Merge pull request 'feat: Migrate Gemini SDK to google-genai (#231)' (#236) from issue-231-migrate-gemini-sdk-google-genai into main

Reviewed-on: #236

.ai/workflow-contract.json

@@ -52,7 +52,8 @@
       "ONE PR for the parent issue. The PR closes the parent and all sub-issues.",
       "Commits reference the specific sub-issue index they implement.",
       "Sub-issues should be small enough to fit in a single AI context window.",
-      "Plan milestones map 1:1 to sub-issues."
+      "Plan milestones map 1:1 to sub-issues.",
+      "Each sub-issue receives its own plan comment with duplicated shared context. An agent must be able to execute from the sub-issue alone."
     ],
     "examples": {
       "parent": "#105 'feat: Add Grafana dashboards and alerting'",
@@ -103,8 +104,9 @@
     "[SKILL] Problem Analysis if complex problem.",
     "[SKILL] Decision Critic if uncertain approach.",
     "If multi-file feature (3+ files): decompose into sub-issues per sub_issues rules. Each sub-issue = one plan milestone.",
-    "[SKILL] Planner writes plan as parent issue comment. Plan milestones map 1:1 to sub-issues.",
+    "[SKILL] Planner writes plan summary as parent issue comment: shared context + milestone index linking each milestone to its sub-issue. M5 (doc-sync) stays on parent if no sub-issue exists.",
-    "[SKILL] Plan review cycle: QR plan-completeness -> TW plan-scrub -> QR plan-code -> QR plan-docs.",
+    "[SKILL] Planner posts each milestone's self-contained implementation plan as a comment on the corresponding sub-issue. Each sub-issue plan duplicates relevant shared context (API maps, state changes, auth, error handling, risk) so an agent can execute from the sub-issue alone without reading the parent.",
+    "[SKILL] Plan review cycle: QR plan-completeness -> TW plan-scrub -> QR plan-code -> QR plan-docs. Distribute milestone-specific review findings to sub-issue plan comments.",
     "Create ONE branch issue-{parent_index}-{slug} from main.",
     "[SKILL] Planner executes plan, delegates to Developer per milestone/sub-issue.",
     "[SKILL] QR post-implementation per milestone (results in parent issue comment).",
@@ -123,7 +125,7 @@
       "execution_review": ["QR post-implementation per milestone"],
       "final_review": ["Quality Agent RULE 0/1/2"]
     },
-    "plan_storage": "gitea_issue_comments",
+    "plan_storage": "gitea_issue_comments: summary on parent issue, milestone detail on sub-issues",
     "tracking_storage": "gitea_issue_comments",
     "issue_comment_operations": {
       "create_comment": "mcp__gitea-mcp__create_issue_comment",

.env.example

@@ -0,0 +1,36 @@
+# MotoVaultPro Environment Configuration
+# Copy to .env and fill in environment-specific values
+# Generated .env files should NOT be committed to version control
+#
+# Local dev: No .env needed -- base docker-compose.yml defaults are sandbox values
+# Staging/Production: CI/CD generates .env from Gitea variables + generate-log-config.sh
+
+# ===========================================
+# Stripe Price IDs (environment-specific)
+# ===========================================
+# Sandbox defaults used for local development
+STRIPE_PRO_MONTHLY_PRICE_ID=price_1T1ZHMJXoKkh5RcKwKSSGIlR
+STRIPE_PRO_YEARLY_PRICE_ID=price_1T1ZHnJXoKkh5RcKWlG2MPpX
+STRIPE_ENTERPRISE_MONTHLY_PRICE_ID=price_1T1ZIBJXoKkh5RcKu2jyhqBN
+STRIPE_ENTERPRISE_YEARLY_PRICE_ID=price_1T1ZIQJXoKkh5RcK34YXiJQm
+
+# ===========================================
+# Stripe Publishable Key (baked into frontend at build time)
+# ===========================================
+# VITE_STRIPE_PUBLISHABLE_KEY=pk_test_...
+
+# ===========================================
+# Log Levels (generated by scripts/ci/generate-log-config.sh)
+# ===========================================
+# Run: ./scripts/ci/generate-log-config.sh DEBUG >> .env
+#
+# BACKEND_LOG_LEVEL=debug
+# TRAEFIK_LOG_LEVEL=DEBUG
+# POSTGRES_LOG_STATEMENT=all
+# POSTGRES_LOG_MIN_DURATION=0
+# REDIS_LOGLEVEL=debug
+
+# ===========================================
+# Grafana
+# ===========================================
+# GRAFANA_ADMIN_PASSWORD=admin

