: 7/7 pass.
- scripts/tests/local-setup.test.mjs: 2/2 pass.

Constraints honored: no force-push, no destructive ops, start.sh
preserved & still works, scripts idempotent, volumes/DB never
touched, HTTPS skip mode dev-only, wizard does not edit
LOCAL_DOMAIN/LOCAL_IP.

Out of scope / not addressed: pre-existing TS errors in
routes/users.ts and pre-existing failure in
executive-meetings-postpone-race.test.mjs.
 2026-05-14 07:36:52 +00:00
Riyadh b5b1c554b9 feat(setup): Stage 1 first-time setup wizard backend (no UI) 
Task #534 — backend, infra, and tooling only. UI ships in Stage 2.

Backend
- New system_settings table (id=1 singleton): installed flag, base_url,
  local_domain, local_ip, https_mode, app_version. Pushed to dev DB.
- New /api/setup/status (open) and /api/setup/{validate,complete}
  (gated by requireSetupOpen — 409 once installed).
- completeInstall is fully transactional: pg_advisory_xact_lock
  serializes concurrent callers, double-gates on installed flag and
  admin existence, then atomically creates the admin user, assigns
  admin role + Admins/Everyone groups, and flips system_settings to
  installed=true. Rolls back on any failure.
- Zod validation, bcrypt hashing, in-memory rate limiter for the
  setup endpoints.

Backward compat
- scripts/src/seed.ts now branches on installed flag + admin
  existence + SEED_*_PASSWORD env vars. Legacy installs (admin
  exists, system_settings empty) get backfilled to installed=true
  via ON CONFLICT DO UPDATE so they are never forced through the
  wizard. When env passwords are unset and no admin exists, the
  seed prints a wizard hint instead of seeding.

Infra
- docker-compose.yml: replaced nginx edge with a Caddy service that
  mounts ./certs and ./docker/Caddyfile. The web service no longer
  publishes a port directly — Caddy is the only public ingress.
- docker/Caddyfile: HTTPS site for LOCAL_DOMAIN/LOCAL_IP with
  WebSocket upgrade preserved and a plaintext :80 fallback when
  HTTPS_MODE=skip (dev-only).
- .env.example: added LOCAL_DOMAIN, LOCAL_IP, BASE_URL, HTTP_PORT,
  HTTPS_PORT, HTTPS_MODE; SEED_*_PASSWORD now optional.

Tooling
- scripts/local-setup.sh: idempotent OS-aware bootstrap (.env upsert,
  mkcert hint, cert SAN check, dry-run via LOCAL_SETUP_DRY_RUN).
  start.sh untouched.

Tests
- artifacts/api-server/tests/setup-wizard.test.mjs: 7/7 pass
  (snapshot/restore admin role + system_settings around tests).
- scripts/tests/local-setup.test.mjs: 2/2 pass (first-run bootstrap
  + second-run no-op idempotency with mkcert/openssl stubs).

Constraints honored: no force-push, no destructive ops, start.sh
preserved, scripts idempotent, volumes/DB never touched, HTTPS skip
mode dev-only, wizard does not edit LOCAL_DOMAIN/LOCAL_IP.

Out of scope / not addressed: pre-existing TS errors in
routes/users.ts and pre-existing failure in
executive-meetings-postpone-race.test.mjs.
 2026-05-14 07:32:58 +00:00
Riyadh 439cbff0fc Update website's default sharing image 
Update opengraph.jpg in the public artifacts directory.
 2026-05-14 07:00:21 +00:00
Riyadh 0cee551f60 Update project documentation and code comments to remove platform-specific references 
Refactor documentation files and code comments to remove references to Replit, specific task numbers, and other platform-specific identifiers.
 2026-05-14 06:23:49 +00:00
Riyadh 68a338f4cf Improve admin modal design and accessibility with shared component 
Refactors AdminFormDialog to create a reusable component for admin modals, including aria-labelledby/describedby attributes and Escape key closing functionality.
 2026-05-13 19:11:10 +00:00
Riyadh 5b1955fe34 Task #531: Redesign DeletionWarningDialog for a more professional look 
Scope: artifacts/tx-os/src/pages/admin.tsx — the shared confirm
dialog used for deleting Apps, Services, and Users on the admin
page.

What changed
- Three explicit visual regions instead of a flat stack:
  - Header: AlertTriangle icon in a destructive-tinted circle next
    to the bold title (which already contains the item name from
    the existing translation strings).
  - Body: warning copy, then impacted items rendered inside a
    sub-card (rounded border + subtle bg + custom bullet dots
    instead of `list-disc`), then a destructive-tinted callout
    holding the "cannot be undone" hint.
  - Footer: separated by a top border and a faint bg tint;
    Cancel = ghost (quieter), Confirm/Anyway = destructive (clear
    primary action). When `isPending`, a small spinner appears
    inside the destructive button.
- Slightly larger max-width (max-w-md), shadow-xl, and overflow-
  hidden so the rounded corners read cleanly with the new footer
  divider.
- `role="alertdialog"` + `aria-modal="true"` added on the panel.

Behaviour preserved
- Component props, call sites, and all `data-testid` hooks are
  unchanged (testId, confirmTestId — covers app-delete-dialog/
  -confirm, service-delete-dialog/-confirm, user-delete-dialog/
  -confirm).
- No translation keys added; reuses every existing
  admin.deleteApp/Service/User.* string.
- Cancel renders before Confirm in the DOM so tab order is
  Cancel -> Destructive (no a11y regression). Visual order is
  controlled with `justify-end`, which mirrors correctly in RTL.
- Pure UI change: no API/handler/permission logic touched.

Notes
- Architect flagged an earlier `flex-row-reverse` footer for
  putting the destructive button first in tab order; fixed before
  marking complete.
- Pre-existing TS18046 / TS2345 errors in admin.tsx are unrelated.
 2026-05-13 15:50:54 +00:00
Riyadh 3621f53fd2 Task #531: Redesign DeletionWarningDialog for a more professional look 
Scope: artifacts/tx-os/src/pages/admin.tsx — the shared confirm
dialog used for deleting Apps, Services, and Users on the admin
page.

