- ThrottlerGuard runs first, so requests with wrong tokens or API keys
count against the rate limit before any database lookup.
- TRUST_PROXY (hop count or Express list) replaces the hard-coded 1.
- Every API answer except the public lot pages is Cache-Control: no-store
without ETag; helmet sends default-src 'none' / frame-ancestors 'none'
(Swagger keeps the page policy when enabled) and same-site CORP.
- Request log lines carry the path only - no query strings (the SSO
callback's code and state no longer reach the logs).
- In production the server refuses to start without an https
FRONTEND_ORIGIN or with COOKIE_SECURE=false.
Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
- Rotation claims the old token with a conditional update; of two
concurrent refreshes only one gets a successor, so a stolen token cannot
fork into two live sessions.
- A token rotated out less than 10 s ago (parallel tab) is refused without
revoking every session; later reuse still revokes them all.
- Sessions end at most SESSION_MAX_AGE (default 7 days) after login: the
login time is inherited by every rotated token and caps its expiry.
- The refresh cookie is cleared with the same attributes it was set with.
Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
- Per-account lockout: five failed attempts (password, MFA code or a
step-up confirmation) lock the account for 15 minutes, on top of the
per-IP throttle - guesses spread over many addresses no longer help.
- A TOTP code is accepted once: the time step is stored and that step or
any earlier one is refused (also between two concurrent requests).
- Login runs bcrypt on every path and answers unknown email, wrong
password, deactivated and locked accounts with the same message; failed
logins for unknown emails no longer write attacker-chosen text into the
audit log.
- Turning MFA off needs password and a current code; turning it on needs
the password. Both end all other sessions.
- New audit action ACCOUNT_LOCKED.
Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
- Deliveries connect through a lookup hook that checks the address
actually dialled, so a host that resolves to a private address after
registration (DNS rebinding) or a tampered row cannot reach internal
services. Redirects are still not followed.
- More blocked ranges (CGNAT/100.64/10 incl. cloud metadata, multicast,
reserved, benchmarking, NAT64, 6to4, site-local, IPv4-mapped IPv6);
only ports 80/443, no credentials in the URL, https in production.
- 5 s timeout, response body discarded; dispatch no longer waits for
receivers, so a hanging endpoint cannot stall the user's request.
- Signature covers "timestamp.deliveryId.body" with X-ApiCore-Timestamp
and X-ApiCore-Delivery headers, so receivers can reject replays.
- Signing secrets are stored AES-256-GCM-encrypted; older plain rows still
work.
Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
backup:configure could point automatic backups at any writable directory
(e.g. a statically served one) and set a cron of every second; an invalid
expression deleted the running job before failing, silently ending
backups.
- A path entered in the UI is resolved under BACKUP_ROOT (default
./backups) and refused if it leaves it; env values stay unrestricted.
- The schedule must be a five-field cron that runs at most hourly.
- reload() validates before replacing the running job.
Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
Backups carried every password hash, encrypted MFA secret and webhook
secret in plain JSON, and an import upserted arbitrary rows - including
users - so anyone holding backup:import, or tricking an admin into
importing a file, could overwrite the admin account with a known password.
- Export (download and scheduled) is AES-256-GCM over the gzipped bundle,
key from BACKUP_ENCRYPTION_KEY or derived from JWT_ACCESS_SECRET via
HKDF; header is authenticated, decompression is capped. Scheduled files
are written 0600 in a 0700 directory; run-now returns only the file name.
- Import requires the current password (and TOTP when MFA is on).
- Only a sealed file restores users, roles, grants, webhooks and audit
entries; an old unsealed file restores business data only and reports
what it skipped.
- Audit entries are append-only on import.
- Imported documents: entityType checked against the enum, MIME type
allowlisted and matched against the file signature, path kept under the
storage root (no more directory traversal via entityType).
- Permission cache is cleared after an import.
Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
SSO picked the local account by the email/preferred_username claim, which
Entra does not verify. Anyone able to present a matching email (another
tenant with a multi-tenant issuer, or an editable attribute) could sign in
as that user without a password.
- The first SSO login links the account to "<iss>|<oid>" (new unique
column User.ssoSubject); afterwards only that identity signs in, and a
matching email from another identity is refused.
- Multi-tenant issuers (common/organizations/consumers) disable SSO.
- Emails are compared case-insensitively and stored lower-case, so
[email protected] and [email protected] can no longer be two accounts; the migration
lower-cases existing emails where that does not collide.
- An admin changing a user's email clears the link.
Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
A role with users:update could reset an administrator's password or MFA,
change the admin's email to one it controls, or give a colluding account
the ADMIN role (with users:assign-role) - and so become administrator.
Every management action (update, deactivate, activate, reset password,
reset MFA) and every role assignment now requires the target role's grants
to be a subset of the caller's own, like role editing already did. The
target role is read from its grants, so a deactivated admin still counts
as an admin. Changing another user's email revokes their sessions.
Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
All JWTs (access, MFA setup, MFA challenge, SSO completion) were signed
with the same secret and the access-token strategy checked only signature
and expiry. The MFA challenge token returned after the password step
therefore worked as a full access token: a password alone was enough to
use the API, skipping the second factor.
Every token now carries a pinned issuer, algorithm (HS256) and a
per-purpose audience; the strategy accepts only the access audience and
rejects any payload carrying a helper-token purpose. Existing access
tokens (no audience) stop working; clients renew them via the refresh
cookie.
Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
- WinterCheck per colony, winter and season (autumn/spring), every field
editable; loss causes limited to the COLOSS monitoring categories.
- Summary per winter and apiary: loss rate lost/wintered, marked
provisional while colonies are not yet checked in spring - an unchecked
colony never counts as survived.
- A loss marks the colony deceased only on explicit request, in the same
transaction and audited.
- Permissions winter-checks:read/create/update granted by migration to
beekeeper and staff roles; winter checks included in backups.
- Page "Überwinterung": next action first, one button per row until the
colony is wintered, German number format.
Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
- Treatment records supplier, purchase proof and prescribing vet, and
keeps the apiary the colony stood at on the treatment day.
- Holder name and address in site settings (never in public output;
regression test in public.service.spec).
- treatment-log.pdf: one sheet per apiary, identical treatments grouped,
missing mandatory fields printed as FEHLT, uncertain locations marked.
- Shared PDF table helpers extracted from the record card.
Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
New report colony-record-card (button "Stockkarte (PDF)" on the colony
page): master data, current queen and queen history, every inspection as
a table row (strength, eggs, open/capped brood, queen seen, queen cells,
gentleness, swarm drive, feed, measures, notes), treatments with batch
number and withdrawal period, feedings, varroa counts and harvests.
Landscape A4, table header repeated on each page. Whole history by
default, one calendar year with ?year=.
Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
Two new rules, "Kontrolle nach Behandlung" and "Kontrolle nach
Fütterung", create a check task when a treatment or feeding is recorded.
The due date counts from the measure itself (end of treatment, else its
date; feeding date) plus the configured number of days, and the task
names the product and date. At most one open check per colony and rule.
Both rules ship switched off and without an interval: the right timing
depends on the product (package leaflet) and season, so the beekeeper
sets it. Every task-creating rule can now carry a "due after N days"
setting in the automation page.
Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
Label sheets printed on plain A4 are cut with a lever paper cutter. Each
template can now print small hairline cut marks in the sheet margin: one
at both ends of every cut line (left/right edge of each column, top/bottom
edge of each row), 1 mm away from the label and never closer than 2 mm to
the paper edge, so they print on normal printers and never touch a label.
Labels without a gap share one mark. Marks are drawn on every sheet for
the full grid, since the cutter cuts across the whole sheet.
On by default (new column cropMarks, default true); switch it off in the
template designer for pre-die-cut label sheets.
Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
Integrations such as the Wabenhain shop sync no longer need a service
account with a password (and without MFA). An admin creates a key under
Verwaltung > Automatisierung > API-Schlüssel.
- Key format ack_<8>_<43>; only its SHA-256 hash is stored, the key is
shown exactly once. Optional expiry, "last used" timestamp, revocation
takes effect immediately; create/revoke are audit-logged without the key.
- A key is not a user: JwtAuthGuard sets request.apiKey, PermissionsGuard
admits it only to routes whose permission is among its scopes and never
to user-only routes. Scopes come from a short allowlist (shop-sync:read).
- Managing keys requires the new permission api-keys:manage.
- Keys are excluded from backups on purpose; a restore means issuing a new
key.
Additive migration (new table).
Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
GET /api/v1/shop-sync returns exactly the data ApiCore releases to the
shop: harvest facts, article master data and bottlings with their
harvest ids. Every query uses an explicit select, so notes, colonies,
treatments and fields added to the models later never leave ApiCore
unless deliberately added to the allowlist.
- New permission shop-sync:read (granted to ADMIN automatically, to no
other role): the shop's service account needs nothing else, in
particular no access to harvest or extraction records with notes.
- contractVersion 1; only breaking changes bump it.
- Always the full set: small volume, and a full comparison heals missed
runs by itself.
Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
Start page (AuDHD-oriented, per the agreed "Heute zuerst" layout):
- One highlighted next step: started work first (easy re-entry after an
interruption), then overdue, due today, then urgent undated tasks.
Primary action "Erledigt", secondary "Beginnen", link to the related
colony/apiary/queen.
- "Außerdem heute": at most five further tasks with a large honey-cell
checkbox; undated automatic suggestions are included so a queenless
colony does not sit unseen in the full list.
- Completing is the one authored motion moment: the cell fills, the tick
draws, the row folds away and the next step arrives with a short focus
pull. Every completion can be undone from the toast (10 s), restoring
the previous status.