.gitea/workflows/production.yaml

@@ -99,6 +99,7 @@ jobs:
             docker-compose.yml
             docker-compose.blue-green.yml
             docker-compose.prod.yml
+            .env.example
           sparse-checkout-cone-mode: false
           fetch-depth: 1
 
@@ -115,11 +116,20 @@ jobs:
           mkdir -p "$DEPLOY_PATH/secrets/app"
           cp "$GITHUB_WORKSPACE/secrets/app/google-wif-config.json" "$DEPLOY_PATH/secrets/app/"
 
-      - name: Generate logging configuration
+      - name: Generate environment configuration
         run: |
           cd "$DEPLOY_PATH"
+          {
+            echo "# Generated by CI/CD - DO NOT EDIT"
+            echo "STRIPE_PRO_MONTHLY_PRICE_ID=${{ vars.STRIPE_PRO_MONTHLY_PRICE_ID }}"
+            echo "STRIPE_PRO_YEARLY_PRICE_ID=${{ vars.STRIPE_PRO_YEARLY_PRICE_ID }}"
+            echo "STRIPE_ENTERPRISE_MONTHLY_PRICE_ID=${{ vars.STRIPE_ENTERPRISE_MONTHLY_PRICE_ID }}"
+            echo "STRIPE_ENTERPRISE_YEARLY_PRICE_ID=${{ vars.STRIPE_ENTERPRISE_YEARLY_PRICE_ID }}"
+            echo "VITE_STRIPE_PUBLISHABLE_KEY=${{ vars.VITE_STRIPE_PUBLISHABLE_KEY }}"
+            echo "GRAFANA_ADMIN_PASSWORD=${{ secrets.GRAFANA_ADMIN_PASSWORD }}"
+          } > .env
           chmod +x scripts/ci/generate-log-config.sh
-          ./scripts/ci/generate-log-config.sh "$LOG_LEVEL"
+          ./scripts/ci/generate-log-config.sh "$LOG_LEVEL" >> .env
 
       - name: Login to registry
         run: |

.gitea/workflows/staging.yaml

@@ -124,11 +124,20 @@ jobs:
           mkdir -p "$DEPLOY_PATH/secrets/app"
           cp "$GITHUB_WORKSPACE/secrets/app/google-wif-config.json" "$DEPLOY_PATH/secrets/app/"
 
-      - name: Generate logging configuration
+      - name: Generate environment configuration
         run: |
           cd "$DEPLOY_PATH"
+          {
+            echo "# Generated by CI/CD - DO NOT EDIT"
+            echo "STRIPE_PRO_MONTHLY_PRICE_ID=${{ vars.STRIPE_PRO_MONTHLY_PRICE_ID }}"
+            echo "STRIPE_PRO_YEARLY_PRICE_ID=${{ vars.STRIPE_PRO_YEARLY_PRICE_ID }}"
+            echo "STRIPE_ENTERPRISE_MONTHLY_PRICE_ID=${{ vars.STRIPE_ENTERPRISE_MONTHLY_PRICE_ID }}"
+            echo "STRIPE_ENTERPRISE_YEARLY_PRICE_ID=${{ vars.STRIPE_ENTERPRISE_YEARLY_PRICE_ID }}"
+            echo "VITE_STRIPE_PUBLISHABLE_KEY=${{ vars.VITE_STRIPE_PUBLISHABLE_KEY }}"
+            echo "GRAFANA_ADMIN_PASSWORD=${{ secrets.GRAFANA_ADMIN_PASSWORD }}"
+          } > .env
           chmod +x scripts/ci/generate-log-config.sh
-          ./scripts/ci/generate-log-config.sh "$LOG_LEVEL"
+          ./scripts/ci/generate-log-config.sh "$LOG_LEVEL" >> .env
 
       - name: Login to registry
         run: |

backend/src/core/config/config-loader.ts

@@ -41,14 +41,6 @@ const configSchema = z.object({
     audience: z.string(),
   }),
 
-  // External APIs configuration (optional)
-  external: z.object({
-    vpic: z.object({
-      url: z.string(),
-      timeout: z.string(),
-    }).optional(),
-  }).optional(),
-
   // Service configuration
   service: z.object({
     name: z.string(),

backend/src/core/config/feature-tiers.ts

@@ -29,7 +29,7 @@ export const FEATURE_TIERS: Record<string, FeatureConfig> = {
   'vehicle.vinDecode': {
     minTier: 'pro',
     name: 'VIN Decode',
-    upgradePrompt: 'Upgrade to Pro to automatically decode VIN and populate vehicle details from the NHTSA database.',
+    upgradePrompt: 'Upgrade to Pro to automatically decode VIN and populate vehicle details from the vehicle database.',
   },
   'fuelLog.receiptScan': {
     minTier: 'pro',

backend/src/features/notifications/README.md

@@ -97,7 +97,7 @@ Templates use `{{variableName}}` syntax for variable substitution.
 