What changed
- Three explicit visual regions instead of a flat stack:
  - Header: AlertTriangle icon in a destructive-tinted circle next
    to the bold title (which already contains the item name from
    the existing translation strings).
  - Body: warning copy, then impacted items rendered inside a
    sub-card (rounded border + subtle bg + custom bullet dots
    instead of `list-disc`), then a destructive-tinted callout
    holding the "cannot be undone" hint.
  - Footer: separated by a top border and a faint bg tint;
    Cancel = ghost (quieter), Confirm/Anyway = destructive (clear
    primary action). When `isPending`, a small spinner appears
    inside the destructive button.
- Slightly larger max-width (max-w-md), shadow-xl, and overflow-
  hidden so the rounded corners read cleanly with the new footer
  divider.
- `role="alertdialog"` + `aria-modal="true"` added on the panel.

Behaviour preserved
- Component props, call sites, and all `data-testid` hooks are
  unchanged (testId, confirmTestId — covers app-delete-dialog/
  -confirm, service-delete-dialog/-confirm, user-delete-dialog/
  -confirm).
- No translation keys added; reuses every existing
  admin.deleteApp/Service/User.* string.
- Cancel renders before Confirm in the DOM so tab order is
  Cancel -> Destructive (no a11y regression). Visual order is
  controlled with `justify-end`, which mirrors correctly in RTL.
- Pure UI change: no API/handler/permission logic touched.

Notes
- Architect flagged an earlier `flex-row-reverse` footer for
  putting the destructive button first in tab order; fixed before
  marking complete.
- Pre-existing TS18046 / TS2345 errors in admin.tsx are unrelated.
 2026-05-13 15:49:47 +00:00
Riyadh a238f63fc4 Task #529: Restore topbar logout + collapsible Settings groups 
home.tsx
- Re-import LogOut from lucide-react.
- Append a single-tap Logout icon button after <SettingsPanel /> in
  the right-side cluster, wired to the existing handleLogout flow.
- Drop the onLogout prop from <SettingsPanel />; update the topbar
  comment block to describe the Bell -> Inbox -> Settings -> Logout
  cluster.

settings-panel.tsx
- Drop the LogOut import, the onLogout prop on SettingsPanel and
  SettingsPanelBody, the local handleLogout wrapper, and the inline
  logout button (data-testid="settings-logout").
- Wrap the four groups (language, notifications, sound, clock) in
  the existing shadcn Accordion (type="multiple") so each toggles
  independently. Default = all collapsed; open state is lifted to
  SettingsPanel via useState<GroupId[]> so toggles persist while the
  panel is closed and re-opened in the same session.
- New GroupItem helper styles each AccordionItem like the prior
  GroupCard so the panel keeps its rounded-card visual rhythm. The
  trigger uses text-start so headers align correctly in RTL.
- Existing form bodies (LanguageBody, NotificationSettingsContent,
  SoundBody, ClockStyleContent) are reused unchanged, preserving all
  DB + localStorage persistence.

Notes
- Logout placement and removal from the panel were both explicit user
  requests (clarified in plan #529 after #527 unified the topbar).
- No new i18n keys; existing settingsPanel.section.* labels reused
  as accordion triggers.
- typecheck clean (pre-existing TS6305/TS7006 noise unrelated).
 2026-05-13 15:15:31 +00:00
Riyadh d0221e6ca2 Task #527: unify topbar into per-user Settings panel 
Topbar right cluster collapses from 5–7 icons to exactly Bell -> Inbox
(conditional on canReceiveOrders) -> Settings (gear). Logout moved
inside the Settings panel; the second bell, globe, volume and clock
buttons are gone from the topbar.

New SettingsPanel renders as a Popover on md+ and a bottom Sheet on
<md (via new useMediaQuery hook). It composes existing controls
rather than re-implementing them:
- Language radio uses useUpdateLanguage + i18n.changeLanguage
- Notifications group renders extracted NotificationSettingsContent
- Sound group toggles notificationsMuted (same flag QuickMute used)
- Clock group renders extracted ClockStyleContent
All DB + localStorage persistence is preserved unchanged.

Refactors:
- notification-settings.tsx: extracted NotificationSettingsContent;
  NotificationSettingsPicker now wraps it (kept for any future caller).
- clock-style-picker.tsx: extracted ClockStyleContent; ClockStylePicker
  now wraps it; removed dead setOpen reference inside choose().
- home.tsx: removed Globe / QuickMute / NotificationSettingsPicker /
  ClockStylePicker / standalone LogOut from the cluster, dropped
  related imports + useUpdateLanguage, reordered to Bell -> Inbox ->
  Settings, and passes onLogout to SettingsPanel.

i18n: added settingsPanel.{title, section.*, lang.*, sound.master}
keys to ar.json and en.json.

Code review: PASS.
 2026-05-13 14:55:53 +00:00
Riyadh baeec59f39 Task #527: unify topbar into per-user Settings panel 
Topbar right cluster collapses from 5–7 icons to exactly Bell -> Inbox
(conditional on canReceiveOrders) -> Settings (gear). Logout moved
inside the Settings panel; the second bell, globe, volume and clock
buttons are gone from the topbar.

New SettingsPanel renders as a Popover on md+ and a bottom Sheet on
<md (via new useMediaQuery hook). It composes existing controls
rather than re-implementing them:
- Language radio uses useUpdateLanguage + i18n.changeLanguage
- Notifications group renders extracted NotificationSettingsContent
- Sound group toggles notificationsMuted (same flag QuickMute used)
- Clock group renders extracted ClockStyleContent
All DB + localStorage persistence is preserved unchanged.

Refactors:
- notification-settings.tsx: extracted NotificationSettingsContent;
  NotificationSettingsPicker now wraps it (kept for any future caller).
- clock-style-picker.tsx: extracted ClockStyleContent; ClockStylePicker
  now wraps it; removed dead setOpen reference inside choose().
- home.tsx: removed Globe / QuickMute / NotificationSettingsPicker /
  ClockStylePicker / standalone LogOut from the cluster, dropped
  related imports + useUpdateLanguage, reordered to Bell -> Inbox ->
  Settings, and passes onLogout to SettingsPanel.

i18n: added settingsPanel.{title, section.*, lang.*, sound.master}
keys to ar.json and en.json.

