A subscription froze its planVersionId at sale (reproducible pricing). There was
no way to move a sold sub onto a different VERSION of the SAME plan — needed when
an admin publishes v2 with different timeframes (e.g. mujor-naten-cdo-dite v1
"every day" → v2 "weekdays only") and wants an existing subscriber on it, or back
on v1.
Backend (PUT /api/subscriptions/:id):
- accept planVersionId; honored only with the subscription:plan permission
(stronger than subscription:update — a plan-management action). Non-privileged
caller sending a change → 403, not silently dropped.
- validated to belong to the sub's EXISTING planId (a different plan = a
different price basis = a re-sale → 400).
- price/currency/period/planId stay frozen; only planVersionId moves. The swap is
server-logged for audit (the row is mutable master data, not on the ledger).
Past signed entry/exit events keep their own windowTariffVersionId, so history
reprices identically — only future access uses the new version's windows.
Frontend (SubscriptionManager):
- pass the session user through the route (like RolesManager).
- admin-only "Versioni" picker in the edit modal: lists every version of the
sub's plan by effective date + a timeframe summary (days + window, or 24/7),
current pre-selected. The plan itself stays read-only. Sends planVersionId only
when it changed.
- i18n: subs.version/versionHint/versionCurrent/versionOnlyOne/everyDay/allDay
in both sq + en.
Verified on a writable DB copy: version changed, price + planId frozen,
cross-plan version rejected. Live DB untouched. build+lint 14/14.
Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
An out-of-window subscriber entry stamped a FIXED windowOwedMinor = the whole
gap to window-open (e.g. 800 ALL for a 13:21 arrival to a 20:00 window) and
deferred it to exit. That over-charged anyone who left before the window
opened — a 1-hour visit was billed as 6.5 hours.
The amount isn't knowable at entry: a subscriber may enter early, leave after
an hour, come and go several times before the window opens, and linger past
window-close. They should pay only for the minutes actually parked outside the
window (capped at the window edges) — exactly what minutesOutsideWindow already
computes.
So the entry now stamps a MARKER only (outOfWindow: true + windowTariffVersionId
for reproducible pricing), no fixed amount. The exit gate and booth quote price
it live via windowOwedBetween(entry → settle-time), which already caps at the
window edges (early entry stops accruing at window-open; the in-window portion
of a crossing stay is free; the late-exit tail keeps accruing until payment).
Both already called that one function, so they agree.
- subscription-flow: entry stamps outOfWindow marker; the advisory slip is now a
scannable out-of-window TICKET (Code128 + QR of the occurrence id).
- shared LedgerPayload: add outOfWindow; mark windowOwedMinor/windowGap*/
windowCurrency deprecated read-only (historic signed events still type-check).
- BoothScreen: window-charge badge keys on outOfWindow (or the old stamp).
- ActiveSessions: drop the always-on "Open barrier" for subscribers — the
assist-open / window-charge payment live in the pay modal, so the list can't
one-click past an unpaid out-of-window charge.
Verified the live model on a DB copy: 13:21→14:30 = 200 ALL; 19:55(in grace)→
23:00 = 0; 19:00→21:30 (crosses into window) = 100 ALL. Existing signed
occurrences left untouched (immutable). build+lint 14/14, shared 87/87.
Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
Add apps/desktop, a thin Tauri v2 shell wrapping the SAME @parking/web SPA so
the desktop and browser UIs never drift: dev loads the Vite dev server (HMR),
prod bundles the web app's dist/. No business logic in the shell (device/auth/
ledger stay in @parking/server); deny-by-default capabilities.
apps/web (single UI source of truth):
- lib/origin.ts: centralize the backend origin (API_BASE/apiUrl/wsUrl from
VITE_API_BASE); no-op in the browser, lets the desktop build target Fastify.
- lib/kiosk.ts: block the right-click context menu in PROD only (dev keeps it +
devtools).
- lib/desktop-updater.ts: prompt-on-update auto-update (no-op in browser/offline)
→ downloadAndInstall + relaunch; i18n update.* keys (sq+en).
- .env.production: VITE_API_BASE wired to the Fastify origin for the bundle.
Desktop:
- window starts maximized (not fullscreen — operator keeps OS access).
- auto-update via tauri-plugin-updater + -process; self-hosted endpoint is a
PLACEHOLDER to fill in. Updater keypair: pubkey embedded in tauri.conf.json;
private key + password kept OUTSIDE the repo (~/.parking-updater-keys) and as
TAURI_SIGNING_* build secrets.
- Turbo build is a no-op; the real signed bundle is `pnpm --filter
@parking/desktop bundle` (verified → .deb/.rpm/.AppImage + .sig signatures).
Verified: cargo check clean; turbo run build lint 14/14 green; i18n parity holds;
no key/sig/bundle artifacts in the repo.
Wiki (security + desktop analysis recorded alongside):
- new concepts/tpm.md (TPM 2.0: how it works, sealed-LUKS auto-unlock + non-
extractable signing key, limits — live-root, bus-sniff — TPM-vs-ATECC608 by
platform).
- new decisions/desktop-shell-tauri.md (Tauri v2 over Electron; best-case Ubuntu
26.04 LTS, worst-case Windows+WSL → kiosk browser; full as-built).
- pull-the-disk attack trace on append-only-event-chain; ATECC608 not-in-a-PC
caveat; cross-links from disk-os-hardening / threat-model.
- open-questions #11 (appliance WebKitGTK), #12 (TPM hardening impl), #13
(startup verifyChain self-check); index/overview/log/standing-decisions.
Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
A subscriber entering outside their plan's window owes a deferred charge, but
nothing printed — they had no paper proof a fee was pending. Print a best-effort
ADVISORY slip at entry ("PARKIM — JASHTË ORARIT"): holder, entry time, "entered
out-of-window (window opens HH:MM)", and the key line "⚠ fee computed at exit"
+ the occurrence number. It is NOT a payable ticket and carries NO amount — the
total is computed at the booth on settlement, combining early-entry AND any
late-exit time into one number (windowOwedBetween over the whole stay).
Best-effort like the Z-report / subscription card: printed AFTER the barrier
opens and fully swallowed, so a missing/failed printer never blocks entry. New
printWindowChargeNotice (booth-print.ts) via the generic printReport; wired into
the subscription entry flow when an out-of-window entry charge applies.
