c0ecf2fa3bebd46bdfd0169efd73b89653d1dfed
5 Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
 Jakob Wennberg
|
c7a75d069d |
feat(ai): job-shaped AI service with OpenAI-compatible backend, extraction-first; stop extracting every inbox document twice (#1740)
* feat(ai): job-shaped AI service with OpenAI-compatible backend, extraction-first; stop extracting every inbox document twice Sovereign plan WS1 PR1 (#1406 Tier 2, extraction-first, aligned with the AI surface audit). lib/ai grows a job-shaped service (generateText / generateStructured / extractFromDocument; no streaming members yet, see plan rule R3): - services/anthropic-family delegates to the existing createAiClient() and sends the exact request literals the inbox extractor sent before (request-shape tests deep-equal them), so hosted Bedrock stays byte-identical. - services/openai-compatible talks to any chat-completions endpoint (BYO Swedish provider) via Vercel AI SDK 6.x, exact-pinned and guarded: images as parts, PDFs rasterized with poppler (AI_PDF_MODE) or sent natively, AI_VISION / AI_STRICT_JSON declared, honest skips (ai_no_vision, pdf_rasterizer_missing) instead of fake failures. - config.ts: AI_PROVIDER/AI_BASE_URL/AI_API_KEY/AI_MODEL and per-tier AI_*_MODEL with the legacy BEDROCK_* names kept as the same overrides; getAiStatus() is the single source of truth for "is AI wired up". - provider.ts: openai-compatible in the auto-detect chain (after Bedrock and the direct API); createAiClient() refuses it loudly. Document extraction moves onto the service and gets the audit's fixes: - Inbox documents were extracted TWICE (pipeline A ran inside uploadDocument() before the inbox row existed, so its dedupe branch never fired; 3 707 + 1 666 calls / 30 d). The inbox now declares extractionOwner on the upload, the extension yields, and the inbox mirrors its single outcome onto document_attachments from every writer (sync, deferred, attach, retry, MCP). - Every "no extraction will ever happen" outcome is stamped (skipped:no_ai_entitlement / ai_unconfigured / system_generated / ...); the status route maps the quiet ones to 'disabled' on the first poll instead of a 30 s client timeout. Prod showed 309 of the 327 never-extracted uploads were the paywall working silently. - Self-generated documents (our own invoice PDFs, payout files) are no longer OCR'd. - Agent invoke answers 503 ai_unconfigured when the deployment has no assistant backend, distinct from the paywall. Guard: new direct-ai-client antipattern check (shrink-only allowlist of the pre-abstraction SDK callers) plus exact pins for @anthropic-ai/sdk, ai and @ai-sdk/openai-compatible. Verified: 15 958 unit tests green, guards, lint ratchet, typecheck, and a live smoke against hosted Bedrock through the new service (ping, streamed tool turn, thinking+cache, PDF extraction). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * feat(ai): make AI_API_KEY optional for OpenAI-compatible endpoints (keyless local model servers) A local model server (llama.cpp's server, Ollama /v1, LM Studio, vLLM) usually has no auth. Before, the OpenAI-compatible backend required both AI_BASE_URL and AI_API_KEY to count as configured, so running Accounted on a local model meant setting a meaningless placeholder key. - resolveAiProvider / hasAiCredentials: a base URL alone is now enough. - services/openai-compatible: only send Authorization: Bearer when AI_API_KEY is set, so a keyless server is never handed an empty bearer; a hosted provider that needs a key still sets it. - Docs (SELF-HOSTING Option 3: local-model example, key marked optional), DECISIONS. Verified: with no AI_API_KEY, just AI_BASE_URL + AI_MODEL, getAiStatus() reports configured=true / provider=openai-compatible (live). lib/ai suite 71 green; tsc, guards, lint clean. Bedrock/Anthropic logic unchanged. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> --------- Co-authored-by: Jakob Wennberg <311770904+jakobwennberg-oss@users.noreply.github.com> Co-authored-by: Claude Fable 5 <noreply@anthropic.com> |
||
|
Jakob Wennberg
|
1dce9227a4 |
feat(assistant): make the thumbs actually report something (#1236)