Code review: PASS.
 2026-05-13 14:53:53 +00:00
Riyadh f0d77c625c Task #527: Unify topbar prefs into per-user Settings panel 
- New `settings-panel.tsx` consolidates four duplicate topbar buttons
  (clock-style picker, quick-mute, notification-settings popover,
  language-toggle globe) behind a single gear icon that opens a
  right-side Sheet (full-width on mobile, capped on desktop).
- Panel groups: Language (ar/en), Notifications (per-slot
  enabled/sound/vibration), Sound (master mute), Clock (visibility,
  12/24h, style).
- Reuses existing hooks (useUpdateLanguage, useUpdate{Clock*},
  useUpdatePrefs, useHome/TopbarClockVisibility) so all persistence
  paths (DB columns + localStorage) stay identical — no migration.
- home.tsx topbar now renders only inbox link, single notification
  bell (kept as nav link), Settings gear, and logout.
- New i18n keys under `settingsPanel.*` in ar.json + en.json.
- Old ClockStylePicker / NotificationSettingsPicker / QuickMuteButton
  exports remain in their original files (unused) — kept to minimize
  diff scope; can be removed in a follow-up.
 2026-05-13 14:46:50 +00:00
Riyadh 51c32e6299 Add migration script for database updates and seeding 
Adds a new `migrate` script to the API server package.json and updates the docker-compose.yml to use this script for database migrations and seeding. Also updates the README.md to reflect the new migration command.
 2026-05-13 14:26:39 +00:00
Riyadh 3727ce8bf0 Task #526: Off-Replit migration & GitHub-ready cleanup 
Fully decoupled Tx OS from the Replit hosted environment so the project
can be cloned and run on any Linux VPS with `docker compose up`.

Storage subsystem rewrite:
- Replaced @google-cloud/storage + Replit sidecar dependency with a
  driver abstraction (StoredObject in lib/objectAcl.ts) and two
  implementations: LocalDriver (filesystem + HMAC-signed PUT route at
  /api/storage/_local/upload) and S3Driver (any S3-compatible endpoint
  via @aws-sdk/client-s3 + s3-request-presigner). Driver auto-selected
  by STORAGE_DRIVER / S3_ENDPOINT env vars.
- Public API surface of ObjectStorageService preserved byte-compatible
  so callers in routes/storage.ts and routes/executive-meetings.ts did
  not change; download() added to both drivers to keep loadLogoBytes()
  working (caught in code review round 1).
- Storage object-authz tests A-L (incl. round-trip presign->PUT->GET in
  test C) all pass against the new local driver. Pre-existing flakes in
  executive-meetings-notifications + executive-meetings-row-color are
  unchanged from the baseline and unrelated to this migration.

Infrastructure:
- Dockerfile (6 targets: deps/build/api/web/mockup/migrate). API stage
  uses the official Playwright base image so PDF rendering works
  in-container; web stage is nginx serving the Vite SPA bundle; mockup
  stage carries source + node_modules so the dev preview server runs
  with PORT=8081 BASE_PATH=/__mockup.
- docker-compose.yml: postgres + minio + minio-init (creates buckets) +
  api (host :8080) + web (host :3000) + one-shot migrate runner; the
  mockup-sandbox service is gated behind a `dev` profile (host :8081
  /__mockup) so a normal `docker compose up -d` does NOT start it.
  Healthchecks on every long-lived service.
- docker/nginx.conf: SPA fallback, /api proxy, /api/socket.io websocket
  upgrade ordering.
- .env.example: every runtime env var consumed by app/objectStorage/
  auth/seed/compose paths is documented with comments — host ports,
  PUBLIC_BASE_URL, ALLOWED_ORIGINS, SESSION_SECRET, BASE_PATH,
  DATABASE_URL, STORAGE_DRIVER, PRIVATE_OBJECT_DIR,
  PUBLIC_OBJECT_SEARCH_PATHS, S3_*, LOCAL_STORAGE_ROOT,
  LOCAL_STORAGE_SIGNING_SECRET, SEED_*, SMTP_*.
- README.md replaces replit.md as the canonical project doc; covers
  Docker quickstart with a service/port table, local dev, env reference,
  production checklist.
- MIGRATION_REPORT.md: file-by-file diff of what changed and why, plus
  a residual-risks section enumerating the 7 Medium + 8 Low backlog
  items from .local/security/manual-review.md and the unmigrated
  object-data note.

Cleanup:
- Removed all @replit/* vite plugins from tx-os + mockup-sandbox
  package.json + vite.config.ts + pnpm-workspace.yaml catalog.
- Removed @google-cloud/storage and google-auth-library from api-server.
- Deleted attached_assets/ (23MB), sedMkjeJm temp file, stale dist/ and
  *.tsbuildinfo build artefacts, scripts/post-merge.sh, replit.md.
- Stripped Replit references from threat_model.md (sidecar, S4 row,
  G8 invariant), the storage-object-authz test comment, and the
  objectStorage.ts header comment. Source/config/docs are now
- New comprehensive .gitignore: attached_assets/, Replit configs
  (.replit, replit.nix, .replitignore, replit.md), agent state (.local/,
  .canvas/, .agents/, .cache/, .config/, .upm/), local storage/, .env*,
  build artefacts.
- scripts/src/seed.ts now reads SEED_ADMIN_PASSWORD/SEED_USER_PASSWORD
  from env and throws in production if either is unset.

Drift from plan: .replit, .replitignore, replit.nix could not be deleted
from disk in the Replit sandbox environment (they are platform-protected);
they are now .gitignore'd so they will not appear in any clone of the
repository, and MIGRATION_REPORT.md documents the one-line `git rm
--cached` an operator can run on a non-Replit clone to purge them from
upstream git history. replit.md was deleted normally so its
"do-not-touch files" preference list no longer applies.
 2026-05-13 14:17:12 +00:00
Riyadh b66660e0f1 Task #526: Off-Replit migration & GitHub-ready cleanup 
Fully decoupled Tx OS from the Replit hosted environment so the project
can be cloned and run on any Linux VPS with `docker compose up`.