(The "both charges at the booth" requirement was already satisfied by the
windowOwedBetween fix — verified: early-entry + late-exit minutes combine in one
calc at lookup/exit. This commit only adds the entry paper trail.) Build+lint 12/12.
Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
The timeframes model was a coarse weekday/weekend split, which couldn't express
"open Saturdays" or different rules on a specific day — and it didn't match the
V2 tariff, which already has a proper per-day-of-week picker (Hën–Die).
Replace PlanTimeframes { weekday, weekend } with { days[], fromMin, toMin }: the
allowed window applies only on the selected days (0=Sun..6=Sat; empty = every
day); on unselected days the subscriber parks free. A "night plan, free
weekends" is just days [Mon..Fri] with a 20:00→08:00 window — the exact case
from before, now expressible alongside any other day combination.
outOfWindowGap reworked to the days model (per-day membership test instead of
the weekend helper); the plans editor reuses the tariff composer's Mon-first
checkbox row and the shared tariff.dow0..6 labels. No production plans carry
timeframes yet (feature shipped today), so the shape changed directly with no
migration. Unit tests updated + extended (Saturday-only, every-day, weekday
night); 81 shared tests pass. Build+lint 12/12.
Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
Three subscriber enhancements driven by real scenarios (migration 0011, all
additive columns — backward-compatible).
1. QUANTITY. One subscription covers N cars (a family pays once for two). Sale
amount = span price × quantity; maxConcurrent defaults to the quantity so all
N cars can be inside. Quantity rides in the payment payload.
2. PLAN TIMEFRAMES → TARIFF BRIDGE. A plan may restrict WHEN a subscriber may
park (e.g. weekday 20:00→08:00, weekend all-day). A scan outside the window is
NOT refused — the out-of-window minutes are charged at the normal TRANSIENT
tariff (the subscriber is a transient for that time):
- early entry: arrival → window-open, DEFERRED (signed as windowOwedMinor on
the vehicle_entry payload), collected at exit;
- late exit: window-close → departure, and exit is GATED
(sub.refused.unpaidWindow) until paid at the booth.
Pure, tz-aware outOfWindowGap in @parking/shared (12 unit tests); pricing
reuses computeFee + the active tariff version
(apps/server/src/subscription-window.ts). The exit refusal is a host-ONLINE
business gate — the fail-open rule still governs the offline path.
3. RESERVED SPOTS. Site toggle reserve_subscriber_spots: occupancy holds
max(0, quantity − itsCarsInside) per active subscription, so transients see
"full" sooner; effectiveFree = capacity − count − reserved. Subscribers are
never gated by full.
UI: quantity field + ×N quote (SubscriptionManager); timeframes editor
(SubscriptionPlansManager); reserve checkbox (SiteSettings); booth pay modal
shows an "OUT-OF-WINDOW" charge and takes payment to clear the exit gate.
Verified on a copy of the live DB: qty 2 = 2× price; a night-plan 19:30 entry →
30min/15,000 ALL owed, stamped + paid → gate clears, chain verifies; the reserve
toggle holds a qty-2 sub's 2 spots. Build+lint 12/12; 80 shared tests pass.
Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
Re-model subscription pricing from per-row, operator-typed prices into an
admin-composed, versioned PLAN CATALOG (the tariff pattern). The operator now
SELLS by picking a plan over a date span; the price is LOOKED UP, never typed —
removing the fat-finger risk on a money field — and day/week/month periods make
the hotel "guest stays 1–N days" case a daily plan over a check-in→check-out span.
- Schema/migration 0010: new `subscription_plans` (immutable, effective-dated,
keyed by a stable planId; period day/week/month + per-period price + active
flag). `subscriptions` gains planId/planVersionId; period enum widened. Seeds a
"Monthly" plan from the existing site default price (no data loss).
- Pricing (pure, unit-tested in @parking/shared): periods = ceil(span / period),
amount = periods × per-period price. Ceil = any started period is full (hotel
practice). `resolvePlanVersion` picks the latest active version ≤ sale instant.
- Backend: new admin-only plan CRUD (`subscription:plan` permission); reworked
sell path derives the amount from the plan; `POST /api/subscriptions/quote`
returns a server-computed quote so the operator can't override it. The
signed-payment sale fix is unchanged — only the amount SOURCE moved; payload
now carries planId/planVersionId/periods. Updates never re-sell (price frozen).
- Frontend: SubscriptionManager sell form swaps the price field for a plan
picker + start/end dates + a live quote line. New SubscriptionPlansManager
(Setup tab) for the admin catalog. i18n (sq+en) for both.
Verified on a copy of the live DB: 0010 applies (existing subs intact), a
3-night hotel sale prices to 2,400 ALL, appends one signed payment with
planVersionId, chain verifies. Build+lint 12/12; 68 shared tests pass.
Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
Creating a priced subscription wrote only the mutable `subscriptions`
master row and appended NOTHING to the signed ledger — so the cash an
operator collected showed in the live feed, drawer, and shift Z-report
nowhere, leaving no signed trace. A booth operator could sell
subscriptions and pocket the money untraceably — the exact
operator-as-adversary path the append-only signed ledger exists to close.
Found live: 3 priced subscriptions (27,000 ALL) had zero payment events.
Selling a priced subscription now appends a signed `payment` event at
create time: amount = priceMinor x months (full multi-month prepay),
operator-chosen tender (cash->drawer / card->bank), payload
{ subscriptionSale: true, permitId, operator, months }. Folds into the
shift Z-report/drawer with no new summing logic; the feed badges it
"subscription sale" and resolves the holder name. The create response
returns the recorded { sale }; subscriptionRoutes now takes the EventLog
and ShiftService.
Not hard-gated on an open shift (a sale can happen outside the booth money
path) — it warns instead. The 3 historical off-book sales are not
back-fillable (append-only forbids forging dated events) — reconcile via
cash_movement or a Z-report note.
Verified against a copy of the live DB with the real signing modules:
signed payment appended, hash-chain still verifies, lands in shift cash
totals. Build + lint 12/12.
Wiki: subscription "Collecting the fee" deferred -> BUILT (+ the off-book
hole and why); shift sale-folds-in; threat-model worked example
("store the price != account for the sale").
Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
The ANPR plate was saved (device_events kind:"read") but had no UI. Extend
GET /api/snapshots/by-identity/:identity to also return plates[] (plate, confidence,
region, direction, snapshotId, at) for that session, and render each as a cyan
"Plate: AA558EE 100%" chip in the SnapshotStrip — so it shows in both the booth
event-detail modal and the pay modal, beside the evidence photo, no separate screen.
Deduped by plate+direction; session:read gated; i18n sq+en.
Verified: by-identity returns plates[] for a seeded read (200, AA558EE 0.999 Albania
entry). Build + lint green.
Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
Rework the ANPR trigger to the real design: when a transient presses the button or a
subscriber passes QR/RFID, the entry/exit fires and takes its evidence snapshot — that
is the moment to recognize. snapshotAsync now takes the VisionClient and, after storing
each snapshot from an opt-in (config.anpr) camera, runs ANPR on the SAME image and
records the plate against the SAME session identity (device_events kind:"read" with
plate/confidence/region/snapshotId/source:"entry-exit-snapshot"). One image serves both
evidence and plate extraction; recognition fires only on a real entry/exit — no polling.
The entry/exit/subscription flows take an optional VisionClient and pass it through;
server.ts wires it. Removed the polling VisionReader and VISION_POLL_MS/VISION_DEDUPE_MS.
Advisory + fire-and-forget: a low-confidence/no-plate result records nothing, a vision
failure never delays or changes the open, and the plate does not feed the access
decision. Verified e2e: a simulated entry snapshot on an anpr camera (live fast_alpr)
stored the snapshot for the session and recorded {identity, plate:AA558EE, 0.999,
region:Albania, snapshotId}. Build + lint green.
Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
Make the vision service genuinely configurable (was env-only).
- SetupWizard: an "ANPR" checkbox on the camera form (writes config.anpr; persisted
only when on; sq+en) — opt-in is no longer raw JSON.
- DeviceMonitor optionally takes the VisionClient and probes /health each tick, emitting
a "vision" pseudo-device → a Vision chip (ready/degraded/offline + recognizer) in the
booth footer when VISION_ENABLED, no chip when off. Widened the DeviceStatus category
union (server + web) + footer maps + devices.catVision. Verified: ready/fast_alpr when
up, 0 chips when disabled.
- apps/vision/.env.example (Python service) + a VISION_* block in apps/server/.env.example
(Node side) + a Configuration section in opencv-anpr-service.md covering all four
layers and the caveats: the two processes share the VISION_ prefix but need SEPARATE
.env files; bind /analyze to 127.0.0.1; cache model weights at deploy; an unbound anpr
camera recognizes but every read is refused.
Build + lint green.
Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
Answers "is a recognized plate saved?" — now yes, for both transient and subscriber, as
an ANPR audit trail independent of whether it matched anything.
VisionReader now stores the snapshot bytes in `snapshots` keyed by identity=PLATE — the
same identity the flow signs its anomaly/event with — so GET /api/snapshots/by-identity/:plate
(the booth event-detail modal's snapshot strip) shows the car's photo against that
anomaly with no UI changes. It also records an unsigned device_events{kind:"read"}
breadcrumb (plate, confidence, region, model, snapshotId, and the dispatch outcome) as a
queryable recognition log. Switched from emitRead to calling ReadDispatcher.dispatch
directly (like qr-reader) to capture that outcome.
Non-blocking: a refused read (no session / unpaid / unknown plate) just returns
rejected — no barrier hold — and is logged with its snapshot for investigation. Plate
stays advisory (exit demands payment; subscription matches only a bound plate).
Verified e2e: a recognized AL plate with no open session signed exit.refused.noSession
(identity=plate), stored a 555KB snapshot under that plate, recorded the read breadcrumb
(accepted:false, reason "no open session"), and by-identity returned the image — the
refused read is fully investigable with its picture. Build + lint green.
Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
A VisionReader polls each opt-in camera (config.anpr===true, off by default) every
VISION_POLL_MS, captures a snapshot, recognizes via VisionClient, and on a confident
plate emits deviceEvents.emitRead({kind:"plate", value}) — the same event a physical
plate reader sends, so the existing ReadDispatcher routes it to the subscription/exit
flow unchanged (no flow rewrite).
The plate stays advisory by construction: the exit flow still demands a covering
payment, the subscription flow only matches a bound plate. Guards: low-confidence reads
dropped; debounce (VISION_DEDUPE_MS) so a parked car doesn't re-fire; per-camera
in-flight guard; idle when vision is off or no camera opts in. #recognizeOn is public
for a future on-demand (loop-edge/API) trigger.
Verified end-to-end: an in-memory anpr camera (AL plate image) + live fast_alpr service
→ VisionReader emitted exactly one {kind:"plate",value:"AA558EE"} onto the bus; debounce
held it to 1 emit over 7 polls. Build + lint green. Updates opencv-anpr-service
(trigger-wiring + per-camera opt-in marked done).
Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
Node-side adapter to the apps/vision ANPR microservice (localhost HTTP: POST /analyze
with snapshot bytes, GET /health), returning a normalised VisionResult or null. Enforces
"advisory, never sole authority" at the boundary: opt-in (VISION_ENABLED, default off),
fail-soft (any error/timeout/unreachable → null, never throws into the lane → ticket
fallback), and re-applies the confidence floor (VISION_MIN_CONFIDENCE) on top of the
service's own low_confidence flag. Per-request AbortController timeout so a slow call
can't hang the barrier. Constructed in server.ts.
Verified: fail-soft (disabled/unreachable → null, no throw) and live end-to-end (Node
client → running fast_alpr service → AA558EE 0.999, region=Albania). NOT yet wired into
the read bus — the opt-in snapshot→DeviceReadEvent{kind:"plate"} trigger is the next
step. Build + lint green. Updates opencv-anpr-service (adapter gap marked done).
Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
Record the verdict: the ANPR service is worthy to consume NOW as an advisory plate
IDENTITY source (Job 1) — the flows already treat a kind:"plate" read as first-class
(exit signs source:"lpr"; subscription matches read plate vs bound plates), so it feeds
an existing input with no flow rewrite. It is NOT worthy as the sole authority to open a
transient barrier (a plate is not a payment; spoofing needs Job 2 vehicle verification,
unbuilt) — gated by the confidence floor with ticket/manual fallback. Lists the four
gaps before consumption (VisionClient adapter, opt-in trigger, field accuracy,
weight-provenance). Next step is the adapter, not more model work.
Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
Benchmarked fast-alpr's four fast-plate-ocr models via the full pipeline on real AL
plates (AA558EE, AA687KE), CPU. All four read both correctly; the default
cct-xs-v2-global-model wins on confidence (0.999/1.000) AND speed (33-39ms) and returns
region=Albania. The "European 40+country" model is WORSE here (~0.77 confidence, one
synthetic misread) — overturning the "EU model better for AL" assumption from the prior
research. Decision: no config change. Resolves the AL-accuracy-benchmark open item
(results table + finding added to opencv-anpr-service); weight-provenance remains the
one open recognizer item. Re-benchmark on real on-site captures once cameras installed.
Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
Settle WHERE the host-side ANPR service lives and how it joins the build: in this
monorepo at apps/vision/ (not a separate repo), still a separate OS process called
over localhost HTTP, wired into the Turbo graph via a thin package.json shim whose
scripts shell to Python tooling (uv/uvicorn/ruff/pytest). Co-located source honors the
vision-service runtime+license isolation decision (AGPL reach is a linking boundary,
not a folder); the fast-alpr MIT baseline removes most of the split-repo pressure
anyway. New page vision-service-packaging; updates vision-service, opencv-anpr-service,
the CLAUDE.md layout, index, log. Not built yet — packaging decision only.
Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
Research note from the recognizer-options query. fast-alpr v0.4.0 (MIT) — a swappable
YOLOv9-detector + CCT-OCR pipeline on ONNX Runtime, CPU-only and offline — fits the
decided vision-service architecture and is MIT end-to-end (code + published weights),
so the ANPR path may not need the scoped AGPL exception. Flags the open caveats:
verify model-weight provenance, and benchmark AL-plate accuracy (default global vs.
the 40+ country EU model). fast-alpr is plate-only, so the vehicle-verification job
stays ours to build. Decision kept open. Updates opencv-anpr-service (new "Recognizer
evaluation" section + licensing nuance), vision-service (open/next), index, log.
Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
Sync the two canonical reference pages with this session's features: local-jwt-auth
gains the new log resource / log:read permission in the RBAC grid (links app-logs);
first-run-setup notes the one-car-one-ticket presence-loop/cooldown guard the admin
configures on a relay (links entry-double-press).
Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
The dynamic-RBAC management routes are themselves grantable (role:* and
user:*), so a non-admin holding them could self-escalate: edit their own
role to add a permission they lack, mint a privileged role, assign someone
the admin role, or reset/delete a more-privileged account. Found by the
commit security review (2× HIGH).
Fix — enforce the RBAC invariant "you cannot grant beyond yourself":
- roles.ts: role:create/update reject any permission not held by the caller
(escalates()). An admin holds the full set, so it stays unrestricted.
- users.ts: user:create/update reject assigning a role whose permissions
exceed the caller's; update/password-reset/delete reject acting on a user
whose current role exceeds the caller's (exceedsCaller()).
The existing no-lockout + builtin-admin protections are unchanged.
Verified: 10-assertion inject test — manager (role:* + user:* but no
tariff:update, not admin) gets 403 on self-grant, minting a privileged role,
assigning/resetting/deleting an admin; admin stays unrestricted; the manager
can still create peers + in-scope roles (not over-blocked). Full build green.
Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
Replace the hardcoded role enum (admin/operator/cashier/readonly, checked
literally as requireRole("admin",...) across ~15 routes) with dynamic RBAC:
roles are DATA, route guards check a PERMISSION.
@parking/shared defines a code-defined grid: RESOURCES (user/role/tariff/
subscription/site/device/shift/payment/session/event/report) × Action
(create/read/update/delete + domain verbs void/cash) -> PERMISSIONS
(resource:action, e.g. tariff:update, payment:create, event:void).
DB: new roles + role_permissions tables; users.role enum -> role_id FK;
migration 0007_rbac (create tables, seed the builtin admin role + all 26
perms, seed operator/cashier/readonly composable roles matching old
behaviour, rebuild users to swap the column copying all rows).
auth.ts: JWT payload role -> roleId; permissionsFor(roleId) with an
in-memory cache + bumpPermsCache(); requirePermission(...perms) preHandler;
requireAuth for /me & /language; initAuth(db) wires the resolver once. Every
route guard mapped to a permission; device ingress (devices/qr-reader) stays
auth-free by design. New routes/users.ts (user:* CRUD, bcrypt 12, last-admin
guard) + routes/roles.ts (role:* CRUD, builtin-protected, perms validated
against the grid, cache bump on write). auth/me + /login return
{roleId, roleName, permissions, language}. seed-admin -> roleId:'admin'.
Frontend: SessionUser carries permissions + can() helper; router nav/route
guards gate by permission (requirePerm replaces adminOnly); SiteSettings
edit gated by site:update; new UsersManager + RolesManager (permission
checkbox grid; admin role locked); i18n nav.users/roles + blocks (sq+en).
Decisions: one role per user; protected built-in admin (no-lockout: the last
admin can't be deleted/downgraded); JWT carries roleId, perms resolved
per-request so role edits apply immediately (no re-login).
Verified: full build green; 20-assertion inject test passes (cashier 403s on
tariff publish + user list, admin passes, granting a perm applies on the next
request, last-admin + builtin-role protections return 409); migration 0007
applied to a copy of the live DB (incl WAL/shm) — existing admin maps to
role_id='admin', all rows preserved. Append-only event chain untouched
(event:void gates appending a void, not a delete).
Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
After a completed payment the customer always gets a transparency record:
entry time, payment time, duration parked, amount + tender. One shared
ESC/POS renderer (renderReceipt + ReceiptData in @parking/devices), two
modes: VOUCHER = those figures PLUS the scannable Code128 barcode and an
emphasised walk-back-grace line, so the one slip both proves payment and
self-exits at a distant exit reader (replaced the old barcode-only voucher);
STANDALONE = detail-only, auto-printed at payment when no voucher is issued.
Figures fold from the SIGNED ledger (latest payment event); printed on the
booth printer (failover to dispenser). Best-effort: a printer fault never
blocks the exit that already happened — the modal shows a note and offers
"Reprint receipt".
Server: booth-print.ts printPaymentReceipt() + receiptFigures(); routes
POST /api/voucher (voucher) + new POST /api/receipt (standalone/reprint).