- Status is shown as icon plus word (never colour alone); tasks created by
automation carry a "Vorschlag" badge, recurring and inspection tasks
their own - a suggestion never looks like work already done.
- Friendly empty, loading and error states; key figures moved into a
compact overview below; overdue tasks are no longer listed twice.
Motion: per-device setting Voll / Dezent / Aus in the profile (default
Voll); the OS reduce-motion setting always lowers Voll to Dezent.
New 44 px "touch" button size for gloves at the apiary; themed text
selection and caret.
Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
When the server saved an inspection, harvest, treatment, feeding, varroa
measurement or swarm check but the answer was lost (dead zone at the
apiary), the form queued the entry and the sync later created it a second
time - including a second automation task. Each opened form now creates
one Idempotency-Key, sends it with the request and stores it with the
queued entry, so the retry returns the first record. Entries queued before
this change still sync without a key.
Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
Data-integrity fixes found in a systematic review of ApiCore:
- Bottling booked the honey batch amount from a value read outside the
transaction and wrote an absolute result back, so two concurrent
bottlings (double tap, two devices) could both pass the check and one
deduction was lost. Bookings now use a conditional UPDATE inside the
transaction; corrections and deletions give amounts back atomically.
- Outgoing stock movements could drive stock below zero; the removal is
now a conditional decrement that refuses when stock is insufficient.
- Weak, conspicuous and queenless colonies were treated as "not active"
and could not be moved, split or merged - uniting a queenless colony
is the textbook merge. Only terminal statuses lock a colony now; merging
a colony into itself and dissolving twice are rejected.
- Moving a queen to another colony failed on the unique currentQueenId
because the old colony kept pointing at her. The old colony now releases
her (and becomes QUEENLESS); a queenless colony receiving a queen becomes
ACTIVE; dissolved colonies cannot receive a queen.
- Completing a task twice moved completedAt and re-sent the webhook;
reopening a task kept completedAt. Recurring templates created tasks
without a due date and caught up missed weeks one stale task at a time.
- Automation rules created a new task and warning on every trigger (every
stock removal below minimum). Tasks now record their source trigger and
at most one open task per rule and object exists. The varroa threshold
(mites per day) is only compared with natural-mite-fall rates, not with
sample-based counts in another unit.
- New opt-in Idempotency-Key support for record-creating endpoints, so
the offline queue can retry after a lost response without duplicates.
- Permission catalog sync crashed replicas booting concurrently on a
fresh database (unique violation); it now uses createMany/skipDuplicates.
- Server-side validation for inspection rating scales (1-5), unique
colony/harvest lists and net weight <= gross weight.
- Stale e2e expectations updated (permission count, trigger list, real PNG
fixture for the file-signature check).
Migration is additive only (nullable Task column, new table).
Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
The backend accepted every field of these records already; the forms simply
did not offer them, so from the app they were still frozen.
- Inventar: only the condition could be changed. A mistyped inventory number,
a wrong purchase or maintenance date, a renamed item - all stuck for the life
of the record. New edit dialog covering every field; the inline condition
select stays as a shortcut.
- Produkt: the status was shown as a badge but could not be set anywhere, so a
product could never be discontinued or brought back.
- Aufgabe: the status could only ever move to DONE through the "Erledigt"
button. IN_PROGRESS and CANCELLED were unreachable, and an accidentally
completed task could not be reopened.
- Aufgabenvorlage: priority, assignee and the active flag were missing.
Deactivating was one-way, since only a deactivate endpoint was wired up -
the active checkbox now reactivates through PATCH.
- Königinnen-Bewertung: notes and the evaluated colony were missing.
Verified by extracting the field set each form submits and comparing it with
the matching update DTO: every form now covers its DTO. The one gap left is
re-linking a document to a different record, which still has no UI.
Co-Authored-By: Claude Opus 5 <[email protected]>
renderLabelsPdf resolved the jarSize placeholder to "500 ml". Everywhere else
in the app jarSize is a fill weight in grams - the bottling forms label it
"Glasgröße (g)", the batch list and the public Herkunft page print "g", and
BottlingService divides by 1000 to get kilograms. Only the printed label said
millilitres, so every label produced so far declares the wrong net quantity for
a food sold by mass.
The placeholder mapping moved into an exported buildPlaceholderValues() so the
wording can be asserted directly. It could not be before: the render output is
a PDF buffer whose text a test cannot read, which is how the wrong unit shipped
unnoticed. Three tests now cover it, including one that rejects any volume unit
outright.
Labels already printed with "500 ml" have to be reprinted - the code change
does not correct paper.
Co-Authored-By: Claude Opus 5 <[email protected]>
test/jest-e2e.json listed the ESM-only packages ts-jest must still transform,
but @nestjs/schedule (12.0.1, "type": "module") was not among them. Since
AppModule imports ScheduleModule for the backup scheduler, every e2e spec died
on "SyntaxError: Unexpected token 'export'" while loading AppModule - the whole
suite, not one test.
That silently disabled test/authorization-coverage.e2e-spec.ts, the check that
asserts every non-public route carries @RequirePermission or
@AnyAuthenticatedUser. It is exactly the check that would have caught the
displaced decorators fixed in the previous commit, and it had been dead since
the scheduler shipped.
With @nestjs/schedule and its cron dependencies added to the allowlist, the
authorization-coverage spec compiles and passes, confirming no route is left
undecorated. The remaining 36 e2e specs still need a Postgres instance on the
port .env.test points at; they fail on connection, not on code.
Co-Authored-By: Claude Opus 5 <[email protected]>
Inserting the new PATCH handlers directly above an existing route put them
between that route's @RequirePermission decorator and its method, so the
decorator bound to the new handler instead:
- DELETE /documents/:id lost documents:delete
- POST /honey-batches/:id/measurements lost honey-batches:measure
- POST /queens/:id/evaluations lost queens:evaluate
and each new PATCH ended up carrying two @RequirePermission decorators, of
which only one takes effect.
This did not open anything up. PermissionsGuard fails closed by decorator
absence: a non-public route with neither @RequirePermission nor
@AnyAuthenticatedUser is denied outright, so the three routes would have
returned 403 for every caller including an administrator. The effect was a
broken endpoint, not a bypass.
Each decorator is back on the route it belongs to, and the duplicates are
gone. test/authorization-coverage.e2e-spec.ts asserts exactly this contract
across every route and is the check that should run before a route change
ships.
Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
Both could only be created and deleted. For a webhook that is actively harmful:
recreating it issues a new signing secret, so the receiver that verifies the
X-ApiCore-Signature header against the old one starts rejecting every delivery
without anyone changing it on purpose.
Backend:
- New PATCH /webhooks/:id behind a new webhooks:update permission, covering the
target URL, the event selection and the active flag - the last of which had
no way to be set after creation at all, so a misbehaving receiver could only
be deleted. The signing secret is deliberately left alone and stays out of
the audit log.
- A changed webhook URL runs through the same assertPublicWebhookUrl SSRF check
as a new one; without that the edit would have been a way around the guard
create() applies. Covered by a test.
- New PATCH /documents/:id behind a new documents:update permission, for
re-linking a document filed against the wrong record and correcting its
display name. Re-linking validates the new target with the existing
assertEntityExists; the stored file is untouched.
Frontend:
- New webhook edit dialog. The event labels and the toggle helper moved out of
the create dialog into lib/webhooks/labels.ts so both use one copy.
- Documents can be renamed from the attachment card.
Re-linking a document to a different record has no UI yet: it needs a two-step
picker (colony, then inspection) that is not worth building unprompted.
Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
Both records could be created but never changed. A water-content reading
decides whether a batch may be sold as honey at all, and queen ratings drive
breeding decisions - a slipped digit in either is exactly the kind of wrong
value that must not be left standing.
- New PATCH /honey-batches/measurements/:measurementId behind
honey-batches:measure, and PATCH /queens/evaluations/:evaluationId behind
queens:evaluate. Both are addressed by the child record's own id, so the
routes take two path segments and do not collide with PATCH /:id.
- Both log before and after data, like every other correction.
- The measurement list on the honey batch page and the rating list on the queen
page gained an edit dialog; the rating dialog reuses the existing criteria
list rather than restating it.
Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
Four fields were still frozen because changing them would have bypassed the
record that makes the change traceable, and a user's login data could not be
corrected at all. Locking a field is the wrong trade: a wrong value stays in
the system forever. The fix is to let the edit write the same history entry the
dedicated endpoint writes.
- A colony's apiary is editable through PATCH and writes the same
ColonyApiaryHistory entry POST /colonies/:id/move writes, so the location
history never skips a stand the colony was recorded on.
- A honey batch's storage container is editable through PATCH and writes the
same HoneyBatchRelocation entry POST /honey-batches/:id/relocate writes.
- A stock item's quantity is editable and books the difference as an ADJUSTMENT
movement, so the stored level stays the sum of its movements.
- A role's key is editable (system roles were already rejected outright by
RolesService.update, which keeps the seed's ADMIN lookup intact).
- A user's e-mail - the login identity - is editable. A typo there previously
locked the account out for good.
- New POST /users/:id/reset-password lets an administrator set someone else's
password: there is no self-service reset flow, so a forgotten password was
the one thing no edit could repair. It revokes every refresh token and clears
the permission cache exactly as a self-service change does, refuses to act on
your own account (that goes through /profile/password, which proves ownership
by asking for the current one), and never writes the password to the log.
Frontend:
- Users had no edit UI at all, although PATCH /users/:id existed. New
EditUserDialog covers name, e-mail, role and an optional new password.
- Colony, honey batch and stock item forms gained the newly editable fields,
each with a note saying the correction is recorded as a move, a relocation
or a stock adjustment.
- The role dialog gained the key field.
Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
An inspection, treatment, feeding, varroa measurement or swarm control check
filed against the wrong colony could not be repaired: the colony was frozen at
creation time, and with no delete endpoint either, the only remedy was leaving
a wrong record in the colony's history. The same held for a harvest booked
against the wrong apiary or with a colony forgotten, for an extraction linked
to the wrong harvests, and for a honey batch whose number or source extraction
was mistyped.
Backend:
- The five colony-scoped update DTOs now cover the full create DTO, colonyId
included.
- Harvest update accepts apiaryId and colonyIds, extraction update accepts
harvestIds. Both validate the ids first and replace the whole join-row set
inside the existing transaction; the join rows carry no data of their own,
so replacing beats diffing.
- Honey batch update accepts batchNumber, extractionId and status. Status was
previously only ever set to BOTTLED by the bottling module, so SOLD and
CONSUMED were unreachable through the API.
- Colony and storage container codes, and a task's relatedType/relatedId, are
editable.
- apiaryId on a colony and storageContainerId on a honey batch stay out of
PATCH on purpose: both have dedicated endpoints that also write the history
entry (ColonyApiaryHistory, HoneyBatchRelocation) the traceability chain
depends on.
- AllExceptionsFilter now also translates Prisma P2003 to a 400 and P2025 to a
404. Services write straight against the foreign keys with no application
level existence check, so before this a stale or mistyped relation id came
back as a bare 500 and filled the log with fake server errors.
Frontend:
- New shared fields: ColonySelectField (used by all five colony-scoped edit
dialogs), HarvestSelectField (both extraction dialogs) and TaskRelatedField.
- Re-filing a record invalidates the lists of both the colony it left and the
one it joined, otherwise one of the two stays stale.
- The harvest dialog gained an apiary select and a colony checklist that keeps
already-linked colonies visible even when they moved or were dissolved; the
honey batch dialog gained batch number, source extraction and status.
- Extraction and harvest forms refuse to submit an empty relation set instead
of letting the backend reject it.
Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
A bottling batch could only be corrected in four of its seven fields after
creation: honey batch, jar size and jar count were frozen at creation time,
an assigned product could not be cleared again, and there was no way to
remove a record entered by mistake at all. A mistyped jar count is a
data-entry error, not a fact of production, so it has to stay correctable.
Backend:
- UpdateBottlingBatchDto now covers every field. `productId: null`
explicitly clears the product assignment; omitting it leaves it alone.
- BottlingService.update rebalances the affected honey batch amounts:
the previously bottled amount is given back before the corrected amount
is taken again, so a correction never double-charges the source batch.
Moving a batch to a different honey batch credits the old one and debits
the new one in the same transaction.
- New DELETE /bottling-batches/:id gives the bottled amount back to the
honey batch and keeps the full record in the audit log (action DELETE,
beforeData). The honey batch status is deliberately left untouched - the
status it held before being bottled is not recoverable here.
- New permission bottling:delete, checked server-side. The catalog sync
grants it to system roles on the next boot.
Frontend:
- The edit dialog exposes all fields, offers "Kein Produkt" to clear the
assignment, keeps a discontinued product selectable so opening the dialog
cannot silently drop it, and shows backend errors inline.
- The form is mounted only while the dialog is open, so a cancelled edit
never leaks its values into the next one.
- New delete button behind a confirmation that names the amount credited
back and warns that the public origin page for the lot number stops
resolving.
- Edit and delete are additionally hidden without the matching permission.
Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
No code anywhere caught Prisma's P2002 (unique constraint violation).
Colony/apiary/queen codes, product SKUs, batch numbers, and equipment
inventory numbers are all @unique DB columns with no application-level
duplicate check ahead of the insert, relying on the constraint itself
- so a beekeeper reusing a code by mistake got a bare 500 "Internal
server error" with zero indication what went wrong, and the incident
polluted server logs as a fake error rather than a routine 4xx.
AllExceptionsFilter now recognizes P2002, returns 409 with a specific
German message naming the conflicting field when it's a single known
column (falling back to a generic duplicate message otherwise), and
logs it at warn level like any other client-side rejection.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Four more mutations had no error feedback: equipment condition change,
task completion, and notification mark-as-read all failed silently.
More importantly, the offline sync badge computed a result.failed
count from syncPendingInspections() but never displayed it - an entry
rejected by the server (e.g. a validation rule) would sit in the
offline queue indefinitely with zero indication to the beekeeper why
it never syncs. Both the failed-count and outright sync-mutation
rejection now surface a toast.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Backend business-rule guards (inactive apiary blocking colony
create/move, active colonies blocking apiary deactivation, negative
quantity on IN/OUT stock movements, unavailable queen blocking colony
assignment, inactive colony blocking split/merge/move) threw English,
UUID-laden exception text that the frontend either discarded in favor
of a generic message or - in colony-actions.tsx's move/split/merge/
assign-queen dialogs - didn't handle at all, so the guard fired with
zero visible feedback. Translated the messages to plain German and
wired the affected mutations to display them, so a beekeeper sees the
actual reason a save was rejected instead of silence or "failed".
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Two unguarded async calls silently swallowed errors: document download
used a bare "void downloadDocument(...)" with no catch, so a failed
download (network error, expired session) did nothing visible; the
webhook secret copy button had no try/catch around
navigator.clipboard.writeText, which can throw on denied permission or
an insecure context. Both now show a toast on failure.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Export mutations (annual report PDFs, colony/honey-batch CSV exports)
had no onError handler, so a failed download (network error, expired
session) silently did nothing with zero user feedback. Same silent-
failure class as the earlier onError sweep on entity-save mutations,
here applied to the 5 remaining export/download call sites.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Nearly every *-edit-dialog.tsx and *-edit-card.tsx in the app - the
save mutation for editing a colony, queen, treatment, inspection,
feeding, varroa measurement, swarm control check, harvest, extraction
(both variants), honey batch, bottling batch, task, stock item,
storage container, or product - had an onSuccess handler but no
onError. A failed save (a validation error, including several added
earlier this session, a permission error, a network hiccup) produced
no feedback at all: no toast, no inline message, the dialog just sat
there having silently done nothing. The user's most likely read on
that is "it saved," when it didn't.
Added the same onError: () => toast.error("<Entity> konnte nicht
gespeichert werden") pattern already used elsewhere in the app (e.g.
apiary deactivate/reactivate) to every one of these mutations. Also
covers assign-colony-dialog.tsx, which is the exact spot where the
queen-availability guard added earlier this session would otherwise
have failed with zero visible feedback.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
The previous two commits added @Min(0)/@Max(100) validation and a
sign-inversion guard on the backend, but the number inputs themselves
had no min/max attributes - a user could still type a negative weight,
quantity, or an out-of-range percentage and only find out via a
generic "failed" toast after submitting, with no indication of why.
Added min={0} (and max={100} on water-content percentages) to every
input for a field the backend now bounds: harvest weights/super
count, honey-batch amount/water content, honey measurements, feeding
amount, extraction amount/water content, stock item quantity/min
threshold, and varroa mite count/duration/calculated value - across
both the create and edit form for each.
The stock-movement quantity input is the one exception requiring
care: min is only applied when type !== "ADJUSTMENT", matching the
backend guard, since a correction movement must still be allowed to
go negative. Added a short explanatory line under that field too.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
None of these numeric business fields had a lower bound, so a
negative honey weight, jar count, feeding amount, mite count, or
stock quantity was silently accepted - each would corrupt whatever
total or statistic it later feeds into. Added @Min(0):
- Harvest: honeySuperCount, grossWeight, netWeight
- HoneyBatch: amount (+ @Max(100) on waterContent, a percentage)
- HoneyMeasurement: waterContent (+ @Max(100))
- Feeding: amount
- Extraction: amount (+ @Max(100) on waterContent)
- StockItem: quantity, minThreshold
- VarroaMeasurement: durationMinutes, miteCount, calculatedValue
Left temperature fields alone (extraction, inspection) - genuinely
can be negative (winter inspections, outdoor storage) and no
confident domain bound exists to apply instead. Update DTOs for these
entities all extend PartialType(PickType(Create...Dto, [...])) from
@nestjs/swagger, which preserves validation decorators, so no
separate change was needed there.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
addMovement() negates quantity internally for OUT (delta = -quantity)
and applies it directly for IN. Nothing stopped a negative quantity
from being submitted, so e.g. { type: 'OUT', quantity: -5 } would
compute delta = -(-5) = +5 - silently ADDING stock while being
recorded in the audit trail as a removal. The inverse held for IN.
ADJUSTMENT is the one type that must stay allowed to be negative,
since that's the only way to correct a stock count downward, so this
is a service-level guard (BadRequestException) rather than a DTO
validator - class-validator's @ValidateIf skips every validator on a
property when its condition is false, including @IsNumber, so it
can't express "always a number, but only sometimes positive" cleanly;
verified this empirically before choosing the service-level approach.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
No error.tsx existed anywhere in the app - an uncaught render error on
any page (a beekeeper's most field-critical moment: mid-inspection,
no signal, nowhere to go) showed nothing recoverable, not even a
message, let alone a way back to a working page.
Two boundaries, matching Next.js App Router's error-boundary scoping:
- (app)/error.tsx covers every authenticated page. It only replaces
the crashed page's content - the sidebar/header/bottom nav stay
mounted and usable, so the user can navigate to a different page
without a full reload.
- app/error.tsx is the outer fallback (login, the public Herkunft
page, or a crash in (app)/layout.tsx itself, which only a strictly
higher boundary can catch). Its "Zum Dashboard" action is a plain
<a> reload rather than client-side navigation, since a crash this
high may have left client state corrupted.