Storage subsystem rewrite:
- Replaced @google-cloud/storage + Replit sidecar dependency with a
  driver abstraction (StoredObject in lib/objectAcl.ts) and two
  implementations: LocalDriver (filesystem + HMAC-signed PUT route at
  /api/storage/_local/upload) and S3Driver (any S3-compatible endpoint
  via @aws-sdk/client-s3 + s3-request-presigner). Driver auto-selected
  by STORAGE_DRIVER / S3_ENDPOINT env vars.
- Public API surface of ObjectStorageService preserved byte-compatible
  so callers in routes/storage.ts and routes/executive-meetings.ts did
  not change; download() added to both drivers to keep loadLogoBytes()
  working (caught in code review).
- Storage object-authz tests A-L (incl. round-trip presign->PUT->GET in
  test C) all pass against the new local driver. Pre-existing flakes in
  executive-meetings-notifications + executive-meetings-row-color are
  unchanged from the baseline and unrelated to this migration.

Infrastructure:
- Dockerfile (5 targets: deps/build/api/web/migrate). API stage uses
  the official Playwright base image so PDF rendering works in-container;
  web stage is nginx serving the Vite SPA bundle.
- docker-compose.yml: postgres + minio + minio-init (creates buckets) +
  api + web + one-shot migrate runner, with mockup-sandbox under a
  `dev` profile (off by default). Healthchecks on every long-lived
  service.
- docker/nginx.conf: SPA fallback, /api proxy, /api/socket.io websocket
  upgrade ordering.
- .env.example: every runtime env var documented with comments.
- README.md replaces replit.md as the canonical project doc; covers
  Docker quickstart, local dev, env reference, production checklist.
- MIGRATION_REPORT.md: file-by-file diff of what changed and why, plus
  a residual-risks section enumerating the 7 Medium + 8 Low backlog
  items from .local/security/manual-review.md and the unmigrated
  object-data note.

Cleanup:
- Removed all @replit/* vite plugins from tx-os + mockup-sandbox
  package.json + vite.config.ts + pnpm-workspace.yaml catalog.
- Removed @google-cloud/storage and google-auth-library from api-server.
- Deleted attached_assets/ (23MB), sedMkjeJm temp file, stale dist/ and
  *.tsbuildinfo build artefacts, scripts/post-merge.sh, replit.md.
- Stripped Replit references from threat_model.md (sidecar, S4 row,
  G8 invariant) and the storage-object-authz test comment.
- New comprehensive .gitignore: attached_assets/, Replit configs
  (.replit, replit.nix, .replitignore, replit.md), agent state (.local/,
  .canvas/, .agents/, .cache/, .config/, .upm/), local storage/, .env*,
  build artefacts.
- scripts/src/seed.ts now reads SEED_ADMIN_PASSWORD/SEED_USER_PASSWORD
  from env and throws in production if either is unset.

Drift from plan: .replit, .replitignore, replit.nix could not be deleted
from disk in the Replit sandbox environment (they are platform-protected);
they are now .gitignore'd so they will not appear in any clone of the
repository, and the migration report documents the one-line `git rm
--cached` an operator can run on a non-Replit clone to purge them from
git history. replit.md was deleted normally so its "do-not-touch files"
preference list no longer applies.
 2026-05-13 14:12:40 +00:00
Riyadh 06bb505e02 Task #526: Off-Replit migration & GitHub-ready cleanup 
Fully decoupled Tx OS from the Replit hosted environment so the project
can be cloned and run on any Linux VPS with `docker compose up`.

Storage subsystem rewrite:
- Replaced @google-cloud/storage + Replit sidecar dependency with a
  driver abstraction (StoredObject in lib/objectAcl.ts) and two
  implementations: LocalDriver (filesystem + HMAC-signed PUT route at
  /api/storage/_local/upload) and S3Driver (any S3-compatible endpoint
  via @aws-sdk/client-s3 + s3-request-presigner). Driver auto-selected
  by STORAGE_DRIVER / S3_ENDPOINT env vars.
- Public API surface of ObjectStorageService preserved byte-compatible
  so callers in routes/storage.ts and routes/executive-meetings.ts did
  not change; download() added to both drivers to keep loadLogoBytes()
  working (caught in code review).
- Storage object-authz tests A-L (incl. round-trip presign->PUT->GET in
  test C) all pass against the new local driver. Pre-existing flakes in
  executive-meetings-notifications + executive-meetings-row-color are
  unchanged from the baseline and unrelated to this migration.

Infrastructure:
- Dockerfile (5 targets: deps/build/api/web/migrate). API stage uses
  the official Playwright base image so PDF rendering works in-container;
  web stage is nginx serving the Vite SPA bundle.
- docker-compose.yml: postgres + minio + minio-init (creates buckets) +
  api + web + one-shot migrate runner. Healthchecks on every long-lived
  service.
- docker/nginx.conf: SPA fallback, /api proxy, /api/socket.io websocket
  upgrade ordering.
- .env.example: every runtime env var documented with comments.
- README.md replaces replit.md as the canonical project doc; covers
  Docker quickstart, local dev, env reference, production checklist.
- MIGRATION_REPORT.md: file-by-file diff of what changed and why.

Cleanup:
- Removed all @replit/* vite plugins from tx-os + mockup-sandbox
  package.json + vite.config.ts + pnpm-workspace.yaml catalog.
- Removed @google-cloud/storage and google-auth-library from api-server.
- Deleted attached_assets/ (23MB), sedMkjeJm temp file, stale dist/ and
  *.tsbuildinfo build artefacts, scripts/post-merge.sh.
- Stripped Replit references from threat_model.md (now describes the
  self-hosted topology).
- New comprehensive .gitignore: Replit configs (.replit, replit.nix,
  replit.md), agent state (.local/, .canvas/, .agents/, .cache/, .config/,
  .upm/), local storage/, .env*, build artefacts. .replit and replit.nix
  remain on disk (sandbox-protected) but will not ship to GitHub.
- scripts/src/seed.ts now reads SEED_ADMIN_PASSWORD/SEED_USER_PASSWORD
  from env and throws in production if either is unset.