Both ESC/POS drivers gained printReceipt(). Web: BoothPayModal auto-prints
after a non-voucher payment + reprint button; api.ts printReceipt().
CP852 fixes found on a real printout: (1) uppercase Ë was mapped to 0xEB
(that's ű) — correct byte is 0xD3; (2) Intl.NumberFormat injects a NO-BREAK
SPACE (U+00A0/U+202F) that isn't in CP852 and printed as "?" — line() now
normalises it to a plain space ("1000 Lekë"); (3) grace line wrapped
mid-word — split into two short lines.
Full build green; both receipt modes render-verified; routes live.
Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
The rongta-printer "Deployment" section still described a single
2026-06-14 unit. The live site now runs two: entry-dispenser 10.0.10.9
(Cashino, `cashino` ping-only driver) and booth-receipt 10.0.10.10
(Rongta, full status-page monitoring). Follow-up to 3e6773a.
Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
Rounds out subscriptions across enrollment, the barrier flow, and the booth.
- RFID credentials enabled with a "Read card" enrollment flow: the operator
arms ONE chosen reader (CredentialCapture, single-shot + ~30s TTL); that
reader's next read is captured into the form and NOT dispatched to the access
flow — the OTHER reader keeps serving live entry/exit. Routes:
/api/subscriptions/readers + /capture/{arm,cancel} + poll.
- Enter with one credential, exit with another: sessions are keyed by a
per-occurrence id (SUBSESS-<short>), not the credential value, with
permitId in the payload. Direction is decided by the barrier the reader sits
at (entry-lane→entry, exit-lane→exit; "both" infers); a fleet (maxConcurrent>1)
admits several cars and exits any with any credential, FIFO (oldest first).
- Booth treats a subscription occurrence as PREPAID: never quoted/charged; the
pay/exit modal shows a subscription mode (snapshots + a single audited
Open-barrier action) to assist a faulty exit reader / missing card;
reopenBarrier authorizes paidAt!=null OR subscription. Active Sessions badges
"abonim" and labels by holder name (not the raw key).
- Plus a per-read diagnostic log in the QR-reader route (serial → device →
verdict/dir), which surfaced the earlier duplicate-reader-IP misroute.
Verified via buildServer+inject + reader-scan/TCP-capture simulations
(enrollment isolation, cross-credential + FIFO fleet, prepaid-not-charged,
subscription reopen, unpaid-transient guard). Updated wiki (subscription,
booth-exit-flow). No migration.
Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
Builds out subscription credentials on top of the rename.
- Operator chooses the credential type; only QR is live (RFID shown disabled
"soon"). Backend/schema keep accepting both — re-enabling RFID is UI-only.
- QR codes are AUTO-GENERATED server-side (SUB-<base32>, crypto-random,
globally-unique-checked) — the customer/operator never picks the value.
RF stays operator-entered (the physical card id). Reader output decided =
TCP/IP full string (Wiegand-numeric fallback noted).
- Multi-month: form takes a `months` count → server sets validTo =
validFrom + N months (day-clamp); one record/one window; total = N×monthly.
- The QR card is PRINTED so the operator can hand it over: real ESC/POS 2D QR
(GS ( k) added to the Rongta driver (printSubscriptionCard); auto-print on
create (best-effort — never fails the create; returns {printed,printError})
+ reprint via POST /api/subscriptions/:id/print and a "Print code" button.
Verified via buildServer+inject incl. a TCP capture of the on-wire QR bytes
(autogen+uniqueness, Jan31+3mo→Apr30, auto-print, GS ( k QR with embedded
code, reprint, no-QR→409). Updated wiki (subscription, rongta-printer). No
migration.
Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
The "permit/lejet" feature is really a subscription. Full rename of the
mutable master data, plus a recurring monthly price.
- DB (migration 0004, data-preserving ALTER RENAME): permits→subscriptions,
permit_credentials/_plates→subscription_*, sessions.permit_id→subscription_id.
- Pricing: per-subscription priceMinor + period(monthly) + currency, with a
site default (site_config.subscription_monthly_price_minor) pre-filling the form.
- Server: subscription-flow.ts (SubscriptionFlow), routes/subscriptions.ts
(/api/subscriptions). Web: SubscriptionManager, route, i18n (sq Abonimet/en).
- The signed ledger `permitId` payload is intentionally kept — immutable
hash-chained history; renaming it would break verification of past events.
Deferred (wiki notes): fee collection into the ledger/shift (a shift-attributed
payment), LPR/ANPR plate source, time-of-day access windows (overnight subscriber).
Also carries the device-footer UI surface (api DeviceStatus, router mount,
i18n devices) due to shared-file overlap with the preceding footer commit.
Verified end-to-end on a fresh DB and migration on a live-DB copy (sessions
preserved). Live DB migrated. Full monorepo builds clean.
Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
File concept pages for the operator-UI architecture ([[booth-console]]: stack,
/api/ws live feed, anti-CSWSH) and [[i18n]] (per-user server-stored language;
resolves a dangling code-comment link). Qualify the stale 'plain React' note on
react-vite-spa. Backfill log entries for the live WebSocket, frontend foundation,
and i18n builds (which had none), plus a reconciliation lint entry. Catalog
booth-exit-flow + the two new pages in index; fix the concept count (27→41).
A parking lot is one pool of spaces with a flexible set of entry/exit
points — no "lane". Direction is a property of each RELAY inside an access
controller; readers/cameras bind to a controller relay and inherit it.
Schema:
- drop `lane` from ledger_events, device_events, sessions
- rename lane_devices -> devices (no lane/direction columns)
- access config.relays=[{relay,direction,button?}]; reader/camera
config.controllerId+relay binding
- fresh 0000_baseline migration (history reset; dev data was throwaway)
Signed ledger:
- remove `lane` from canonicalize(); bump signer keyId sw-hmac-v1 -> v2
(v1 events won't verify under v2 — intentional, gated per-event by keyId)
Server:
- new device-resolve.ts (replaces lane-map.ts): relayForButton,
relayForDevice, firstRelayByDirection, devicesByDirection
- entry-flow: button terminal -> its relay; exit/permit: reader's bound
relay; dispatcher resolves the bound relay + inherited direction
- camera snapshots fire by direction site-wide, async, never block open
- DeviceConfig widened to nested JSON for relays[]
Web:
- wizard: no lane selector; add controllers (relay map + entry-button
terminal) first, then bind readers/cameras/printers to a controller relay
Wiki: new entry-exit-points.md (replaces lane-direction); reworked
entry-exit-readers, parking-session, first-run-setup, device-registry,
append-only-event-chain, device-events; removed stale lane/LaneMap mentions.