Both are in German, matching the app's existing card/honeycomb visual
language, and explicitly reassure the user that already-saved data is
unaffected.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
The colony annual PDF report's treatment section only ever printed
date/product/amount - activeIngredient, batchNumber, and
withdrawalPeriodDays (added earlier this session specifically for
Wartezeit compliance) existed on the Treatment record but never made
it into the one document a beekeeper would actually hand to a
Veterinäramt inspector or keep as their Bestandsbuch-style compliance
record. Silently missing the withdrawal period from the printed
record undermines the whole point of tracking it.
Also adds the first tests for annual-reports.ts (previously untested).
Co-Authored-By: Claude Sonnet 5 <[email protected]>
The service worker only ever cached same-origin static assets/pages -
every real data call (colonies, apiaries, tasks, honey batches, ...)
was excluded, both because it targets a different origin (API_URL
points at a separate backend service, per .env.example) and because
of an explicit /api path exclusion. In practice this meant a
beekeeper opening the app at an apiary with no signal saw nothing at
all: not even data fetched minutes earlier on the same device, despite
the offline write-queue built earlier assuming the app itself stays
usable while offline.
The backend can't be matched by origin the way static assets are (the
service worker has no static knowledge of the runtime-configured
API_URL), so it's matched by path instead - every backend call goes
through /api/v1/. /api/runtime-config is explicitly excluded and
always fetched live, since a stale cached value would point the app
at the wrong backend. Error responses (4xx/5xx) are now never cached
either, closing a latent bug that could otherwise serve a stale 401
back after a token refresh.
Since the cache now holds real business data, it's cleared on logout
so it can't leak into the next person's session on a shared device.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
The colony-apiary and queen-availability validation just added
backend-side had no matching frontend guard, so a user could still
pick an inactive apiary or open the assign-colony dialog on an
unavailable queen and only find out it fails after submitting:
- Colony create/move apiary pickers now fetch only ACTIVE apiaries.
- The queen "Volk zuweisen" button is disabled once her status is
LOST/REPLACED/SOLD/DECEASED (QUEEN_UNAVAILABLE_STATUSES, mirroring
the backend's QUEEN_NO_LONGER_AVAILABLE_STATUSES), instead of only
failing after the form is filled out and submitted.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
assignColony() never checked the queen's own status, so a queen marked
LOST/DECEASED/REPLACED/SOLD - exactly the statuses the queen-loss fix
now treats as "not really available" and releases her from her old
colony for - could still be assigned to a brand new colony, which
makes no domain sense (you can't put a dead queen in a hive).
Rejects with ConflictException up front, before even looking up the
destination colony; correcting a mistaken status update first is the
intended fix path.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Colony.apiaryId was never checked against the target apiary at all -
create() would hit a raw Prisma FK-constraint error for a nonexistent
apiary instead of a clean 404, and both create() and move() would
silently succeed against a deactivated apiary, orphaning the colony
the same way (just from the opposite direction) as the apiary-side gap
fixed in the previous commit.
Adds a shared assertApiaryActive() check, following the same
findUnique-then-status pattern already used for cross-entity checks
elsewhere (e.g. QueensService.assertMotherQueenExists), used by both
create() and move().
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Colony.apiaryId is a plain FK, never checked against the apiary's own
status. Without a guard, deactivating an apiary silently orphaned any
colonies still living there: they kept showing as ACTIVE, but their
apiary had disappeared from every "active apiaries" list, dashboard
count, and colony-creation picker - a confusing half-state with no
warning at all.
ApiariesService.deactivate() now throws ConflictException when active
colonies remain, so the beekeeper has to move or dissolve them first -
matching the same assertActive-style guard pattern already used
elsewhere (colonies, e.g. move()/merge()/split()).
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Colony.currentQueenId is unique across colonies. dissolve() and merge()
set status to DISSOLVED/MERGED but never cleared currentQueenId, so a
colony that still had a queen when it was dissolved or merged kept
permanently "owning" her - any later attempt to assign that queen to a
different colony (e.g. requeening another colony with her) would fail
with a unique-constraint conflict, since the dead colony's stale
reference was still occupying that slot.
Both operations now clear currentQueenId and close the open
QueenColonyHistory row for that queen, mirroring the bookkeeping
assignColony() and the queen-status-transition fix already do
elsewhere. The merge survivor keeps its own queen; a merged-away
colony's queen is released, not auto-transferred, since which queen
"wins" a merge is a beekeeper decision.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
A very common real-world event - a queen dying over winter, absconding,
being superseded, sold, or replaced - had no side effect on the colony
record at all. Colony.currentQueenId kept pointing at a queen who was
no longer really there, colony.status stayed whatever it was, and the
QUEENLESS dashboard alert could never fire from this path.
QueensService.update() now detects a transition into
LOST/DECEASED/REPLACED/SOLD and, inside the same transaction, clears
currentQueenId on whichever colony currently points at that queen,
flips its status to QUEENLESS (unless the colony is already in a
terminal lifecycle status like DISSOLVED, which is left alone), and
closes the open QueenColonyHistory row - mirroring the bookkeeping
assignColony() already does when a queen changes colonies.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
formatGeneratedNumber() passed the admin-supplied {SEQ:N} width straight
into String.padStart() with no upper bound. A careless or malicious
format value (e.g. {SEQ:999999999}) would try to allocate a
correspondingly huge string - on the backend that runs on every single
honey-batch/bottling creation once such a format is saved, and in the
newly added frontend preview it would run on every keystroke while
typing the format. Caps the effective width at 20 digits, far beyond
any realistic sequence number, in both the backend implementation and
its frontend port so they stay identical.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Admins configuring the honey-batch/bottling auto-numbering format
(WH-{YYYY}-{SEQ:4}) had no way to see what it would actually produce
before saving - a typo'd placeholder or unexpected format silently
breaks every batch number from then on. When the rule is active, the
format card now renders the next number it would generate, live as
the format is edited, using the same formatGeneratedNumber logic the
backend uses (ported to the frontend, no shared package between the
two apps) and the real next sequence value (sequenceValue + 1).
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Audit requested by the user after finding queens couldn't set
hatchYear: many edit dialogs (and some create forms) were missing
fields the backend DTOs already accept, so that data could never be
entered or corrected anywhere in the UI. Fixes per entity:
- Colony: status (WEAK/QUEENLESS/CONSPICUOUS) was never settable
anywhere, so the COLONY_ATTENTION dashboard alert could never fire.
Added as a dropdown restricted to the non-terminal statuses;
DISSOLVED/SOLD/DECEASED/MERGED stay on their dedicated confirm
actions, which also set deactivatedAt.
- Apiary: description was missing from the edit form.
- Inspection: 10 fields (weather, temperature, broodAmount,
broodPattern, openBroodPresent, cappedBroodPresent,
queenCellsPresent, swarmMood, gentleness, combBehavior,
honeySuperStatus) existed in the schema/DTOs but nowhere in the UI -
added to both the create and edit forms.
- Varroa measurement: durationMinutes was on the create form but
missing from edit.
- Feeding: batchNumber missing from both create and edit.
- Treatment: startDate, endDate, activeIngredient, batchNumber
missing from both create and edit.
- Task: assignedUserId had no UI anywhere; added to create/edit,
degrading gracefully if the current user's role lacks users:read.
- Extraction (honey-batch traceability variant): amount was missing,
while the other edit dialog for the same entity already had it.
- Bottling: productId missing from edit.
- Harvest: grossWeight missing from edit; grossWeight and
honeySuperCount missing from create.
- Queen: breeder, matingType, matingStatus, inseminationData,
motherLine, fatherLine added to the edit form (hatchYear/
markingColor/markingNumber were already fixed in a prior commit).
Left alone deliberately: fields the backend DTO itself excludes via
PickType (e.g. HoneyBatch's jarSize/jarCount/totalAmount, Bottling's
quantities) since mutating those post-hoc would desync inventory
deduction, and Colony's terminal statuses/code which have dedicated
lifecycle actions.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
VarroaMeasurement.calculatedValue was already used as "Milben/Tag" by
the VARROA_THRESHOLD automation (thresholdMitesPerDay) and the annual
PDF report, but no client ever actually computed it - the frontend
form only collects raw miteCount and durationMinutes. That meant the
threshold check was comparing a raw, un-normalized mite count against
a per-day threshold: e.g. 12 mites counted over 5 days (2.4/day,
healthy) would incorrectly trigger the same alert as 12 mites counted
over 1 day (a real problem).
VarroaService now derives calculatedValue = miteCount / duration in
days for NATURAL_MITE_FALL measurements, on create and (only when no
value exists yet) on update, without ever overwriting an explicitly
provided or already-stored value. Other methods (Puderzucker,
Waschmethode, Drohnenbrutkontrolle) use a different, non-time-based
metric and are left for manual entry rather than guessing an
unverified formula.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
The backend has always accepted hatchYear/markingColor/markingNumber
(Queen DTOs, Prisma schema), and the queens list even already renders
a "Schlupfjahr" field - but neither the create nor the edit form ever
exposed inputs for them, so no user could actually set hatchYear
through the UI. That silently broke the OLD_QUEEN dashboard alert
just added: it can only ever fire for colonies whose queen has a
hatchYear, and there was no way to record one.
Both forms now include Schlupfjahr, Zeichenfarbe and Zeichennummer.
Marking color is auto-suggested from the hatch year using the
internationally standardized queen marking color convention (last
digit of the year -> white/yellow/red/green/blue), editable like any
other suggestion.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Queen.hatchYear was already captured but never used anywhere - no
signal existed for "this colony's queen is overdue for evaluation or
replacement," despite that being one of the most common productivity
and swarming risk factors in beekeeping. Adds an OLD_QUEEN dashboard
alert alongside the existing overdue-task/swarm-check ones, flagging
any active colony whose current queen has completed 3+ seasons.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Treatments (Ameisensäure, Oxalsäure, etc.) can require a legally
mandated withdrawal period before honey from that colony may be sold.