* feat(assistant): make the thumbs actually report something The thumbs up/down under an assistant answer shipped wired to nothing. They lit up, the vote died in component state, and the code said so in a comment nobody reading the UI could see. An affordance that looks like it reports something and does not is worse than no affordance: it spends the user's goodwill once, and silently. They now post to a new /api/agent/feedback, which emits the SAME agent.feedback event the gnubok_feedback MCP tool emits, with actorType 'user'. The product team already queries event_log for that type, so chat votes land in the backlog they read rather than in a second place someone has to remember to look at. event_log takes the payload as jsonb and already treats agent.* as telemetry, so there is no migration. The conversation id is caller-supplied, so it gets the same ownership check /api/agent/invoke got: without it a member could file feedback against a colleague's thread and the backlog would carry conversations the reporter never saw. Mutation-checked, three tests fail when the guard is removed. The pressed state is set only after the server accepts the vote, so the button never claims a report that never arrived, and a vote does not toggle off: it is append-only telemetry, and offering an undo we cannot honour would be a control that lies. Changing your mind sends the other sentiment instead. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * fix(assistant): drop the unused free-text field from the feedback route compliance-swarm flagged the `comment` field as an undocumented PII exposure: free text embedded verbatim into the agent.feedback payload and written to event_log under telemetry retention, with no data-classification decision, next to a PostHog policy that treats the same class of data differently. The finding is right, and the field was worse than it looked: no caller ever sent one. The UI posts sentiment and a turn index. So this is dead API surface whose only effect was to accept whatever a user might type, in an accounting product, into a 180-day log: client names, personnummer, case details. Removed rather than documented. A comment box is a reasonable thing to want, but it needs its own classification and redaction decision made with the UI in front of it, not inherited from an unused parameter. The test now asserts the property instead of the field's absence: a caller that posts a comment anyway must not get it stored anywhere in the payload. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5 <noreply@anthropic.com> |
||
|
Jakob Wennberg
|
f0f3050f54 |
fix(assistant): stop the chat loading in stages (#1210)
* fix(assistant): stop the chat loading in stages PR2 of the assistant UI makeover (dev_docs/assistant_redesign_plan.md section 7). No redesign; this is the "it loads in different stages" complaint, traced to four separate staging points and one dead link. Resumed conversations rendered a column of EMPTY bordered cards until the markdown chunk arrived, then filled in all at once and reflowed the thread. The chunk was deferred with a null fallback, which is invisible while a reply streams (nobody reads that fast) but very visible on hydrate, where every assistant bubble is already text. The chunk is now prefetched as soon as any chat surface mounts, and until it resolves the raw text renders instead of nothing, so a bubble is never blank. Clicking the assistant launcher showed NOTHING until the sheet chunk loaded: the dynamic import had no loading state at all. It now renders a skeleton in the same geometry, and the chunk is warmed on idle so the click usually hits an already-loaded module. /chat's route skeleton drew a 320px sidebar while ChatSidebar mounts collapsed as a 48px rail, so every load snapped one to the other. The skeleton now matches what actually mounts, per breakpoint. The first turn read agent_profiles twice: once in the route to build the intent's prompt template, once again in run-turn for the system prompt. The route now hands its result over. Ranked memory is deliberately NOT shared: the two queries differ (the route's selects fewer columns and orders without is_pinned, and run-turn needs ids to stamp last_accessed_at), so reusing it would silently change both the prompt and memory touch. Command palette's "Fråga Anna: ..." pointed at /chat?prompt=, but only /chat/new reads ?prompt=, so the typed question was silently dropped and the user landed on an empty state. Verified: 9526 unit tests pass, lint clean and tsc clean on every touched file, guards pass. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * fix(assistant): make the prefetches fail-safe and bounded Review follow-ups on the staged-loading batch. A rejected markdown import left the cached promise permanently rejected, so every bubble for the rest of the session stayed on the plain-text fallback and the rejection went unhandled. The cache is now cleared on failure so a later surface retries, and the rejection is swallowed. requestIdleCallback can defer indefinitely on a page that never goes idle; the 2s fallback only applied where the API is missing. The idle request now carries a 2s timeout, and the warm import cannot produce an unhandled rejection either. Adds the first-turn test for the profile-summary handover: it asserts the value read for the prompt template is what reaches the turn, so a regression that re-introduces the second read (or drops the template) fails. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5 <noreply@anthropic.com> |