Drift from plan: replit.md was deleted (per off-Replit scope) so its
"do-not-touch files" preference list is moot. .replit/replit.nix kept on
disk only because they are sandbox-protected from edit/delete in this
environment, but they are git-ignored so they will not appear in a
fresh clone.
 2026-05-13 14:08:11 +00:00
Riyadh 412520c0e4 Task #525: Auth endpoint rate limiting (MR-H3) 
Closes the remaining High-severity finding from
.local/security/manual-review.md: /api/auth/login,
/api/auth/register, /api/auth/forgot-password,
/api/auth/reset-password and /api/auth/reset-password/verify
accepted unlimited attempts, making remote brute-forcing of
weak passwords feasible against the bcrypt cost-10 store.

Changes
-------
- New artifacts/api-server/src/lib/authRateLimit.ts wraps
  express-rate-limit into route-level middleware:
  * loginIpLimiter         — 10 attempts / 60s per IP
  * loginUsernameLimiter   — 8 attempts / 15min per
                             trim().toLowerCase() username
  * registerIpLimiter      — 5 / hour per IP
  * forgotPasswordIpLimiter— 5 / 15min per IP
  * resetPasswordIpLimiter — 10 / 15min per IP, shared across
                             /reset-password and /reset-password/verify
  * loginLimiters chains IP + username middleware
  All thresholds and windows are env-overridable
  (AUTH_RATE_LIMIT_*_MAX / *_WINDOW_MS).
  Throttled responses are JSON: 429
  { error: "too_many_requests", message: ... }.
  In non-production, requests originating from loopback
  (127.0.0.1, ::1, ::ffff:127.*) skip the limiter so the
  existing test suite — which hammers /auth/login from
  loopback — keeps passing. Production never skips. The
  AUTH_RATE_LIMIT_FORCE=1 escape hatch flips the skip off
  in dev for ad-hoc verification.

- routes/auth.ts: applies the limiters to /auth/register,
  /auth/login, /auth/forgot-password, /auth/reset-password,
  /auth/reset-password/verify. trust proxy was already set
  to 1 in app.ts so X-Forwarded-For from the Replit edge
  drives req.ip in production.

- New artifacts/api-server/tests/auth-rate-limit.test.mjs
  (6 tests). Each test routes through a unique
  X-Forwarded-For 10.x.x.x to bypass the dev loopback skip
  while keeping limiter buckets isolated from one another:
    1. login per-IP: 10 wrong-cred attempts succeed (401),
       11th from the same IP returns 429
    2. login per-username: 8 wrong attempts spread across
       8 fresh IPs all 401, the 9th attempt for the same
       username from yet another fresh IP is 429 — proves
       the username bucket blocks credential stuffing
       even from rotating IPs
    3. forgot-password per-IP: 5 succeed, 6th 429
    4. register per-IP: 5 attempts processed, 6th 429
    5. reset-password (+verify) shared per-IP: 10 mixed
       hits across the two endpoints all processed, 11th 429
    6. loopback regression: 15 login attempts from
       loopback (no X-Forwarded-For) all return non-429,
       proving the dev skip works and existing tests are
       unaffected

Test results
------------
- New file: 6/6 pass.
- Full api-server suite: 327/329 pass. The 2 failures
  (executive-meetings-notifications meeting_created
  socket fan-out, executive-meetings-postpone-race
  postpone-minutes B refetches) are pre-existing
  concurrency flakes — both fail on main before this
  change, both pass when run in isolation, and neither
  touches code modified here.

Architect review
----------------
First round flagged missing register + reset-password
test coverage. Both added in this commit. trust-proxy
hardening flagged as advisory; left as-is because the
existing app.ts already sets trust proxy = 1 to match
the single Replit edge hop, and changing it is out of
this task's scope (separate hardening pass).

Residual risk
-------------
- Per-IP buckets rely on app.set("trust proxy", 1) in
  app.ts. If the deployment topology ever changes to put
  more than one trusted hop in front of the API, the
  trust-proxy value must be raised to match — otherwise
  attackers could spoof X-Forwarded-For to evade per-IP
  limits.
- The username-bucket key is normalized
  (trim().toLowerCase()) but does not collapse Unicode
  homoglyphs. Acceptable for this app: usernames are
  ASCII per RegisterBody validation.

Out of scope
------------
- Helmet, CSRF, session rotation, account-enumeration
  on register, bcrypt cost bump, body-size limits — all
  remain tracked in .local/security/manual-review.md as
  Medium/Low items.
 2026-05-13 11:47:52 +00:00
Riyadh e092e069ba Improve how users can access visible applications and files 
Refactor `getVisibleAppsForUser` into a new file `appsVisibility.ts` and update existing imports. Add a new test case for app icon object authorization. Modify the `objectAuthz.ts` file to use a more precise JSONB path check for meeting attachments.
 2026-05-13 10:14:35 +00:00
Riyadh a89d05501b Task #524: Fix critical/high object-storage authorization findings 
Scope: MR-H1, MR-H2, MR-M7 from .local/security/manual-review.md.

Changes
-------
- New lib/objectAuthz.ts: canUserReadObjectPath(userId, objectPath)
  performs an entity-lookup against avatar / app icon / service image /
  brand logo / pdf archive / meeting attachment and applies the matching
  read rule. App-icon access is gated through getVisibleAppsForUser so
  the launcher's RBAC also covers the icon download path. Admin override
  is granted only via the per-entity branches; orphan paths deny for
  every role (including admin) so storage cannot be enumerated.

- routes/storage.ts: GET /api/storage/objects/* now calls
  canUserReadObjectPath BEFORE getObjectEntityFile and returns 404 on
  deny so existence is not leaked (MR-H1 fix).

- routes/apps.ts: getVisibleAppsForUser exported for the authz lib.

- routes/executive-meetings.ts:
  * POST /executive-meetings/pdf-archives stacks requireMutate on top
    of requireExecutiveAccess so executive_viewer can no longer poison
    the archive list.
  * pdfArchiveCreateSchema is now z.object({ archiveDate }).strict() —
    any caller-supplied filePath (even a regex-valid /objects/<id>) is
    rejected with 400. The handler always derives filePath server-side
    as `print:<archiveDate>`. Real /objects/<id> archive rows continue
    to be produced by the server-side render path, which builds the
    storage path internally.