Nothing in the app tracked this before, so a beekeeper had no
safeguard against harvesting into an active withdrawal window.
Treatments now optionally record withdrawalPeriodDays (from the
product's Packungsbeilage). The new harvest form fetches each
candidate colony's active withdrawal window and shows an inline
warning naming the product and the date it clears, computed as
max(endDate ?? date) + withdrawalPeriodDays across a colony's
treatments, restricted to windows still in the future.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Schwarmkontrolle already captures nextCheckDate per check, but it was
only ever visible by opening that one colony's tab. During swarm
season a missed follow-up check can mean losing half the colony, so
this now surfaces as a dashboard alert (OVERDUE_SWARM_CHECK) like
overdue tasks already do - only the most recent check per colony is
considered, so a completed follow-up clears the warning.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
iOS Safari ignores the web manifest for home-screen icons entirely and
needs its own apple-touch-icon link, which Next.js only emits when
app/apple-icon.tsx (or a literal apple-icon.png) exists. Without it,
iOS falls back to a screenshot of the open page instead of the brand
mark. Mirrors the existing public/icon.svg design via next/og.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
TaskTemplate had zero seed data - a fresh deployment showed a
completely empty recurring-task list with no hint of what to even put
there. Seeds 12 YEARLY templates, one per month, covering the standard
German-beekeeping annual cycle (Frühjahrsdurchsicht, Schwarmkontrolle,
Einfütterung, Varroa-Restentmilbung/Oxalsäure, Wintersitz-Kontrolle,
etc.) so a new beekeeper gets a ready-made starting point instead of a
blank page. Runs once (guarded by a taskTemplate.count() check, same
idempotency shape as the existing admin-user seed) and only ever adds
rows - admins can edit or deactivate any of them afterward.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Temperature and rain were already tracked, but wind matters just as
much for whether it's safe/sensible to open a hive - bees get more
defensive and frames are harder to handle in wind. Adds
wind_speed_10m_max to the Open-Meteo request and a derived
inspectionSuitable flag per day (>=15C, <=1mm rain, <=25km/h wind;
null when a value is missing rather than guessing). The apiary weather
card now shows a clear "good/not good for a Durchsicht today" banner
plus wind alongside each day's forecast.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Only Durchsicht had an offline fallback (queue to IndexedDB, sync on
reconnect); Behandlung, Fütterung, Varroamessung, Schwarmkontrolle and
Ernte just failed outright with no signal - exactly the situation a
beekeeper is in at the actual hive. Generalizes lib/offline/db.ts and
sync.ts to six queues (version-bumped IndexedDB schema, existing queued
inspections preserved), wires the same "server rejected it vs. no
network" fallback into all five remaining */new pages, and updates the
sync badge to count/drain all six.
Also: doc-hygiene pass (ARCHITECTURE.md/TASKS.md) to replace the stale
pg_dump/CronJob backup plan with the in-app BackupModule that actually
shipped earlier this session.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
The dashed rectangle drawn around every label slot (a cut/crop line)
wasn't needed and is gone, along with the width/height params that
existed only to draw it. Dates on the label (bottling date, MHD) now
use { day: '2-digit', month: '2-digit' } so single-digit values print
with a leading zero (05.03.2026, not 5.3.2026) - the default 'numeric'
formatting was dropping it.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
A batch that doesn't fit in one container (e.g. 125 kg across 5 x 25 kg
buckets) had no way to be represented - HoneyBatch.storageContainerId
is a single nullable FK. Adds HoneyBatchContainerAllocation as an
additive join table (container + amount per batch), independent of the
existing single-container/relocate() flow which still covers the
simple one-container case unchanged.
- GET/PUT/DELETE /honey-batches/:id/containers[/:containerId], reusing
the existing honey-batches:relocate permission.
- Assigning a container is an upsert (@@unique on batch+container), and
the total assigned is validated against the batch's own amount when
known.
- New "Aufteilung auf mehrere Behälter" card on the honey batch detail
page, alongside (not replacing) the existing storage card.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
The sidebar nav sent both entries straight to their "new" creation
form, with no list page to ever browse what had been created -
Schleuderung creation didn't even redirect to a detail page, just to
/colonies, so a saved extraction left no trace anywhere in the UI.
- New /harvests and /extractions list pages (mirrors the existing
products/apiaries list pattern).
- New /extractions/[id] detail page + edit dialog, matching the
harvests one that already existed but had nothing linking to it.
- Sidebar "Ernten"/"Schleuderungen" now point at the list pages instead
of the creation forms.
- Extraction creation now redirects to its detail page instead of
/colonies.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
One Entra enterprise application now covers both interactive login and
the calendar sync's client-credentials Graph access, instead of
requiring a second app registration. GraphClientService reads
OIDC_ISSUER_URL/OIDC_CLIENT_ID/OIDC_CLIENT_SECRET (same as SSO) and
derives the tenant ID from the issuer URL's path, rather than three
dedicated M365_CALENDAR_* env vars. The app registration just needs the
"Calendars.ReadWrite" Application permission added alongside its
existing delegated login scopes.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Adds a one-way push (ApiCore -> M365) of Aufgaben/Durchsichten/
Behandlungen/Fütterungen/Ernten/Königinnen-Zuordnungen into the calendar
of a Microsoft 365 group, via the Graph API.
- CalendarSyncModule (app-only client-credentials Graph auth, separate
from the interactive SSO app registration; new env vars
M365_CALENDAR_TENANT_ID/_CLIENT_ID/_CLIENT_SECRET).
- Outbox + minute-interval cron worker (CalendarSyncOutboxEntry) instead
of a synchronous push, since Graph calls are too slow/flaky to block a
domain create request on; failed syncs retry up to 5 times.
- Per-source-type toggle plus a group-ID field, configurable in
Verwaltung > Automatisierung > Kalender-Sync (Verwaltung > Backup, an
env var still wins on those fields where one applies to related config).
- Hooked into Tasks/Inspections/Treatments/Feedings/Harvests/Queens at
their existing create (and Task update/complete) points; entries render
as all-day Graph events to avoid timezone ambiguity.
- Documents the design decision in ARCHITECTURE.md (Annahme 9), since
this is a net-new capability the lastenheft doesn't scope.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
2 links center side-by-side, 3 sit in one row, and 4 form a centered
2x2 block - previously flex-wrap could leave an uneven number stranded
awkwardly on its own line.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
- SiteSettings gains websiteUrl (e.g. online shop link) alongside the
existing social links, editable in Verwaltung > Automatisierung.
- /herkunft/[lot] now shows the apiary's Standort instead of its internal
Name (so multiple apiaries at one location read as one origin), adds
small icons per social link, and surfaces previously-fetched-but-unused
data (Honigsorte, Abfülldatum, stat cards) to make the page a richer
customer-facing info page.
- Backup schedule (path/cron/retention) is now also configurable via the
GUI (Verwaltung > Backup), stored in a new BackupSettings singleton;
an env var always wins over the GUI value per field, and the scheduler
reloads its cron job live when GUI settings change.
- New Automatisierung rules let an admin define a format (e.g.
"WH-{YYYY}-{SEQ:4}") that auto-generates HoneyBatch.batchNumber and
BottlingBatch.lotNumber on creation, with an atomic per-rule sequence
counter and a manual reset. Manual entry still works when the rule is
inactive.
Backup export/import now covers Role, User, RolePermission,
Notification, AuditLog and Document (with actual file content), not
just business data - matching "wirklich alle Daten". Role/User use
natural keys (role.key, resource+action) so a bundle restores cleanly
into a different database.
Adds BackupSchedulerService (@nestjs/schedule) for cron-based automatic
backups to a local path (BACKUP_STORAGE_PATH), with retention pruning
and a file-mtime guard against duplicate runs across replicas. New
GET /backup/schedule and POST /backup/schedule/run-now endpoints, plus
a status card and manual trigger in the frontend Backup settings page.
Menü unter "Verwaltung" wie gewünscht neu strukturiert:
- "Benutzer & Rollen" (/settings/users-roles): Benutzerverwaltung und
Rollen & Rechte als zwei Reiter auf einer Seite (vorher zwei
getrennte Menüpunkte).
- "Automatisierung" (/settings/automation): Webhooks, Wiederkehrende
Aufgaben, Etikettenvorlagen und Social Media Links als Reiter
gebündelt - dazu (eigene Ergänzung, da sonst zwei fast gleich
benannte Menüpunkte "Automatisierung"/"Automatisierungen"
nebeneinander gestanden hätten) auch die bestehenden
Automatisierungsregeln als fünfter Reiter "Regeln".
- "Backup" (/settings/backup): neuer voller Daten-Export/Import
(Commit zuvor) plus der bisherige CSV-Import, beides auf einer Seite.
Aktiver Reiter wird als ?tab=... in der URL gehalten statt nur im
Client-State, damit z.B. "Speichern" im Etiketten-Designer wieder auf
den richtigen Reiter zurückführt.
Technisch: bestehende Seiteninhalte in Reiter-Komponenten extrahiert
(PageHeader entfernt, Rest unverändert übernommen), Dialog-Unterkomponenten
in die neuen Ordner verschoben (git mv, Historie bleibt erhalten). Die
Etiketten-Editor-Routen (/settings/label-templates/new, /[id]) bleiben
unverändert bestehen, nur der Rücksprung nach dem Speichern zeigt jetzt
auf den neuen Reiter. settings/layout.tsx's Berechtigungsliste um
label-templates:read, site-settings:read, backup:export/import ergänzt -
schließt nebenbei eine bestehende Lücke (eine Rolle mit ausschließlich
diesen Rechten wurde vorher faelschlich von der ganzen
Verwaltungssektion ausgesperrt.