The reader sends Connection: keep-alive but only acts on the verdict (beep,
output) once the TCP socket closes. Fastify's default kept the connection alive,
so the reader waited out a ~10s keep-alive timeout before beeping — even though
the server replied in ~15ms. Every vendor demo replies Connection: close and
shuts the socket. Set reply.header('connection','close') on the QR endpoint.
Verified the header is now sent; symptom was correct accept/reject with a ~10s
lag before the beep.
The QR reader is a push device and the setup wizard assigns random-UUID ids, so
'id = serial' can't be set via the UI. Add a dedicated gee-qr-reader driver
(reader category) with a single 'serial' config field; the admin assigns it
normally and enters the device's serial (its cjihao).
The QR endpoint now resolves the lane by matching lane_devices.config.serial to
the scan's cjihao (instead of row id == cjihao), so no DB hand-editing. An
unassigned serial resolves to no lane -> status:0, gracefully.
Verified via inject through the real /api/setup/assign: assign {serial:
H05M2AFA} -> .jsp scan with a matching permit QR -> status:1 (accept) + open;
re-scan -> permit exit; unknown card -> status:0; unassigned serial -> status:0.
Hardware capture: the GEE/Fondvision reader (serial H05M2AFA) scans + sends +
beeps fine — the earlier 'no beep' was just nothing answering :3000. Real
request: GET /qa/mcardsea.jsp?cardid=...&cjihao=H05M2AFA&... — the 'server
language' setting (JSP here) selects the URL EXTENSION, so it posts .jsp, not
.php. Our route was .php-only and would have 404'd it.
Register the endpoint at php/jsp/asp/aspx/cgi so it works whatever the device is
configured to. cjihao (serial) is the lane key: assign the reader as
lane_devices.id = its serial.
The reader HTTP-GETs on each scan and beeps/acts on our JSON reply (host-in-the-
loop, synchronous). New route GET/POST /qa/mcardsea.php parses the SDK query,
runs the scan through the read dispatcher (permit match -> permit flow; else
transient exit), and replies the SDK verdict: status 1=valid (beep 2x) /
0=invalid (beep 1x), output, time-sync.
Refactored the read flows to return a ReadOutcome {accepted, direction, reason}
so the reply reflects the real accept/reject decision (ReadDispatcher.dispatch,
ExitFlow.handleAt, PermitFlow.run). Fire-and-forget readers ignore it.
Reader's lane is keyed off its serial (cjihao) as lane_devices.id for now;
endpoint is public (reader has no auth, on the device subnet).
Verified via inject: valid permit QR -> status:1 + open; re-scan -> permit exit;
unknown QR -> status:0; barrier-less lane -> status:0.
The QRCode SDK v1.6.5 settles the reader protocol (supersedes the earlier
serial guess). On each scan the reader HTTP-GETs the host
(/qa/mcardsea.php?cardid&mjihao&cjihao&status&time); the host replies JSON
{data:[{...,status,output}],code:0}. Reply status 1=valid(beep 2x)/0=invalid
(beep 1x); output 0=Access/1=WG26/2=WG34; time syncs the clock. The GET's status
low digit is the direction (1=in/0=out).
Key: the beep/accept is decided by the SERVER REPLY, not locally -- the 'no
beep' during bring-up was a plain-text reply, not a scan failure. Host-in-the-
loop and synchronous. 'Server language' only selects the URL path; transport is
plain HTTP.
New source page qrcode-sdk; updated gee-qr-er80 (protocol resolved), index.
The reader on hand is a GEE-QR-ER80 QR/DataMatrix/1D barcode access reader
(not an EM4100 prox-card reader as first guessed). Interfaces: Wiegand 26/34,
RS-232, RS-485, USB, TCP/IP; 4-15 VDC; Linux-supported. Variant on hand: -Q-W
(QR scanner, Wiegand/RS-232/485).
This is the QR-ticket scanner the design already needed: a host-side reader
whose scans become read-bus events consumed by the (already-built) exit flow
and QR-permit path. Prefer RS-232/485 over Wiegand (Wiegand can't carry a
variable-length QR string; autonomy is moot with the no-ACL Dingtian).
New source + entity pages; updated ticket-encoding, entry-exit-readers, index.
Open (blocks the adapter): the RS-232/485 frame + baud (ASCII CR/LF expected).
A permit is an aggregate (row + credentials + bound plates); create/update
treat it as one unit (child sets replaced on update). GET /api/permits (any
signed-in role, for lookup); POST/PUT/DELETE + POST /:id/revoke (admin only).
Validation: maxConcurrent positive-int-or-null (unbound); a permit must have at
least one credential OR one bound plate. Revoke is the soft common case (keeps
history, barred at the barrier); DELETE hard-removes — past ledger events that
reference it are untouched (append-only audit trail, independent of this row).
Web PermitManager in the admin shell: list + add/edit (holder, car-bound toggle,
validity, credentials, plates), revoke, delete. Makes permits usable without
hand-seeding (companion to the tariff composer).
Verified via inject: validation (empty / maxConcurrent=0 -> 400), create -> 201,
operator can LIST but not write (403), update replaces child rows, revoke ->
revoked, delete -> 204 then 404 with children cleaned.
A credential read now routes by what the credential IS: matches a permit
(card/QR credential or a bound plate) -> permit flow; else -> transient exit
flow. Lane resolved once (readerLaneWithAccess); ExitFlow.onRead -> handleAt so
the dispatcher owns lane resolution.
Permit direction is inferred from session state for that car (the read value is
the per-car session key): no open session -> ENTRY (enforce maxConcurrent, sign
vehicle_entry, open); open -> EXIT (sign vehicle_exit, open, close). Fleet
permit = one session per car; anti-passback falls out naturally.
maxConcurrent enforced as a fold over the signed ledger (null = unbound).
Validity window + status + plate-OR-card identity as designed. No ticket/fee;
every use is a signed event carrying permitId. Refusals (revoked / out-of-window
/ at-capacity) are signed anomalies, barrier stays closed.
