Broad sweep across the whole codebase:
- Blade views (39 files, 315 wraps): tag-text and title/placeholder/alt
attributes wrapped with {{ __() }}. Excludes scripts, styles, @php,
@verbatim, {{ }}, {!! !!}, comments to avoid touching interpolations.
- PHP (54 files, 179 wraps): array 'key' => 'RO value' patterns and
list items with diacritics wrapped with __(). Reverted __() inside
const arrays (PHP disallows non-constant expressions).
- Added 269 new keys to lang/{ru,en}.json (identity fallback for
unknowns → 161 human RU + 163 EN translations added for the most
common enums, stages, roles, statuses, payment methods, vehicle
categories, warehouse, portal, form actions.
Missing translations fall back to RO so the UI never breaks. All 306
tests pass; view cache compiles cleanly.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Rate limiting:
- Shop POST endpoints get per-IP throttles with distinct prefixes so login,
register, password-email, and password-reset have separate buckets:
login/register/pw-reset = 5/min, pw-email = 3/min
- OcrInvoiceService gates per-tenant via RateLimiter (30/hour) so a runaway
uploader can't burn Claude Vision spend
Health monitor (poor-man's monitoring):
- HealthCheckCommand probes DB (SELECT 1), cache write/read, public storage
write/read, and most-recent backup age. On any failure, pushes a Telegram
alert via HEALTH_ALERT_BOT_TOKEN/HEALTH_ALERT_CHAT_ID. Dedups identical
failures within a 30-min window via cache.
- Scheduled every 10 min. Pair with external uptime monitoring (UptimeRobot,
Better Stack hitting /up) for total-outage coverage.
- .env.example documents the two new env vars.
VAPID secret hygiene:
- credentials.md no longer stores the VAPID_PRIVATE_KEY; the source of truth
is the Coolify env on the autocrm app. Doc points to where to read it
(UI or API). Mitigates accidental git leak.
Tests (4 new):
- shop login throttles after 5 attempts (6th = 429); register throttle is
independent of login (separate prefix); health command runs clean; dedup
cache path exercised
Full suite: 138 passed.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Upload an invoice photo → Claude extracts {supplier_name, date, currency,
items, total} as JSON → auto-create a draft Purchase + PurchaseItems →
redirect to edit so the user reviews before confirming/receiving.
OcrInvoiceService:
- Validates supported MIME (jpg/png/webp/gif)
- Reads tenant Claude key (settings.ai.claude_key) — friendly error if missing
- Calls /v1/messages with image content block + structured-output system prompt
- Tolerant parser: strips ```json fences, falls back to first {…} block
- normalize(): computes per-item total when absent, fills overall total
- All return shapes: {ok:bool, data?, error?, raw?, tokens?}
Filament:
- "Import factură (OCR)" header action on Purchases list
- Image file upload → service → matches Supplier by case-insensitive name
(notes the unmapped name if no match) → creates draft Purchase + items →
redirects to the Edit page
Tests (6 new):
- clean JSON parses; markdown fences stripped; malformed → graceful error;
missing key → friendly message + no HTTP; unsupported MIME rejected;
item total computed when missing
Full suite: 123 passed.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The Asistent AI chat can now query the tenant DB directly through Claude
tool use. AiToolExecutor exposes 5 read-only tools (search_clients,
get_vehicle, find_parts, recent_workorders, low_stock_parts) all scoped
to the current tenant via BelongsToTenant.
AiAssistantService::callClaude loops on stop_reason=tool_use up to 5 rounds:
- normalize message history to content blocks
- send `tools` definitions + messages to Anthropic API
- on tool_use: execute each tool, append tool_results as user turn, recall
- on end_turn: collect text + cumulative token counts + tool-call audit in
AiMessage.meta.tools
Single-shot helpers (suggestDiagnosis, suggestPrice, vinRecommendations,
suggestParts) are unchanged — only the conversational chat gets tool-use.
Tests (3 new):
- two-round tool_use → execute → final text; verify 5 tools sent both rounds;
cumulative tokens
- executor finds part by brand
- unknown tool name returns error blob
Full suite: 109 passed.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Schema:
- ai_chats: company_id, user_id, title, provider; index pe activitate
- ai_messages: role (system/user/assistant), content, meta JSON (tokens, latency, model)
Service AiAssistantService (multi-provider):
- ask($chat, $message): persistă mesajul user, build system prompt cu context
tenant (statistici clienți/mașini/cereri/datorii), apelează API-ul providerului,
persistă răspunsul cu meta (tokens, latency)
- callClaude: api.anthropic.com/v1/messages cu claude-sonnet-4-5
- callOpenAI: api.openai.com/v1/chat/completions cu gpt-4o-mini
- callGemini: generativelanguage.googleapis.com cu gemini-1.5-flash
- Try/catch pe toate; eroare devine mesaj asistent fără să crape
System prompt include:
- Numele și orașul companiei
- Statistici curente (clienți, mașini, cereri noi, fișe active, datorii)
- Limita stricta: NU inventează date
Custom Filament Page /app/ai-assistant (group Analiză):
- Sidebar stâng: listă conversații (last 20), buton 'Nouă' + delete cu confirm
- Main: bubble chat (user dreapta albastru, asistent stânga gri)
- Meta jos pe răspuns: provider · latency · tokens
- Empty state friendly cu instrucțiuni configurare
- Loading indicator (3 dots animate) când AI răspunde
- Auto-scroll la mesaj nou
- Enter trimite, Shift+Enter newline
- Auto-titlu chat din primul mesaj user (60 chars)
Settings page extins cu secțiune 'Asistent AI':
- Provider implicit (claude/gpt/gemini)
- 3 chei API (password fields, revealable)
- Key-urile salvate în companies.settings.ai (per tenant, izolat)
Vasyka