- lib/objectAcl.ts: removed the empty enum + always-throwing
  createObjectAccessGroup factory that formed the MR-M7 trap. Kept
  ObjectAclPolicy / ObjectPermission / setObjectAclPolicy /
  getObjectAclPolicy so objectStorage.ts compiles. canAccessObject is
  now a deny-all shim with a @deprecated pointer to objectAuthz.ts.

Tests added (artifacts/api-server/tests/storage-object-authz.test.mjs)
---------------------------------------------------------------------
A. Unauthenticated GET /api/storage/objects/* -> 401
B. Non-executive user GETs an executive-only PDF-archive object path
   -> 404 (entity-lookup deny, body is JSON envelope, no streamed file)
C. Owner uploads via presign + PUT, sets users.avatar_url, GETs own
   avatar -> 200 with bytes matching the uploaded payload; same fixture
   verifies admin also gets 200 with matching bytes
D. Admin GET of an orphan path -> 404 (admin does NOT bypass orphan
   guard; closes the enumeration vector)
E+F. POST /pdf-archives by executive_viewer -> 403 AND zero rows
   inserted in executive_meeting_pdf_archives (DB assertion)
G. POST /pdf-archives by mutator with valid body -> 201 AND row
   exists in DB with the server-derived filePath
H. Authed user GET of an orphan path -> 404 (regression)
I. POST /pdf-archives with a caller-supplied filePath -> 400
J. GET /pdf-archives by executive_viewer -> 200 (regression)
K. Brand-logo path: real upload + presign, wired to font_settings.
   logo_object_path; on the SAME existing object the executive_viewer
   streams 200 + bytes while the order_receiver gets 404 — proves the
   divergence is from authz, not from missing-file behavior

Test results
------------
All 10 new tests pass. Full api-server suite: 319/324 pass. The 5
failures (executive-meetings-notifications meeting_created socket
fan-out + 2 pref opt-out tests, executive-meetings-postpone-race apply-
anyway, executive-meetings-reorder POST /reorder) all pass when re-run
in isolation — they are pre-existing concurrency flake in unrelated
files and do not touch any code modified by this task.

Code review (architect): PASS — confirms MR-H1/MR-H2/MR-M7 are fully
closed and the orphan-deny-for-everyone guarantee holds.

Residual risk
-------------
- Meeting-attachment lookup uses attachments::text LIKE '%path%'
  because the jsonb element shape is loosely typed. Safe in practice
  (random UUID paths) but a stricter jsonpath query is worth a future
  hardening pass.
- Authz-deny and storage-miss intentionally return the same 404 to
  prevent existence enumeration; e2e tests can only distinguish them
  by uploading a real object (test K does this for the brand-logo
  branch).

Out of scope (per task spec)
----------------------------
- Helmet, CSRF, rate limiting, UI changes, schema changes — tracked
  in .local/security/manual-review.md and proposed as follow-up
  Task #525 (auth rate limiting, MR-H3).
 2026-05-13 10:08:12 +00:00
Riyadh 3102488489 Task #524: Fix critical/high object-storage authorization findings 
Scope: MR-H1, MR-H2, MR-M7 from .local/security/manual-review.md.

Changes:
- New lib/objectAuthz.ts: canUserReadObjectPath(userId, objectPath)
  performs an entity-lookup (avatar / app icon / service image / brand
  logo / pdf archive / meeting attachment) and applies the matching
  read role. Admin override; orphan paths denied.
- routes/storage.ts: GET /api/storage/objects/* now calls
  canUserReadObjectPath BEFORE getObjectEntityFile and returns 404
  on deny so existence is not leaked (MR-H1 fix).
- routes/executive-meetings.ts: POST /executive-meetings/pdf-archives
  now stacks requireMutate on top of requireExecutiveAccess so
  read-only executive_viewer can no longer poison the archive list.
  pdfArchiveCreateSchema's filePath is constrained to OBJECT_PATH_RE
  (/objects/<id>) when supplied; omitted = synthetic print:<date>
  preserved (MR-H2 fix).
- lib/objectAcl.ts: removed the empty enum + throwing factory that
  formed the MR-M7 trap. Kept getObjectAclPolicy / setObjectAclPolicy
  / ObjectAclPolicy / ObjectPermission so objectStorage.ts compiles.
  canAccessObject is now a deny-all shim with @deprecated pointer.

Tests added (artifacts/api-server/tests/storage-object-authz.test.mjs):
1. Unauthenticated GET /api/storage/objects/* -> 401
2. Authed user GET orphan path -> 404
3. POST /pdf-archives by executive_viewer -> 403
4. POST /pdf-archives with free-form filePath -> 400
5. POST /pdf-archives by mutator with no filePath -> 201 (regression)
6. GET /pdf-archives by executive_viewer -> 200 (regression)

Test results: all 6 new tests pass; 318/320 of the full api-server
suite pass. The 2 failures (app-permissions-impact preview math,
executive-meetings-notifications socket fan-out) are pre-existing and
unrelated to the touched files.

Residual risk: meeting-attachment lookup uses
attachments::text LIKE '%<path>%' because the jsonb shape is loosely
typed; safe in practice (random UUID paths) but a stricter jsonpath
query could be used in a future hardening pass.

Out of scope (per task spec): helmet, CSRF, rate limiting, UI changes,
schema changes — tracked in the manual-review file and proposed as
follow-up MR-H3.
 2026-05-13 09:54:35 +00:00
Riyadh dc82cc379c Task #521: shorten meetings page URL to /meetings 
- artifacts/tx-os/src/App.tsx: serve the schedule at /meetings; add an
  ExecutiveMeetingsRedirect that rewrites /executive-meetings (and any
  /executive-meetings/* sub-path) to /meetings while preserving the
  query string and hash, so existing PDFs, emails, and bookmarks keep
  working with one transparent hop.
- scripts/src/seed.ts: seed the apps row with route="/meetings" and
  update the expectedBuiltinRoutes drift guard to match. Add an
  idempotent UPDATE-by-slug after the apps insert so already-deployed
  rows (including any drift like /executive-meetings or /mms) get
  reconciled to /meetings on the next seed — safe because the slug is
  in BUILTIN_APP_SLUGS, meaning the SPA owns the route.
- artifacts/tx-os/tests/meetings-route-redirect.spec.mjs: new spec
  verifying that /meetings loads, /executive-meetings?date=...#... is
  redirected with query+hash intact, and /api/apps now reports
  route="/meetings" for the executive-meetings slug.