Verified against stubs: card entry -> inferred exit; fleet cap 2 (F3 rejected
at 2/2, then admitted after F1 exits); plate-bound opens; revoked rejects;
unknown credential falls through to exit reject; verifyChain ok.
Booth reality breaks a fixed clock (relief late/absent, forced double shifts),
and a shift is a separate explicit boundary. Drop expiresIn from the global jwt
config and from login; the token carries no exp. Cookie maxAge = 30 days so a
browser restart doesn't log out an active operator; logout still clears it.
Replace the camera stub with HttpCamera: Hikvision ISAPI and Dahua CGI
snapshots over client-side HTTP Digest (new drivers/http-digest.ts).
healthCheck() now pulls a real frame instead of returning ready/stub.
Snapshot carries bytes (driver fetches); storage/imageRef is the caller's
job, keeping the adapter free of storage deps.
Fix the cosmetic Backend-push-IP field: add pushesToBackend to DeviceDriver
(only Dingtian sets it), expose as pushCapable in the catalog, and gate the
wizard's backend-IP fetch + field on it so pull-only devices hide it.
Verified on hardware (Hikvision 10.0.10.121): healthCheck ready,
captureSnapshot returns a valid JPEG.
Fix two bugs found running the real assign flow: the saved web password
didn't match the device (login stayed admin/admin), and the UDP2 warning
never reached the admin.
Web password:
- Split the conflated field into webPassword (the DESIRED login; blank ->
auto-generate) and webPasswordCurrent (the device's EXISTING password used
as the old cred, default admin). Before, an admin typing a desired password
made harden send it as the old cred -> rotation failed -> but the DB still
saved the typed value, so it claimed a password the device never accepted.
- harden() now rotates current -> desired, VERIFIES by re-authenticating with
the new password, and only returns secrets.webPassword on success (else a
warning, nothing saved). Stores webPasswordCurrent for future re-runs.
- assign strips the typed webPassword/webPasswordCurrent and persists only the
verified secret -- the DB never claims an unapplied password.
Warnings to the UI:
- assignDevice returns warnings[]; SetupWizard shows them in an amber
"saved, but action needed" banner per category. This is how the admin learns
the firmware wouldn't disable UDP2 (finish in the device web UI).
Verified on hardware: after harden the device rejects admin/admin and accepts
the chosen password; the UDP2 warning surfaces.
The string protocol (UDP 60001) has no password field but can fire relays
("11" = relay 1 on), bypassing relay_pw entirely. Proven on hardware: an
unauthenticated packet opened a relay. harden() had left it enabled "for
status reads".
- #status() now reads via the authenticated binary command (relay cmd 0x00)
instead of the string protocol, so the string protocol is no longer needed.
- harden() disables the string protocol (udp2.p=255). BEST-EFFORT: firmware
V3.6J's config API silently refuses to disable udp2 (the device web UI can),
so it's not part of the blocking verify -- harden() re-checks and returns a
warning instead of throwing. After a web-UI disable, the attack is dead and
binary control/status still work (verified on hardware).
- HardenResult gains an optional `warnings[]`; the assign route surfaces them
to the admin and logs them.
- Corrected the false comment claiming relay_pw stops an attacker (it is
defence-in-depth on plaintext UDP, not a boundary).
- Thread localAddress through the driver's UDP/HTTP calls so a multi-homed
host sources device traffic from the device-facing NIC.
- Device web login (webUser/webPassword) is no longer redacted from setup
state -- it's an operational credential for the admin-only device area;
pushPassword/relayPassword stay machine-only.
Wiki: document the vuln + fix, the firmware caveat, and the out-of-band
actuation gap (the log captures host actions only; reconciliation vs. an
independent witness is the real control and is not yet built).
Add the rongta PrinterDevice driver (ESC/POS over raw TCP 9100) and the
device-agnostic pieces around it:
- Roles + failover: each printer declares a role (entry-dispenser/booth-
receipt) and failoverRank; printer-routing.ts picks the best healthy printer
and falls back outside->booth for entry tickets (never the reverse).
- Live status: MonitorableDevice.readStatus()/PrinterStatus capability. The
Rongta driver scrapes the device's own /prn_stat.htm (Cover/Cutter/Paper
End/Near End/Off-Line) rather than hand-decoding DLE EOT, whose reply bytes
on this clone don't match the canonical ESC/POS bit layout (verified on
hardware) -- avoids a false-healthy. Maps to ready/degraded/offline, fail
safe on an unreachable or unexpected page.
- Server PrinterMonitor polls enabled printers (PRINTER_POLL_MS, default 5s),
caches latest, emits "printer-status" on change. Exposed via
GET /api/printers/status and an SSE stream for the booth UI.
Verified against 10.0.10.6: ready when healthy, offline when unreachable
(no throw), bus emits on change and suppresses unchanged reads.
Wiki: new rongta-printer entity, printer-roles-failover and
printer-status-monitoring concepts; BOM/index/log updated.
harden() now rotates the device's default admin/admin web-UI login via
GET /userset.cgi?<old>&<old>&<new>&<new>& (best-effort: a failure logs
and doesn't fail the assign). The new password is stored back in config
(webUser/webPassword) so a re-run can rotate again, and is stripped from
the assign response like the push secret.
Documented the load-bearing caveat: this device's CGI API is fully
UNAUTHENTICATED — config read/write, relay fire, and userset.cgi itself
all return 200 with no credentials (verified on hardware). admin/admin
gates only the browser UI, and there's no inbound-auth setting (only
session_en, which bricks the read API). So the rotation is defence-in-
depth for the UI, NOT a boundary; the signed event log remains the real
anti-fraud guarantee. Verified rotation end-to-end on 10.0.10.5
(success &0&, wrong-old-pw &2&); device left at admin/admin.
Lock down the relay device for the flat (no-VLAN) network.
Relay control:
- pulseOpen/setRelay now use the Dingtian BINARY protocol (:60000) with a
relay password — the only relay option with auth (string :60001 has none, and
is kept only for the read-only status query). Frame verified on hardware.
HardenableDevice capability (driver harden()):
- set a random relay_pw (1-9999); disable unused channels (rs485/can/tcp x2/mqtt
-> p:255), keeping UDP1 binary (control) + UDP2 string (status).
- write-verified (device reboots on apply).
Assign/Save flow now does: fix preconditions -> harden -> set up input push;
the relay password is stored in lane_devices so the runtime device can command
the relay.