||
|
Jakob Wennberg
|
ee8ddb3849 |
fix(assistant): stop cross-user conversation access, bricked threads and lost sessions (#1209)
* fix(assistant): stop cross-user conversation access, bricked threads and lost sessions Hotfix batch (PR1 of the assistant UI makeover, dev_docs/assistant_redesign_plan.md section 7). No visual change; each of these is wrong today regardless of which design lands, and three are unrecoverable per incident. /api/agent/invoke never checked who owns a resumed conversation_id. RLS on agent_conversations/agent_messages is company-scoped, not user-scoped (20260517204000), so a member could post a colleague's conversation id, have their history loaded into the prompt and read it back, while their own turns were appended to that thread. The conversations list route filters on user_id for exactly this reason. Also pins company and intent: resuming a thread from another company would mix ledgers, and resuming under a different intent would swap the tool whitelist under history the model has already seen. A turn persists the assistant message carrying tool_use blocks before the tools run, and their results only after the batch finishes. Dying in between (client disconnect terminating the function, a deploy, a slow tool) left history ending on an unanswered tool_use, which the Messages API rejects on replay: every later turn 400s, and agent_messages is append-only for the BFL trail, so nothing could repair it. History is now patched on read by synthesizing is_error tool_results, leaving the stored trail untouched. check_and_increment_agent_quota is SECURITY DEFINER in public with a caller-chosen p_user_id, so any authenticated user could drain a colleague's minute/day budget and lock them out of every agent endpoint. A plain REVOKE would break the limiter (all three callers use the user's RLS client) and, as it fails open, silently remove the spend cap: the function now refuses to act for anyone but the caller, while service-role connections keep passing an explicit id. The single reject route re-read status and then wrote unguarded, so losing the race with commit's atomic pending -> committing claim stamped `rejected` over an operation that had already posted a verifikat, invisible to the committing-state recovery sweep. Guarded on status like bulk-reject already is; a lost race is now a 409. The sheet's Escape handler listened on window with no defaultPrevented or target check while the sheet is deliberately non-modal, so pressing Esc to dismiss the reject-reason Select inside an approval card, the command palette or any dialog unmounted the sheet and discarded the conversation, the streaming turn and the un-actioned proposal. It now yields to open overlays and to focus outside the sheet. Verified: 9526 unit tests pass, lint clean on touched files, guards pass, and the new pg-real test proves the quota guard against real Postgres (attacker raises 42501, victim counters stay at 0). The four unrelated pg-real failures on this machine reproduce identically with these changes stashed. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * fix(assistant): close anon path on the quota RPC, order the ownership check ahead of writes Review follow-ups on the hotfix batch. The caller guard used auth.uid() alone, which is NULL for the `anon` role just as it is for backend roles, so an unauthenticated caller holding the public anon key (it ships in the browser bundle) could still pick any p_user_id and drain that user's quota. The guard now keys on the request role: anon and authenticated may only ever spend their own quota, backend roles keep passing an explicit id. The default PUBLIC execute grant is revoked as a second layer, with execute granted only to authenticated and service_role. Covered by a new pg test for the anon path. The ownership check ran after the onboarding.intake stamp, so a request that was about to