Neuer Reiter-Baustein src/components/tab-switcher.tsx (generisch, nach
Vorbild von ActiveHistoryTabs). sidebar.tsx's NavItem.permission
unterstützt jetzt auch ein Array (irgendeine der Berechtigungen reicht),
fuer die neuen gebündelten Menüpunkte.
Lokal verifiziert: typecheck/lint/test/build fehlerfrei, Build zeigt
alle neuen Routen korrekt, Live-Test im Browser (ohne Login) zeigt
sauberen Redirect zu /login ohne Konsolenfehler.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Neues Modul src/backup/ fuer Backups und Datenuebertragung zwischen
Deployments: GET /backup/export liefert ein JSON-Bundle mit allen
fachlichen Daten (Bienenstaende bis Etikettenvorlagen, ~30 Tabellen),
POST /backup/import spielt so ein Bundle wieder ein.
Bewusst NICHT enthalten: User/Rollen/Rechte (werden ueber die
Benutzerverwaltung gepflegt, Permission-Zeilen werden ohnehin bei
jedem Start aus dem Code neu synchronisiert), Audit-Log (unveraenderliche
Historie), Notifications (fluechtig), Dokumente (nur die DB-Zeile waere
gesichert, nicht die tatsaechliche Datei auf der Festplatte).
Import laeuft in einer Transaktion, Tabellen in Abhaengigkeitsreihenfolge
(z.B. Bienenstand vor Volk, Volk vor Durchsicht). Selbstreferenzen
(Koenigin -> Mutterkoenigin) werden in zwei Durchgaengen aufgeloest,
damit die Reihenfolge im Export keine Rolle spielt. userId-Felder, die
im Zielsystem nicht existieren (z.B. Uebertragung auf ein anderes
Deployment), werden auf den ausfuehrenden Admin umgebogen statt die
Fremdschluessel-Pruefung scheitern zu lassen. AutomationRule wird ueber
triggerType statt id abgeglichen, da diese Zeilen code-fest sind.
Neue Berechtigungen backup:export/backup:import. 8 neue Unit-Tests
fuer die Sonderfaelle (Selbstreferenz, userId-Remapping,
AutomationRule-Abgleich, Audit-Eintrag).
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Im automatischen Modus (Spalten/Zeilen leer) wird jetzt zusaetzlich
berechnet, wie viele Etiketten pro Blatt passen, wenn man sie um 90
Grad dreht (Breite/Hoehe vertauscht). Passt die gedrehte Anordnung mehr
drauf, wird sie verwendet - die Etiketten selbst bleiben inhaltlich
unveraendert (nur auf dem Blatt gedreht), jedes einzelne Etikett bleibt
beim Abziehen normal lesbar.
Bei explizit gesetzten Spalten/Zeilen (z.B. fuer ein vorgedrucktes
Avery-Raster) greift die Drehung nicht - ein reales Blatt hat feste,
nicht drehbare Felder.
Rotation erfolgt via pdfkit-Rotation um die Mitte des Feldes: ein
Etikett, exakt 90 Grad um seinen eigenen Mittelpunkt gedreht, ergibt
immer eine vertauschte Bounding-Box an derselben Stelle - unabhaengig
von der Drehrichtungskonvention. 2 neue Tests belegen ueber die
Seitenzahl, dass die Drehung greift, wenn sie mehr Etiketten
unterbringt, und dass ein explizites Raster unangetastet bleibt.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Der voran gegangene Fix (proxy.ts nach src/ verschoben, damit Next.js
sie ueberhaupt erkennt) hat eine zweite, bislang tote Stelle live
geschaltet: eine Cookie-Pruefung auf "refreshToken", die zu /login
umleitet, wenn das Cookie fehlt - fuer /settings, /apiaries, /colonies,
/queens.
Dieses Cookie wird vom BACKEND gesetzt (eigene Domain,
apicorebackend.* statt apicore.*, Pfad /api/v1/auth) und ist dem
Browser zufolge nie Teil einer Anfrage an das Frontend - proxy.ts lief
auf der Frontend-Domain und konnte dieses Cookie strukturell nie sehen.
Solange die Datei am falschen Ort lag und nie ausgefuehrt wurde, blieb
das folgenlos; sobald sie lief, hat sie jeden Aufruf dieser Seiten
serverseitig zurueck zu /login geschickt, unabhaengig vom tatsaechlichen
Login-Status.
Die Cookie-Pruefung komplett entfernt statt repariert: es gibt keinen
Weg, ein cross-origin gesetztes Cookie hier sichtbar zu machen, ohne
seinen Schutz (Pfad-Scoping) aufzuweichen. Die echte Session lebt
ohnehin nur clientseitig (Access-Token im Speicher, Refresh-Cookie nur
fuer Backend-Aufrufe) - src/app/(app)/layout.tsx prueft das bereits
korrekt und leitet bei Bedarf weiter. proxy.ts setzt jetzt nur noch die
Security-Header.
Lokal verifiziert: /settings/users ohne jedes Cookie liefert jetzt 200
statt eines Redirects, CSP zeigt weiterhin den korrekten
Laufzeit-API_URL.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Root Cause, zweifach:
1. proxy.ts lag im Projekt-Root statt unter src/ (wo app/ tatsaechlich
liegt). Next.js hat die Datei dadurch nie erkannt - ueberprueft via
Build-Output ("Proxy (Middleware)" fehlte komplett) und via
routes-manifest.json. Das betraf nicht nur die neuen
Security-Header, sondern auch die schon laenger bestehende
Login-Redirect-Logik, die dadurch vermutlich nie serverseitig
gegriffen hat - vorher unbemerkt, weil Client-seitige Auth-Checks
das kaschiert haben.
2. Die CSP (mit dynamischem connect-src aus API_URL) lag in
next.config.ts's headers() - die wird beim Docker-Build EINMAL
aufgeloest und ins Standalone-Output eingefroren, nicht bei jedem
Container-Start neu gelesen, wie im Code-Kommentar faelschlich
behauptet. Jedes Deployment bekam dadurch permanent
"connect-src http://localhost:3001" (den Fallback, da API_URL beim
Build nicht gesetzt ist) - unabhaengig vom tatsaechlich im Pod
gesetzten API_URL. Jede Anfrage vom Frontend ans Backend (Login,
SSO-Status, Herkunftsseite) wurde dadurch von jedem Browser
CSP-blockiert.
Fix: CSP + Security-Header ziehen nach proxy.ts (jetzt korrekt unter
src/) um - Proxy laeuft mit Node.js-Runtime frisch bei jeder Anfrage,
liest process.env.API_URL also korrekt zur Laufzeit. next.config.ts
behaelt nur noch die statische sw.js-Cache-Control-Regel.
Lokal end-to-end verifiziert: Build zeigt jetzt "ƒ Proxy (Middleware)"
(vorher komplett abwesend), CSP-Header zeigt nach Server-Start einen
distinct getesteten API_URL-Wert korrekt an (nicht mehr den
Localhost-Fallback), und ein echter Login-Klick im Browser schickt die
Anfrage jetzt an die richtige, zur Laufzeit gesetzte Backend-Origin.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Beim Umstellen von PermissionsGuard auf Fail-Closed wurde der
@Public()-Check vergessen. JwtAuthGuard ueberspringt bei @Public()
zwar die Authentifizierung, aber PermissionsGuard laeuft als
eigenstaendiger, separat registrierter APP_GUARD trotzdem mit - und
verweigerte seitdem jede Route ohne @RequirePermission/
@AnyAuthenticatedUser, selbst wenn sie @Public() war. Betroffen: health
und health/ready (die gemeldeten 403), aber ebenso Login, SSO-Flow und
die oeffentliche Herkunftsseite.
PermissionsGuard prueft jetzt @Public() als erstes und laesst diese
Routen durch, genau wie JwtAuthGuard es bereits tut. Regressionstest
ergaenzt.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Abfüllungen hatten keine Übersichtsseite - die Navigation ging direkt
auf /bottling/new, sodass eine angelegte Abfüllung danach nirgends
mehr sichtbar war. Neue Seite /bottling listet alle Abfüllungen
(Archiv), Klick führt zur bestehenden Detail-/Bearbeiten-Seite.
Navigationslink zeigt jetzt auf die Liste statt direkt auf "Neu".
Neues Modul site-settings (Singleton-Tabelle SiteSettings) für
YouTube-/Instagram-/TikTok-Links, verwaltbar unter Verwaltung > Website
(neue Berechtigung site-settings:read/update). Die öffentliche
Herkunftsseite (/herkunft/[lot]) zeigt die konfigurierten Links jetzt
zusätzlich zur Chargen-Herkunft an - PublicService liefert sie über
das bereits oeffentliche /public/batches/:lotNumber mit aus, keine
zusaetzliche unauthentifizierte Route noetig.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
JWT_ACCESS_SECRET war zwar als Pflicht-Env-Var geprueft, aber ohne
Mindestlaenge - ein kurzer/erratbarer Wert haette nicht nur das
Faelschen von Access-/MFA-/SSO-Tokens fuer jeden Nutzer erlaubt,
sondern auch (ueber mfa-crypto.util.ts, das den AES-Schluessel fuer
gespeicherte TOTP-Secrets aus genau diesem Wert ableitet) das
Entschluesseln aller gespeicherten MFA-Secrets.
main.ts bricht den Start jetzt mit klarer Fehlermeldung ab, wenn
JWT_ACCESS_SECRET kuerzer als 32 Zeichen ist. .env.example und das
K8s-Secret-Template zeigen jetzt einen ausreichend langen Platzhalter
mit Erzeugungshinweis (openssl rand -hex 32).