 ### Environment Variables
 - `RESEND_API_KEY` - Resend API key (required, stored in secrets)
-- `FROM_EMAIL` - Sender email address (default: noreply@motovaultpro.com)
+- `FROM_EMAIL` - Sender email address (default: hello@notify.motovaultpro.com)
 
 ### Email Delivery
 - Uses Resend API for transactional emails

backend/src/features/notifications/domain/email-layout/base-layout.ts

@@ -7,7 +7,7 @@
 import { EMAIL_STYLES } from './email-styles';
 
 // External logo URL - hosted on GitHub for reliability
-const LOGO_URL = 'https://raw.githubusercontent.com/ericgullickson/images/c58b0e4773e8395b532f97f6ab529e38ea4dc8be/motovaultpro-auth0-small.png';
+const LOGO_URL = 'https://motovaultpro.com/images/logos/motovaultpro-auth0-small.png';
 
 /**
  * Renders the complete HTML email layout with branding
@@ -65,10 +65,10 @@ export function renderEmailLayout(content: string): string {
                       <a href="https://motovaultpro.com" style="${EMAIL_STYLES.footerLink}" target="_blank">Login to MotoVaultPro</a>
                     </p>
                     <p style="${EMAIL_STYLES.footerText}">
-                      <a href="{{unsubscribeUrl}}" style="${EMAIL_STYLES.footerLink}" target="_blank">Manage Email Preferences</a>
+                      <a href="https://motovaultpro.com/settings" style="${EMAIL_STYLES.footerLink}" target="_blank">Manage Email Preferences</a>
                     </p>
                     <p style="${EMAIL_STYLES.copyright}">
-                      &copy; {new Date().getFullYear()} MotoVaultPro. All rights reserved.
+                      &copy; ${new Date().getFullYear()} MotoVaultPro. All rights reserved.
                     </p>
                   </td>
                 </tr>

backend/src/features/notifications/domain/email.service.ts

@@ -16,7 +16,7 @@ export class EmailService {
     }
 
     this.resend = new Resend(apiKey);
-    this.fromEmail = process.env['FROM_EMAIL'] || 'noreply@motovaultpro.com';
+    this.fromEmail = process.env['FROM_EMAIL'] || 'hello@notify.motovaultpro.com';
   }
 
   /**
@@ -33,6 +33,10 @@ export class EmailService {
         to,
         subject,
         html,
+        headers: {
+          'List-Unsubscribe': '<https://motovaultpro.com/settings>',
+          'List-Unsubscribe-Post': 'List-Unsubscribe=One-Click',
+        },
       });
     } catch (error) {
       const errorMessage = error instanceof Error ? error.message : 'Unknown error';

backend/src/features/ocr/CLAUDE.md

@@ -37,7 +37,7 @@ Backend proxy for the Python OCR microservice. Handles authentication, tier gati
 
 | File | What | When to read |
 | ---- | ---- | ------------ |
-| `ocr-client.ts` | HTTP client to mvp-ocr Python service (extract, extractVin, extractReceipt, submitJob, submitManualJob, getJobStatus, isHealthy) | OCR service communication, error handling |
+| `ocr-client.ts` | HTTP client to mvp-ocr Python service (extract, extractVin, extractReceipt, decodeVin, submitJob, submitManualJob, getJobStatus, isHealthy) | OCR service communication, error handling |
 
 ## tests/
 

backend/src/features/ocr/domain/ocr.types.ts

@@ -131,3 +131,21 @@ export interface ManualJobResponse {
   result?: ManualExtractionResult;
   error?: string;
 }
+
+/** Response from VIN decode via Gemini (OCR service) */
+export interface VinDecodeResponse {
+  success: boolean;
+  vin: string;
+  year: number | null;
+  make: string | null;
+  model: string | null;
+  trimLevel: string | null;
+  bodyType: string | null;
+  driveType: string | null;
+  fuelType: string | null;
+  engine: string | null;
+  transmission: string | null;
+  confidence: number;
+  processingTimeMs: number;
+  error: string | null;
+}

backend/src/features/ocr/external/ocr-client.ts

@@ -2,7 +2,7 @@
  * @ai-summary HTTP client for OCR service communication
  */
 import { logger } from '../../../core/logging/logger';
-import type { JobResponse, ManualJobResponse, OcrResponse, ReceiptExtractionResponse, VinExtractionResponse } from '../domain/ocr.types';
+import type { JobResponse, ManualJobResponse, OcrResponse, ReceiptExtractionResponse, VinDecodeResponse, VinExtractionResponse } from '../domain/ocr.types';
 