be rejected could still write intake_completed_at. It now sits directly after the capability gate, ahead of every side effect and ahead of the company and profile reads, which also makes a rejected request cheaper. The tool-result repair matched ids anywhere in the history, but the API needs results in the message IMMEDIATELY after the tool_use. A result persisted after an intervening turn (two turns racing on one conversation) left a shape that still 400s. The repair is now positional, and orphaned or late-duplicate tool_results are dropped, since an unmatched tool_result is rejected just as an unanswered tool_use is. The Escape guard matched the Radix popper wrapper, which stays mounted when a popper is force-mounted; it now requires data-state="open" so a closed popper cannot block Escape for the rest of the session. Both new route errors are Swedish, per the user-facing error rule. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5 <noreply@anthropic.com> |
||
|
Mattsson
|
2c2743eb79 |
Check/salary bankid api (#892)
* fix(bankid): harden login/signup flow — polling, signup rollback, metadata merge, enrichment lookup - middleware: read BankID enrichment from the bankid_enrichment table (the extension_data path has been dead since the multi-tenant refactor), so company-less BankID users land on /select-company instead of the manual wizard - BankIdAuth: hard 6-min poll deadline; every failed poll counts toward the give-up limit; guard overlapping ticks so completion runs exactly once (a double /complete regenerated the magic link and invalidated the first, failing logins intermittently); retry clicks wait out the start cooldown instead of silently no-oping; Swedish messages for 429/unknown start errors - bankid/complete: all-or-nothing signup — delete the created user when the identity insert, app_metadata update, or magic-link generation fails, so a retry starts clean instead of hitting account_exists with an unusable account - bankid/unlink: read-merge-write app_metadata so has_password survives unlink (BankID-only users could otherwise strand themselves with no login method) - login: BankID "create account" CTA now links to /register instead of dismissing the notice; sv.json: fix missing å/ä/ö in settings_bankid strings Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * docs: move secondary guides into docs/, delete dead root files Move DOCKER.md, SELF-HOSTING.md, WHITELABEL.md and extensions.md (renamed EXTENSIONS.md) into a new docs/ folder and update all path references (README, setup.sh, .dockerignore image rules, docker-publish workflow comment, _example-branding, lib/branding/service.ts). Delete two dead root files: customer.json (stray API-test payload) and findings.md (point-in-time swarm audit export, criticals already filed). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Signed-off-by: Emil <emilmattsson14@gmail.com> * fix(api): security & correctness hardening + withRouteContext MFA migration across API routes Audit of ~100 app/api routes. Highlights: Security - agent/conversations: list leaked colleagues' titles + message previews (company-scoped RLS, no user filter) -> user-scoped - calendar/feed PUT: raw body into .update() allowed feed_token fixation on a public unauthenticated URL -> strict schema, content toggles only - bokslutsdispositioner: unbounded schablonintaktRate could inflate the IL 30 kap 25% periodiseringsfond cap base -> bounded - agent profile/composer/onboarding: viewers could rewrite the agent profile while sibling /verify blocked them -> role-gated Correctness - account-totals / listAssets: unbounded queries silently truncated at 1000 rows (under-counted money; skipped assets at year-end depreciation) -> fetchAllRows with stable order (+3 more pagination fixes) - voucher-gaps: swallowed detect_voucher_gaps RPC errors (BFNAR gap view could show "no gaps" when the check never ran) -> surfaced - 5 phantom-success writes (OK on zero matched rows) fixed - assets K3 component-sum validated against stale acquisition_cost -> fixed - invite silent email-send failure -> response carries email_sent; deadlines/calendar cast-then-check JSON crashes -> Zod Convention - ~44 legacy routes converted to withRouteContext (MFA); added Zod validation, corrected status codes, console.* -> lib/logger Response shapes preserved for existing callers. ~110 new tests. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * feat(bookkeeping): save a booking as a reusable template from Bokför direkt Add a "Spara som mall" action to the manual booking dialog so users can capture a kontering they just worked out as a booking template — right where they figured out how something should be booked. - derive amount-parameterised template lines from the concrete booking (settlement = the non-VAT leg nearest the total, 26xx = a VAT line with its rate snapped to the nearest standard rate, the rest = business ratios; line labels come from the loaded BAS chart) - extract the shared TemplateForm out of BookingTemplatesPanel so the booking dialog reuses the same editor, live preview and convertibility hints instead of duplicating them - save via the existing POST /api/settings/booking-templates endpoint Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(bokslut): render arsredovisning RR/BR at ÅRL post level — no kontonummer Bolagsverket rejected a user's filed årsredovisning with "Balansräkning och resultaträkning ska inte innehålla kontonummer": the PDF built every statement row as per-account "1930 Företagskonto" lines while the iXBRL filing path already aggregated to statutory posts, so the two artifacts diverged. The PDF statements now derive from the same K2 risbs mapping the iXBRL document uses (mapTrialBalancesToK2), via a new statement-rows.ts that emits post-level rows in uppställningsform order for both the K2 and K3 templates. Also fixed along the way: - Jämförelseår column (ÅRL 3:5 §) — previous-year trial balances now load and render; the old PDF had no comparatives at all. - mapping.warnings (unmapped accounts, RR ≠ 2099, obalans, reclass nudges) flow into ArsredovisningData.warnings so the wizard flags a non-fileable document before download. - Flerårsöversikt current/previous year overridden with the mapper's strict-3000–3799 Nettoomsattning, mirroring build-input's duplicate-fact rule, so the FB table ties to the RR. - FB eget kapital-table is post-level and drops obeskattade reserver (never eget kapital); K3 equity-changes statement uses real prior-year opening balances with derived utdelning/nyemission residuals that tie the roll-forward exactly to booked UB. - build-input dedupes warnings now that the PDF path runs the same mapping. Regression test asserts no RR/BR label ever contains a four-digit account number again. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(reports): diagnose untransferred prior-year results behind balance-sheet differens Prod incident (97 kr): a multi-year SIE migration lacked one year's omforing av arets resultat; the residual corrupted every later derived opening balance and Balansrakningen showed a bare "Differens: 97 kr" with no explanation. Continuity checking cannot catch this failure mode (prior-year UB and derived IB match per-account by construction) - the invariant that actually breaks is per-year P&L = 0 for all non-latest years. - lib/reports/imbalance-diagnosis.ts: shared detector (findUntransferredResults + buildImbalanceDiagnosis) - Balansrakning/Balansrapport attach imbalance_diagnosis when unbalanced, naming the exact culprit years; rendered in web views + PDF; MCP gnubok_get_balance_sheet inherits the field via spread - SIE import: parse-time warning when a completed year's vouchers leave a P&L residual, plus a post-import DB walk surfacing culprits as warnings and structured details.untransferredResults; the Arcim migration workspace previously dropped result.warnings entirely and now renders them - opening-balance/correct: pre-flight the company lock date and return 409 OB_COMPANY_LOCK_DATE (retryable: false, lock date interpolated in the client message) instead of the retryable 500 that invited blind retries; catch-path maps a raced trigger rejection to the same code Diagnosis runs only on unbalanced paths (zero cost when healthy) and never fails the report or the import. No migration, nothing persisted. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix: production error remediation — FX rates, deadlines, log levels, correction relink Batch of fixes for recurring Vercel runtime errors: - Riksbanken FX rates: persistent read-through cache (exchange_rates table), one retry honoring Retry-After on 429/5xx, bounded ingest concurrency, and an honest fallback — most recent cached observation or null, never a hardcoded rate silently booked into amount_sek. Unrated transactions stay repairable via refresh-exchange-rate. - Tax deadline regeneration inserts replacement rows before deleting the superseded set, so a failed insert no longer wipes a company's deadlines (the 23502 user_id regression did exactly