DELIBERATELY NOT touching the device's HTTP CGI session check (session_en):
enabling it on this firmware breaks the config-READ API (ECONNRESET) and locked
the backend out — required a factory reset to recover. The open CGI API is
accepted as flat-network reality; the signed event log is the real guarantee.
Verified end to end on hardware: assign hardens + configures the device, config
API stays reachable, pulseOpen with the stored password fires the relay, without
it is rejected. wiki: device-input-flow + dingtian-relay updated.
The device pushes button events to the backend via its Input Link URL feature;
the backend decides. No polling — the chosen entry architecture.
packages/devices:
- dingtian driver: configureInputPush() writes the device's input_link_url
config (per-input server/port/path, en=1, active-LOW, plain HTTP) so each
input HTTP-GETs the backend on press/release. Extracted #readConfig/#writeConfig
(with the required command:setconfig injection + post-write reset tolerance).
apps/server:
- routes/devices.ts: public GET/POST
/api/devices/dingtian/:deviceId/input/:n/{on,off} — translates a device push
into an internal device event. Not behind cookie/CSRF (machine call from the
device); trust comes from the signed event log, not this request.
- device-events.ts: internal EventEmitter bus so the entry flow subscribes to
input events without coupling to HTTP. Wired into the server.
Verified on hardware: configured the device, then real presses on all 4 inputs
pushed to the backend (input N on+off, source = device IP). No polling.
wiki: device-input-flow concept (path + trust model for the flat/no-VLAN
network); dingtian-relay updated; index + log.
Neither UHPPOTE nor ZKTeco is used — the Dingtian relay controller was chosen
and verified. Remove their code and re-scope the wiki.
Code:
- delete access-uhppote.ts, uhppoted.d.ts, access.ts (zkteco/esp32-relay stubs),
and the three uhppote-*.mjs hardware test scripts.
- remove the `uhppoted` npm dependency from @parking/devices and @parking/server.
- unregister uhppote/zkteco/esp32-relay from the driver registry; drop their
exports. Catalog access drivers = dingtian only. Build green (5/5).
- refresh now-stale example comments (registry/interfaces/setup/api) to use
current examples; keep the two "UHPPOTE blocker" references that explain why
the precondition capability exists.
Wiki (kept pages, re-scoped):
- uhppote-controller, zkteco-controller -> rejected/historical with callouts;
uhppote-vs-esp32 -> historical (detection-vs-prevention lens still useful).
- re-point all "current device" framing (standing-decisions, bom, overview,
open-questions, device-registry, device-discovery, index) to dingtian-relay.
- transferable concepts (network-isolation, event-log-ingestion, barrier-not-a-
door, threat-model) untouched. Raw source immutable. Links lint clean.
The Dingtian board's inputs are independent of its relays (configurable), so a
button on an input can report to the host WITHOUT auto-firing a relay — solving
the access-controller-button-flow blocker the UHPPOTE/ZKTeco couldn't.
packages/devices:
- access-dingtian.ts: `dingtian` access driver implementing AccessControlDevice
(relay pulse/latch via UDP string protocol :60001), InputDevice (read inputs +
poll-based press/release events, active-LOW), and the new PreconditionDevice.
- PreconditionDevice capability on the interface: a device can report config it
requires for parking and optionally fix it. Dingtian checks input_link_relay
via the HTTP config API and can disable it.
- httpPort config field — the web/config API port is separate from UDP control
(this unit uses 8080, not the default 80).
- Register dingtian; export driver objects from the package.
Verified on real hardware (DT-R004 @ 10.0.10.172): status read, relay pulse,
input events; disabled input_link_relay via the driver, then confirmed pressing
inputs fires NO relay (0000) — host-in-the-loop entry works.
Config-write gotcha recorded: config_set.cgi requires "command":"setconfig"
injected after "status" (GET omits it) or the POST silently no-ops.
apps/server/scripts/dingtian-test.mjs: status / watch / pulse hardware test.
wiki: dingtian-relay verified; button-flow marked RESOLVED; index + log.
- dingtian-relay: relay+input controller (4ch on hand). Inputs are decoupled
from relays (configurable via input_link_relay) — solves the
access-controller-button-flow blocker the UHPPOTE couldn't. Full protocol from
the SDK (UDP string control :60001, `00` status parse, input_link_url push,
multicast discovery). Driver + hardware test still to build.
- dingtian-vs-mqtt: use direct HTTP/UDP now; MQTT skipped (broker = extra infra
+ failure mode + overkill at one-host/few-devices scale) but kept for later
multi-lane scale.
- autonomous-direction: record the roadmap to fully unmanned (no booth) and how
it reshapes the threat model (operator-fraud -> unattended-machine threats),
makes host-in-the-loop entry mandatory, and raises fail-state stakes.
- threat-model: note the unmanned shift. index + log.
gitignore the vendor SDK (dingtian/, 71MB of binaries/examples) — reference
only, protocol captured in the wiki.
Replace the dev-only token shim with real authentication.
Backend:
- @fastify/cookie; JWT carried in an HttpOnly + SameSite=Strict cookie
(parking_token), read from the cookie not the Authorization header.
- Double-submit CSRF: readable parking_csrf cookie + X-CSRF-Token header, both
cross-checked against a csrf claim baked into the JWT; enforced on mutations.
- Routes: POST /api/auth/login (bcrypt, constant-time-ish), POST logout,
GET me. requireRole now verifies the cookie + CSRF + role.
- seed-admin script (pnpm --filter @parking/server seed-admin) for the first
admin; no bootstrap endpoint.
- Removed SETUP_AUTH_BYPASS and catalog.authBypass entirely; setup endpoints
use the cookie admin guard like everything else.
Frontend:
- apiFetch wrapper: credentials:'include' + X-CSRF-Token on mutations.
- Login form; App gates on /api/auth/me and only shows setup to admins; logout.
- Wizard token field removed (auth is the session cookie).
Deploy:
- deploy/nginx.conf: prod reverse proxy, SPA + /api same-origin, TLS, so the
Secure cookies work. Dev stays same-origin via the Vite proxy.
Verified (curl + browser): wrong pass -> 401; login sets cookies; me -> admin;
assign without CSRF -> 403, with -> 201; no cookie -> 401; session persists
across reload. wiki/local-jwt-auth updated.
julian