Betrifft nur main.ts's bootstrap() - e2e-Tests bootstrappen ueber
test/utils/create-test-app.ts direkt via configureApp() und durchlaufen
diese Pruefung nicht, daher keine Testanpassung noetig.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Ein zweiter, unabhaengiger Audit-Durchlauf (frischer Blick, keine
Vorkenntnis meiner eigenen Fixes) bestaetigte alle bisherigen Fixes als
korrekt verdrahtet und fand 3 LOW-severity Robustheitsluecken - keine
davon war ausnutzbar, aber alle sind jetzt behoben:
- HarvestsService.create/ExtractionsService.create pruefen jetzt, dass
referenzierte colonyIds/harvestIds tatsaechlich existieren, bevor die
Transaktion laeuft - vorher fuehrte eine unbekannte Id zu einem
ungefangenen 500 mit rohem Prisma-Fehler statt einer sauberen 400.
- Etiketten-Bildelemente durchlaufen jetzt denselben Magic-Byte-Check
wie Avatar-/Dokument-Uploads, statt pdfkits Bildparser ungepruefte
Bytes zu uebergeben.
- Webhook-Secret-Redaktion im Audit-Log nutzt jetzt explizites delete
statt sich auf Prismas JSON-Serialisierung zu verlassen, die
undefined-Werte verwirft (fragil, falls sich die Serialisierung mal
aendert).
Co-Authored-By: Claude Sonnet 5 <[email protected]>
RolesService.create/update uebernahm eine explizit uebergebene
permissionKeys-Liste ungeprueft. Ein Nutzer mit nur roles:update
konnte damit JEDER Rolle - auch der eigenen, ohne users:assign-role -
beliebige weitere Rechte (bis hin zu allen 92 Katalog-Eintraegen)
hinzufuegen, da das Bearbeiten der eigenen Rollen-Rechte nicht ueber
User.roleId laeuft und daher vom bereits gefixten
users:assign-role-Check nicht erfasst wurde.
Neue Regel (Standard-RBAC-Prinzip "keine Rechte ueber das eigene
Rechte-Set hinaus vergeben"): create/update pruefen jetzt, dass der
aufrufende Nutzer jedes zu vergebende Recht selbst bereits besitzt,
bevor die Rolle angelegt/geaendert wird.
roles.service.ts hatte bislang keine Unit-Tests; neue Spec-Datei deckt
den Fix sowie die uebrigen Methoden (create/update/deactivate/
reactivate) ab.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Eine explizit im Etiketten-Designer eingetragene Spalten-/Zeilenzahl
wurde beim PDF-Rendern ungeprueft uebernommen. Passt sie nicht auf ein
A4-Blatt (z.B. 90mm breite Etiketten x 4 Spalten bei 5mm Rand), ragen
die hinteren Spalten/Zeilen ueber den Blattrand hinaus - der Drucker
schneidet sie dann einfach ab, statt sie zu drucken.
- Backend (label-templates.service.ts): Spalten/Zeilen werden jetzt auf
das tatsaechlich auf A4 passende Maximum begrenzt, bevor gerendert
wird - betrifft auch bereits gespeicherte Vorlagen mit zu grossen
Werten.
- Frontend (label-designer.tsx): zeigt die maximal passende Spalten-/
Zeilenzahl direkt im Seiten-Layout-Panel an und warnt inline, wenn
der eingetragene Wert das Blatt sprengen wuerde.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
PermissionsGuard erlaubte bisher jede Route ohne @RequirePermission
automatisch fuer jeden eingeloggten Nutzer (fail-open) und verliess sich
allein auf test/authorization-coverage.e2e-spec.ts, um eine vergessene
Decorator-Angabe vor dem Deploy zu erwischen - der Guard selbst hatte
kein Default-Deny.
Neuer Decorator @AnyAuthenticatedUser() macht das bisher stillschweigend
erlaubte Verhalten explizit fuer die 12 Routen, die es tatsaechlich
brauchen (ProfileController komplett, NotificationsController komplett,
AuthController.me - alle bereits vorher auf der Allowlist des e2e-Tests).
PermissionsGuard verweigert jetzt jede Route, die weder
@RequirePermission noch @AnyAuthenticatedUser traegt - eine vergessene
Decorator-Angabe schlaegt jetzt zur Laufzeit fehl (403), nicht mehr nur
im Test. Der e2e-Test prueft jetzt denselben Decorator-Vertrag statt
einer separat gepflegten Allowlist, kann also nicht mehr vom
Guard-Verhalten abweichen.
Neue Unit-Tests fuer PermissionsGuard decken beide Codepfade und das
Fail-Closed-Verhalten ab.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Weitere vom Sicherheitsaudit gemeldete Luecken behoben, die bisher als
niedrige Prioritaet offen gelassen wurden:
- SSO-Login (Entra ID) prüft jetzt lokale MFA genauso wie der
Passwort-Login: ist mfaEnabled gesetzt, liefert /auth/sso/complete
denselben mfaRequired-Codepfad wie /auth/login, bevor Tokens
ausgestellt werden. Frontend (Login-Seite, AuthContext) zeigt in dem
Fall den bestehenden Code-Eingabe-Schritt statt die Anmeldung
stillschweigend abzuschliessen.
- Avatar- und Dokument-Uploads pruefen jetzt die tatsaechlichen
Datei-Bytes (Magic Numbers fuer PNG/JPEG/WEBP/PDF) statt nur dem
client-angegebenen MIME-Typ/data:-URL-Praefix zu vertrauen.
- Swagger-UI (/api/v1/docs) ist jetzt standardmaessig deaktiviert
(ENABLE_SWAGGER=true erforderlich), da sie zuvor das komplette
API-Schema ungeschuetzt oeffentlich preisgab.
npm-audit-Warnungen (deepmerge-ts/mysql2 unter prisma) bewusst nicht
gefixt: kein stabiles Prisma-Release loest sie bislang auf, nur eine
Release-Candidate-Hauptversion - das Risiko eines Downgrades/RC-Upgrades
der Datenbankschicht uebersteigt den Nutzen fuer eine Kette, die zur
Laufzeit ohnehin nicht erreichbar ist (nur Prisma-CLI, nur MySQL-Treiber,
dieses Projekt nutzt Postgres).
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Ein Sicherheitsaudit deckte auf, dass jede Rolle mit users:update sich
selbst per PATCH /users/:id auf eine Admin-Rolle heraufstufen konnte,
da roleId ohne eigene Berechtigung und ohne Selbstbezug-Pruefung
akzeptiert wurde. Ausserdem lag ein fest im Code hinterlegtes
Standard-Admin-Passwort vor, das bei fehlender SEED_ADMIN_PASSWORD-Env
greift.
- Neue Berechtigung users:assign-role, getrennt von users:create/update,
fuer jede Rollenzuweisung erforderlich
- Selbstaenderung der eigenen Rolle serverseitig blockiert (auch fuer
Admins), ebenso Selbst-Reset der eigenen MFA ueber die Admin-Route
- Seed erzeugt bei fehlender Env ein zufaelliges statt eines fixen
Passworts
- Wiederverwendung eines bereits rotierten Refresh-Tokens (Diebstahl-
Indiz) widerruft nun alle Sessions des betroffenen Nutzers
- Rate-Limit auf MFA-Enrollment-Bestaetigung ergaenzt
Co-Authored-By: Claude Sonnet 5 <[email protected]>
apicore.encry.de lieferte bisher gar keine Security-Header aus (per
curl bestaetigt) - CSP, X-Frame-Options, X-Content-Type-Options,
Referrer-Policy, HSTS fehlten komplett, waehrend das Backend das
laengst per helmet() macht. Ergaenzt fuer alle Routen. Bewusst ohne
CSP-Nonce (haette dynamic rendering fuer jede Seite erzwungen und
damit die bestehende statische Optimierung gekostet) - stattdessen
script-src/style-src 'unsafe-inline', aber default-src/connect-src/
img-src/object-src/frame-ancestors/form-action/base-uri strikt
eingeschraenkt. Lokal mit dem echten Production-Build (standalone
server + next.config.ts) verifiziert: Login-Seite rendert/hydratisiert
sauber, keine CSP-Verstoesse in der Konsole, Header per curl bestaetigt.
POST /profile/avatar schlug live mit net::ERR_FAILED fehl (Verbindung
abgebrochen, keine Antwort) - selbst bei einer winzigen Testdatei.
CORS-Preflight lief sauber durch, nur der eigentliche multipart-POST
nicht; passt zum bereits dokumentierten Cloudflare-Verhalten bei
multipart-Uploads (siehe .gitea/workflows/*-image.yml). Umgangen durch
Umstellung auf denselben JSON-Content-Type, der fuer alle anderen
Endpunkte nachweislich zuverlaessig funktioniert: Bild wird
clientseitig als data:-URL kodiert und im JSON-Body gesendet.
Globales JSON-Body-Limit auf 5mb angehoben (deckt 2MB-Bild plus
Base64-Overhead).
Eine KI-erzeugte JSON ohne Kenntnis des echten Schemas (z.B. eigene
Feldnamen wie xMm/yMm oder erfundene Elementtypen wie rectangle/svg)
scheiterte bisher mit "Datei enthält keine gültige Etikettenvorlage"
ohne Hinweis, was falsch ist. Jetzt wird pro Element benannt, was
fehlt oder welcher Typ nicht unterstützt wird.