that). Migration makes deadlines.user_id nullable for system-generated rows. - Route wrappers + errorResponse log 4xx outcomes at warn so only genuine 5xx reach Vercel's runtime-error clustering; client-supplied /api/log telemetry demoted to warn as well. - application/json documents (raw PSD2 responses archived per BFL) validate as parseable JSON with object/array root instead of always failing the magic-byte check. - correctEntry surfaces document-relink failures to callers, and the BFL document-immutability trigger now allows relinking underlag from a reversed entry to its correction (migration + pg test). - Middleware clears stale session cookies on /api requests too, using scope 'local' so cleanup doesn't re-trigger the failed token refresh. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(skatteverket): persist token health and stop retrying dead consents Terminal auth errors (SESSION_EXPIRED, REFRESH_EXHAUSTED, MISSING_SCOPE, TOKEN_CORRUPTED) mark the token row needs_reconsent with the error code and timestamp — SKV per-flow refresh tokens live 65 minutes, so once expired nothing recovers without a fresh BankID consent. The AGI kvittens and skattekonto sync crons skip flagged connections instead of failing every night, and the settings panel prompts for re-consent proactively. A successful reconnect resets the row to active. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(banking): allocate distinct BAS ledger slots for PSD2 mirror accounts A bank returning N same-currency accounts used to map them all onto the currency default (1930/1932/1933/1934), tripping the UNIQUE (company_id, ledger_account) constraint per-account — swallowed errors left accounts silently unmirrored. allocatePsd2LedgerAccount now hands out the currency default first, then free 1931–1959 sub-account slots, skipping slots held by any existing row. - Callback persists allocations to accounts_data so the picker pre-fills reality; reconnect reuses previously mirrored ledgers instead of re-deriving (a user remap to 1935 survives). - Selection save resolves effective ledgers up front and rejects duplicates or cross-connection conflicts with a 400 instead of silently skipping the mirror. - Bank error codes + psu_type are forwarded to the settings page for every OAuth error, keying the Handelsbanken corporate fullmakt guidance. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(agent): stage exact journal lines on categorization previews Categorization previews only carried debit/credit accounts, the GROSS amount, and separate VAT rows — read together that looks like an unbalanced 'gross on cost account + VAT debit' entry, and it misled both users and agents into rejecting correct proposals. The MCP preview and the pending-operation PATCH now materialize the exact lines the commit executor will post (net cost line, VAT line, gross bank line, SEK) via buildTransactionEntryLines, and PATCH re-derives them from the new mapping instead of spreading stale staged lines. ApprovalCard and /pending render the verifikat lines, falling back to the legacy summary only for operations staged before this fix. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * feat(bookkeeping): prune unused imported accounts from the chart SIE imports routinely bring in hundreds of accounts that were never used and clutter the kontoplan. New account_usage_counts RPC (one grouped query instead of a count per account) backs GET /api/bookkeeping/accounts/usage, and POST /api/bookkeeping/accounts/prune deletes zero-usage accounts — dry-run first, then an explicit account list capped at 2000. Accounts with journal lines are skipped, never deleted. The chart manager shows a usage column and a prune dialog grouping custom accounts vs unused BAS-seeded ones. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * feat(api): carry dimensions through v1 invoice and supplier-invoice surfaces Credit-note creation now copies default_dimensions and per-line dimensions from the original, so the reversing journal entry nets against the same dimension cells instead of dropping them. List/detail responses expose the dimension fields, and the OpenAPI spec snapshot follows. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * perf: batch serial Supabase round-trips on hot dashboard paths Every dashboard render pays the layout's query chain, so serialized awaits are direct wall-clock: the layout, chat conversation, invoice detail, supplier detail, select-company, and agent-onboarding pages now run their independent lookups in parallel batches, and getCompanyCapabilities folds its disabled-config read into the same round-trip. JournalEntryList hydrates the saved fiscal-year scope optimistically instead of serializing the first entries fetch behind the fiscal-periods request. The supplier detail page filters invoices server-side via a new supplier_id query param instead of fetching the whole company ledger, and the invoice editor (with its framer-motion dependency) lazy-loads so it stops shipping with the invoice list bundle. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * feat(salary): one-click runs, payslip delivery, payments settings, run cockpit Salary P1 batch, driving the 20-click flow toward 3 clicks: - One-click 'Starta lönekörning': POST /api/salary/runs accepts an empty body and resolves defaults server-side — period follows the latest non-corrected run, payment date from the new salary_pay_day setting, series from the per-source-type map. The separate /salary/runs/new page is gone. - Run detail page rebuilt as a step-railed cockpit (progress rail, KPI cards, employee ledger, journal preview) on a deliberately wider canvas; components extracted to components/salary/run/. - Payslip delivery: tokenized public payslip pages (/payslip/[token], backed by salary_payslip_links) plus per-employee email send with PDF — employees need no account, and the middleware exempts the route from auth redirects. - Payments settings: salary pay day, default bank, and pain.001 vs Bankgirot Lön format with per-bank upload instructions and an LB sunset warning (banks retire LB during 2026). - AGI panel: full submission status flows (stale drafts, signing links, kvittens polling, error reports); tax payment panel with skattekonto shortcut and mark-as-paid. - Salary calendar bulk editing, employee benefits/tax-card polish, municipality tax-table lookup improvements. messages/sv+en also carry the strings for the account-prune, skatteverket-reconsent, and banking surfaces committed just before this. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * chore: adopt Next 16 proxy.ts convention + repo housekeeping - Rename middleware.ts to proxy.ts with the proxy() export (Next 16 renamed the middleware convention; behavior unchanged). - Exclude dev_docs/ from tsconfig so stray snippets in planning docs don't break the build type-check. - Ratchet antipatterns-baseline down (raw-route-auth 165 → 119) to lock in the withRouteContext migration from 5cfd2b76. - template-library uses roundOre() instead of inline rounding. - database.md: drop account_balances from the key-tables list. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(bookkeeping): robust service-role detection in correction document relink relink_documents_to_correction() keyed its service-role branch on auth.role(), which reads the singular request.jwt.claim.role GUC that PostgREST v10+ and the pg-real harness no longer populate. Genuine service-role callers (pending-ops executor / MCP approve) landed in the auth gate and could not relink underlag. Read the role from the request.jwt.claims JSON directly, mirroring the canonical link_voucher_rpcs_tenant_guard convention. Validated on staging. Also: harden the salary run page's error paths (res.json().catch) against non-JSON error bodies, and roll back the pg-real service-role case in finally so an aborted transaction cannot poison a pooled connection for the next test. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(documents): restore journal_entry_line_id link durability (BFL 7 kap) Migration 20260704103000 rewrote enforce_document_journal_entry_immutability to guard journal_entry_id but left journal_entry_line_id to the metadata trigger, which exempts draft-linked docs -- and the entry-level trigger only fired on UPDATE OF journal_entry_id, so a line-id-only UPDATE never invoked it at all. That let a set journal_entry_line_id be cleared to NULL, breaking the "link durable from first set" invariant (document-immutability.pg regression). Widen the trigger to fire on journal_entry_line_id too and guard it with the same uuid-durability rule as journal_entry_id (setting NULL -> uuid stays allowed; clearing/re-pointing a set value is blocked, status-independent). The correction-relink GUC path, which legitimately clears line_id when moving underlag to the posted correction, stays exempt. Validated on staging. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Signed-off-by: Emil <emilmattsson14@gmail.com> Co-authored-by: Claude Fable 5 <noreply@anthropic.com> |