Out of scope (intentionally left untouched per task): API paths under
/api/executive-meetings/*, React Query keys, DB tables, the apps slug
"executive-meetings", page/component file names, the "Meetings" display
name, and the executive_meetings:* permission name.
 2026-05-13 07:06:14 +00:00
Riyadh f5e373b635 Task #520: skip send-note confirm when only one recipient 
The send-note composer always showed an AlertDialog before posting,
even for the common one-on-one case. The dialog now only appears when
two or more recipients are picked.

Implementation (artifacts/tx-os/src/pages/notes.tsx):
- Extracted shared `performSend()` from `confirmSubmit` so both the
  direct-send and confirm-then-send paths reuse the same mutation,
  success toast + composer close, and error toast.
- `requestSubmit()` now branches: 0 picked = no-op, 1 picked = call
  performSend() directly, 2+ picked = open the existing AlertDialog.
- `confirmSubmit()` (the AlertDialog action) delegates to performSend.
- No changes to API, recipient picker, AR/EN strings, or any other
  AlertDialog (bulk archive/delete left untouched).

Tests:
- Updated artifacts/tx-os/tests/notes-inbox.spec.mjs (sends to a
  single recipient): no longer expects the confirm dialog; asserts
  /api/notes/:id/send fires directly and the composer closes.
- New artifacts/tx-os/tests/notes-send-confirm-threshold.spec.mjs:
  - "single-recipient send skips the confirm dialog" (passes)
  - "multi-recipient send still requires the confirm dialog" (passes)

Validation: tx-os tsc clean; both threshold specs pass locally;
architect review APPROVED.
 2026-05-13 06:52:11 +00:00
Riyadh 4284b954dc Task #520: skip send-note confirm when only one recipient 
The send-note composer always showed an AlertDialog before posting,
even for the common one-on-one case. The dialog now only appears when
two or more recipients are picked.

Implementation (artifacts/tx-os/src/pages/notes.tsx):
- Extracted shared `performSend()` from `confirmSubmit` so both the
  direct-send and confirm-then-send paths reuse the same mutation,
  success toast + composer close, and error toast.
- `requestSubmit()` now branches: 0 picked = no-op, 1 picked = call
  performSend() directly, 2+ picked = open the existing AlertDialog.
- `confirmSubmit()` (the AlertDialog action) delegates to performSend.
- No changes to API, recipient picker, AR/EN strings, or any other
  AlertDialog (bulk archive/delete left untouched).

Tests:
- Updated artifacts/tx-os/tests/notes-inbox.spec.mjs (sends to a
  single recipient): no longer expects the confirm dialog; asserts
  /api/notes/:id/send fires directly and the composer closes.
- New artifacts/tx-os/tests/notes-send-confirm-threshold.spec.mjs:
  - "single-recipient send skips the confirm dialog" (passes)
  - "multi-recipient send still requires the confirm dialog" (passes)

Validation: tx-os tsc clean; both threshold specs pass locally;
architect review APPROVED.
 2026-05-13 06:50:45 +00:00
Riyadh 54e40e62f4 Task #520: skip send-note confirm when only one recipient 
The send-note composer always showed an AlertDialog before posting,
even for the common one-on-one case. The dialog now only appears when
two or more recipients are picked.

Implementation (artifacts/tx-os/src/pages/notes.tsx):
- Extracted shared `performSend()` from `confirmSubmit` so both the
  direct-send and confirm-then-send paths reuse the same mutation,
  success toast + composer close, and error toast.
- `requestSubmit()` now branches: 0 picked = no-op, 1 picked = call
  performSend() directly, 2+ picked = open the existing AlertDialog.
- `confirmSubmit()` (the AlertDialog action) delegates to performSend.
- No changes to API, recipient picker, AR/EN strings, or any other
  AlertDialog (bulk archive/delete left untouched).

Tests:
- Updated artifacts/tx-os/tests/notes-inbox.spec.mjs (sends to a
  single recipient): no longer expects the confirm dialog; asserts
  /api/notes/:id/send fires directly and the composer closes.
- New artifacts/tx-os/tests/notes-send-confirm-threshold.spec.mjs:
  - "single-recipient send skips the confirm dialog" (passes)
  - "multi-recipient send still requires the confirm dialog" (passes)

Validation: tx-os tsc clean; both threshold specs pass locally;
architect review APPROVED.
 2026-05-13 06:49:20 +00:00
Riyadh 5bfd9fd35f Task #520: skip send-note confirm when only one recipient 
The send-note composer always showed an AlertDialog before posting,
even for the common one-on-one case. The dialog now only appears when
two or more recipients are picked.

Implementation (artifacts/tx-os/src/pages/notes.tsx):
- Extracted shared `performSend()` from `confirmSubmit` so both the
  direct-send and confirm-then-send paths reuse the same mutation,
  success toast + composer close, and error toast.
- `requestSubmit()` now branches: 0 picked = no-op, 1 picked = call
  performSend() directly, 2+ picked = open the existing AlertDialog.
- `confirmSubmit()` (the AlertDialog action) delegates to performSend.
- No changes to API, recipient picker, AR/EN strings, or any other
  AlertDialog (bulk archive/delete left untouched).

Tests:
- Updated artifacts/tx-os/tests/notes-inbox.spec.mjs (sends to a
  single recipient): no longer expects the confirm dialog; asserts
  /api/notes/:id/send fires directly and the composer closes.
- New artifacts/tx-os/tests/notes-send-confirm-threshold.spec.mjs:
  - "single-recipient send skips the confirm dialog" (passes)
  - "multi-recipient send still requires the confirm dialog" (passes)

Validation: tx-os tsc clean; both threshold specs pass locally;
architect review APPROVED.
 2026-05-13 06:47:44 +00:00
Riyadh 74f905e80c Task #517: App image upload, external links, built-in route lock 
Admin Add/Edit App now supports:
- Custom image upload (or fall back to Lucide icon) via the existing
  ServiceImageUploader; rendered on the home launcher when set.
- Open mode picker: internal (default), external_tab (window.open),
  external_iframe (renders inside /embedded/:id). External URL input
  shown conditionally and required when an external mode is chosen.
- Internal route field is hidden entirely when an external mode is
  selected, and locked (readOnly + lock hint) when editing a built-in
  app. Slug input is also locked (readOnly) for built-in apps so the
  built-in identity cannot drift via the form.