- LabelTemplate bekommt columns/rows (optional, leer = automatisch
passend), plus konfigurierbare Raender/Abstaende - damit ein Etikett
exakt auf das vorgedruckte Raster eines physischen Etikettenbogens
(z.B. Avery) gelegt werden kann statt nur automatisch berechnet
- Startposition beim Drucken waehlbar, fuer bereits angebrochene Boegen
(erste Seite startet ab der gewaehlten Position, Folgeseiten wieder
von vorne)
- Vorlage als JSON exportieren/importieren, damit man sich ein Etikett
auch per KI erstellen und danach importieren kann
- Zoom-Regler fuer den Canvas (50-200%), da Etikettgroessen stark variieren
- X/Y-Positionsfelder pro Element fuer praezise Platzierung statt nur Drag
- Element duplizieren-Button
- Entf/Backspace loescht das ausgewaehlte Element
DropdownMenuLabel (rendert Menu.GroupLabel) stand direkt im
DropdownMenuContent statt in einem DropdownMenuGroup - Base UI wirft
dafür beim Öffnen einen ungefangenen Fehler, der den kompletten
React-Baum crasht. Das war die eigentliche Ursache des seit Sitzungsbeginn
gemeldeten "This page couldn't load" beim Klick auf das Profilbild
oben rechts, nicht der Service Worker (der war schon korrekt).
git add -u im letzten Commit hat neue (untracked) Dateien übersprungen:
die Migration für LabelTemplate.elements, die neuen Typ-/DTO-Dateien und
den kompletten Drag-and-Drop-Editor fehlten dadurch im Build.
labels.ts und die ColonyCondition-Migration waren untracked und fehlten
im letzten Commit, dadurch schlug der Docker-Build mit "Module not
found: @/lib/colonies/labels" fehl.
ts-node prisma/seed.ts lief bisher live im Container - der
TypeScript-Compiler dafuer braucht deutlich mehr Heap als das
512Mi-Pod-Limit, daher sporadische "JavaScript heap out of memory"-
Crashes beim Start (manchmal lief's, manchmal nicht, reine
GC-Zufaelligkeit an der Speichergrenze).
Fix: seed.ts wird jetzt wie main.ts beim Docker-Build vorkompiliert
(tsconfig.seed.json, rootDir bewusst auf prisma/ statt Projekt-Root
gesetzt, damit der relative Import '../generated/prisma' nach dem
Kompilieren weiterhin auf denselben Pfad zeigt). Laufzeit-CMD startet
nur noch plain `node dist/seed.js` - kein TypeScript-Compiler mehr im
Container. ts-node/typescript zurueck zu devDependencies (nur noch
fuer den Build-Schritt gebraucht, nicht mehr zur Laufzeit).
incremental in tsconfig.seed.json explizit deaktiviert - ein stale
tsconfig.seed.tsbuildinfo liess tsc beim zweiten Testlauf sonst
faelschlich "nichts zu tun" melden (derselbe Fallstrick wie bei
tsconfig.build.tsbuildinfo).
Verifiziert: tsc/eslint clean, 304/304 Tests, Build reproduzierbar
(zweimal hintereinander ohne Cache-Reset getestet), node dist/seed.js
lief erfolgreich gegen die echte lokale DB (erkannte den bestehenden
Admin korrekt und legte ihn nicht erneut an).
Bisher waren 4xx-Fehler (falsches Passwort, Validierungsfehler,
fehlende Berechtigung) im Log komplett unsichtbar - Nest loggt
standardmaessig nur unbehandelte 5xx-Fehler. Ein Login, das aus
Infra-Gruenden fehlschlaegt (siehe TLS/Cloudflare-Vorfall), war dadurch
allein aus den Logs nicht diagnostizierbar.
- AllExceptionsFilter (global): loggt jede Exception mit Methode/Pfad/
Status (4xx als warn, 5xx als error mit Stacktrace). Client-Antwort
bleibt unveraendert - HttpExceptions liefern weiter exakt ihren
bisherigen Body, unbehandelte Fehler weiter nur eine generische
500-Meldung ohne interne Details.
- requestLoggerMiddleware (global): eine Zeile pro Request (Methode,
Pfad, Status, Dauer), damit auch erfolgreiche Requests im Log
auftauchen - vorher gab's dafuer gar keine Spur.
- AuthService.login()/verifyMfaAndLogin(): konkrete Ursache pro
Fehlschlag (unbekannter User, deaktiviert, falsches Passwort,
falscher MFA-Code) zusaetzlich zum bestehenden Audit-Log-Eintrag.
Nirgends werden Passwoerter, Tokens oder Request-Bodies geloggt - nur
Methode, Pfad, Status, Dauer und die (bereits client-sichere)
Fehlermeldung.
Verifiziert: tsc/eslint clean, 304/304 Tests (7 neu), Build clean.
Bisher wurde nirgends automatisch ein erster Admin-User angelegt -
prisma/seed.ts existierte, war aber weder im Dockerfile-CMD noch sonst
irgendwo verdrahtet. Auf einer frischen Datenbank haette also niemand
sich einloggen koennen.
- Dockerfile-CMD: migrate:deploy && seed && node dist/main.js
- ts-node/typescript von devDependencies zu dependencies verschoben
(seed.ts laeuft als rohes TypeScript zur Laufzeit, kein
Kompilierschritt), tsconfig.json zusaetzlich ins Runtime-Image kopiert
- seed.ts gegen Unique-Constraint-Race bei mehreren gleichzeitig
startenden Pod-Replicas gehaertet (P2002 abgefangen statt Absturz)
- Login-Banner im Log nur beim tatsaechlichen Erst-Anlegen (URL aus
FRONTEND_ORIGIN, Benutzername, Passwort) - bewusst nicht bei jedem
Neustart, sonst wuerden veraltete/geaenderte Zugangsdaten weiter
geloggt und Credentials landen unnoetig dauerhaft in den Logs
Seed ist bereits idempotent per E-Mail-Check (auch bei deaktivierten
Usern), UsersService.assertNotLastSystemRoleUser verhindert ohnehin
das Deaktivieren des letzten aktiven Admins - kein Aenderungsbedarf
dort, siehe Konversation.
Verifiziert: tsc/eslint clean, 297/297 Tests, Build clean. Migration
und Schema live gegen die lokale DB bestaetigt (ueber WSL direkt, da
Windows->WSL-Portweiterleitung weiterhin kaputt ist); der eigentliche
Seed-Lauf selbst konnte nicht End-to-End gegen die DB verifiziert
werden (gleiches Verbindungsproblem), Banner-Ausgabe isoliert
smoke-getestet.
Benutzer koennen jetzt selbst Zwei-Faktor-Authentifizierung einrichten
und ihr Profil verwalten (Name, E-Mail, Passwort, Profilbild).
Backend:
- Neues MfaModule (src/auth/mfa/): AES-256-GCM-verschluesselte
TOTP-Secrets (otplib), QR-Code-Erzeugung, kurzlebige Setup-/
Challenge-JWTs fuer den 2-Schritt-Login-Flow
- Neues ProfileModule (src/profile/): PATCH /profile,
POST /profile/password, POST/DELETE/GET /profile/avatar,
POST /profile/mfa/enroll(/confirm), POST /profile/mfa/disable -
alle self-scoped ueber actor.id, ohne @RequirePermission (siehe
ALLOWLIST in authorization-coverage.e2e-spec.ts)
- AuthService.login() zweistufig: bei aktivem MFA erst
{ mfaRequired, mfaToken } statt Tokens, neuer Endpunkt
POST /auth/mfa/verify schliesst den Flow ab
- Admin-Recovery: POST /users/:id/mfa/reset (users:update) fuer
verlorene MFA-Geraete, da es keinen E-Mail-Reset-Flow gibt
Frontend:
- Login-Seite mit zweitem Schritt fuer den TOTP-Code
- Neue Profilseite (/profile, ausserhalb des admin-only
Settings-Bereichs): Profilbild, Name/E-Mail, Passwort, MFA-Ein-/
Ausrichtung mit QR-Code-Dialog
- Header zeigt jetzt Avatar mit Dropdown (Profil/Abmelden) statt
nur E-Mail-Text
- Admin-Users-Tabelle zeigt MFA-Status + Reset-Button
Haertung nach Security-Review (Backend/Frontend-Diff, kein CRITICAL,
mehrere MEDIUM behoben):
- Access-Token wird bei Passwortwechsel sofort ungueltig
(User.passwordChangedAt + JWT-iat-Vergleich in
PermissionResolverService.isTokenValid, vorher isUserActive) statt
erst nach bis zu 15 Minuten
- E-Mail-Aenderung verlangt jetzt aktuelles Passwort
- POST /profile/password und /profile/mfa/disable haben jetzt das
gleiche strikte Throttling wie Login (SENSITIVE_ACTION_THROTTLE)
- Kein stiller leerer Verschluesselungs-Key-Fallback mehr (wirft
jetzt explizit)
Bewusst zurueckgestellt (dokumentiert, kein aktiver Exploit-Pfad):
dedizierter Verschluesselungs-Key fuer MFA-Secrets statt
JWT_ACCESS_SECRET-Ableitung, dedizierte users:mfa-reset-Permission
statt users:update, separate JWT-Secrets pro Token-Zweck.
Verifiziert: tsc/eslint clean, 297/297 Backend- + 31/31
Frontend-Unit-Tests, Build clean auf beiden Projekten. e2e weiterhin
durch bekanntes WSL2/Postgres-Problem blockiert. Unabhaengiges
Security- und Code-Quality-Review durchgefuehrt (0 CRITICAL, 0 HIGH).