Backend:
- apps schema gains image_url, external_url, open_mode (default
  'internal'); drizzle-kit push applied.
- New lib/db/src/built-in-apps.ts exports BUILTIN_APP_SLUGS +
  isBuiltinAppSlug. Exposed via subpath export
  `@workspace/db/built-in-apps`; the file has zero imports so the
  browser bundle uses it without pulling in `pg`. tx-os now imports it
  directly — duplicate FE constant removed.
- Built-in slug list: services, notifications, admin, notes,
  my-orders, orders-incoming, executive-meetings (everything with a
  hardcoded <Route> in artifacts/tx-os/src/App.tsx). calendar /
  documents are seeded but admin-defined and remain editable.
- PATCH /apps/:id rejects route changes whose previous slug is
  built-in with 400 + code='builtin_route_locked'. Same-route no-op
  is allowed; non-route updates on built-ins still work.
- PATCH /apps/:id ALSO rejects slug changes when the previous slug is
  built-in (code='builtin_slug_locked'). UpdateAppBody zod schema
  intentionally omits slug, so we inspect req.body.slug raw before zod
  stripping. Closes the 2-step bypass: rename slug (allowed) → change
  route (now previous.slug looks non-built-in, allowed).
- POST /apps and PATCH /apps/:id reject externalUrl values that are
  not http:// or https:// (code='invalid_external_url'). Prevents
  shipping javascript:/data:/file: payloads tenant-wide via launcher.

Other:
- New SPA route /embedded/:id and embedded-app page (iframe host with
  back + open-in-new-tab + error/not-embeddable states).
- OpenAPI App / CreateAppBody / UpdateAppBody extended; codegen ran.
- en/ar locales: admin.appImage, appExternalUrl, appOpenMode.*,
  builtinPathLocked, embeddedFrame.*.
- scripts/src/seed.ts: drift guard throws if a seeded built-in slug
  uses a route that does not match the hardcoded SPA route.

Tests:
- New API test apps-builtin-route-lock.test.mjs (6/6 pass): reject
  built-in route change, allow non-route built-in updates, allow
  non-builtin route changes, reject built-in slug change (anti-bypass),
  reject non-http(s) externalUrl scheme + accept https, allow built-in
  same-route no-op.
- New Playwright E2E admin-app-image-external-embedded.spec.mjs
  (passes): launcher renders custom image_url, external_tab opens
  external URL via window.open, external_iframe navigates to
  /embedded/:id and renders <iframe src=externalUrl>.

Out of scope (pre-existing, not introduced here):
- Failing tests in executive-meetings-* and tsc errors in
  api-server/src/routes/executive-meetings.ts.
 2026-05-12 12:43:21 +00:00
Riyadh 50709fb5e2 Task #517: App image upload, external links, built-in route lock 
Admin Add/Edit App now supports:
- Custom image upload (or fall back to Lucide icon) via the existing
  ServiceImageUploader; rendered on the home launcher when set.
- Open mode picker: internal (default), external_tab (window.open),
  external_iframe (renders inside /embedded/:id). External URL input
  shown conditionally and required when an external mode is chosen.
- Internal route field is hidden entirely when an external mode is
  selected, and locked (readOnly + lock hint) when editing a built-in
  app. Slug input is also locked (readOnly) for built-in apps so the
  built-in identity cannot drift via the form.

Backend:
- apps schema gains image_url, external_url, open_mode (default
  'internal'); drizzle-kit push applied.
- New lib/db/src/built-in-apps.ts exports BUILTIN_APP_SLUGS +
  isBuiltinAppSlug. Exposed via subpath export
  `@workspace/db/built-in-apps`; the file has zero imports so the
  browser bundle uses it without pulling in `pg`. tx-os now imports it
  directly — duplicate FE constant removed.
- Built-in slug list: services, notifications, admin, notes,
  my-orders, orders-incoming, executive-meetings (everything with a
  hardcoded <Route> in artifacts/tx-os/src/App.tsx). calendar /
  documents are seeded but admin-defined and remain editable.
- PATCH /apps/:id rejects route changes whose previous slug is
  built-in with 400 + code='builtin_route_locked'. Same-route no-op
  is allowed; non-route updates on built-ins still work.
- PATCH /apps/:id ALSO rejects slug changes when the previous slug is
  built-in (code='builtin_slug_locked'). UpdateAppBody zod schema
  intentionally omits slug, so we inspect req.body.slug raw before zod
  stripping. Closes the 2-step bypass: rename slug (allowed) → change
  route (now previous.slug looks non-built-in, allowed).
- POST /apps and PATCH /apps/:id reject externalUrl values that are
  not http:// or https:// (code='invalid_external_url'). Prevents
  shipping javascript:/data:/file: payloads tenant-wide via launcher.

Other:
- New SPA route /embedded/:id and embedded-app page (iframe host with
  back + open-in-new-tab + error/not-embeddable states).
- OpenAPI App / CreateAppBody / UpdateAppBody extended; codegen ran.
- en/ar locales: admin.appImage, appExternalUrl, appOpenMode.*,
  builtinPathLocked, embeddedFrame.*.
- scripts/src/seed.ts: drift guard throws if a seeded built-in slug
  uses a route that does not match the hardcoded SPA route.

Tests:
- New API test apps-builtin-route-lock.test.mjs (6/6 pass): reject
  built-in route change, allow non-route built-in updates, allow
  non-builtin route changes, reject built-in slug change (anti-bypass),
  reject non-http(s) externalUrl scheme + accept https, allow built-in
  same-route no-op.
- New Playwright E2E admin-app-image-external-embedded.spec.mjs
  (passes): launcher renders custom image_url, external_tab opens
  external URL via window.open, external_iframe navigates to
  /embedded/:id and renders <iframe src=externalUrl>.

Out of scope (pre-existing, not introduced here):
- Failing tests in executive-meetings-* and tsc errors in
  api-server/src/routes/executive-meetings.ts.
 2026-05-12 12:40:20 +00:00