 /** OCR service configuration */
 const OCR_SERVICE_URL = process.env.OCR_SERVICE_URL || 'http://mvp-ocr:8000';
@@ -373,6 +373,55 @@ export class OcrClient {
     return result;
   }
 
+  /**
+   * Decode a VIN string into structured vehicle data via Gemini.
+   *
+   * Unlike other OCR methods, this sends JSON (not multipart) because
+   * VIN decode has no file upload.
+   *
+   * @param vin - 17-character Vehicle Identification Number
+   * @returns Structured vehicle data from Gemini decode
+   */
+  async decodeVin(vin: string): Promise<VinDecodeResponse> {
+    const url = `${this.baseUrl}/decode/vin`;
+
+    logger.info('OCR VIN decode request', {
+      operation: 'ocr.client.decodeVin',
+      url,
+      vin,
+    });
+
+    const response = await this.fetchWithTimeout(url, {
+      method: 'POST',
+      headers: { 'Content-Type': 'application/json' },
+      body: JSON.stringify({ vin }),
+    });
+
+    if (!response.ok) {
+      const errorText = await response.text();
+      logger.error('OCR VIN decode failed', {
+        operation: 'ocr.client.decodeVin.error',
+        status: response.status,
+        error: errorText,
+      });
+      const err: any = new Error(`OCR service error: ${response.status} - ${errorText}`);
+      err.statusCode = response.status;
+      throw err;
+    }
+
+    const result = (await response.json()) as VinDecodeResponse;
+
+    logger.info('OCR VIN decode completed', {
+      operation: 'ocr.client.decodeVin.success',
+      success: result.success,
+      vin: result.vin,
+      confidence: result.confidence,
+      processingTimeMs: result.processingTimeMs,
+    });
+
+    return result;
+  }
+
   /**
    * Check if the OCR service is healthy.
    *

backend/src/features/platform/README.md

@@ -117,7 +117,7 @@ platform/
 When implemented, VIN decoding will use:
 1. **Cache First**: Check Redis (7-day TTL for success, 1-hour for failures)
 2. **PostgreSQL**: Database function for high-confidence decode
-3. **vPIC Fallback**: NHTSA vPIC API with circuit breaker protection
+3. **OCR Service Fallback**: Gemini VIN decode via OCR service
 4. **Graceful Degradation**: Return meaningful errors when all sources fail
 
 ### Database Schema
@@ -164,7 +164,7 @@ When VIN decoding is implemented:
 
 ### External APIs (Planned/Future)
 When VIN decoding is implemented:
-- **NHTSA vPIC**: https://vpic.nhtsa.dot.gov/api (VIN decoding fallback)
+- **OCR Service**: Gemini VIN decode via mvp-ocr (VIN decoding fallback)
 
 ### Database Tables
 - **vehicle_options** - Hierarchical vehicle data (years, makes, models, trims, engines, transmissions)
@@ -269,7 +269,7 @@ npm run lint
 ## Future Considerations
 
 ### Planned Features
-- VIN decoding endpoint with PostgreSQL + vPIC fallback
+- VIN decoding endpoint with PostgreSQL + Gemini/OCR service fallback
 - Circuit breaker pattern for external API resilience
 
 ### Potential Enhancements

backend/src/features/platform/models/responses.ts

@@ -61,19 +61,3 @@ export interface VINDecodeResponse {
   error?: string;
 }
 
-/**
- * vPIC API response structure (NHTSA)
- */
-export interface VPICVariable {
-  Variable: string;
-  Value: string | null;
-  ValueId: string | null;
-  VariableId: number;
-}
-
-export interface VPICResponse {
-  Count: number;
-  Message: string;
-  SearchCriteria: string;
-  Results: VPICVariable[];
-}

backend/src/features/subscriptions/domain/subscriptions.service.ts

@@ -570,11 +570,13 @@ export class SubscriptionsService {
     }
 
     // Update subscription with Stripe subscription ID
+    // Period dates moved from subscription to items in API 2025-03-31.basil
+    const item = stripeSubscription.items?.data?.[0];
     await this.repository.update(subscription.id, {
       stripeSubscriptionId: stripeSubscription.id,
       status: this.mapStripeStatus(stripeSubscription.status),
-      currentPeriodStart: new Date(stripeSubscription.current_period_start * 1000),
+      currentPeriodStart: new Date((item?.current_period_start ?? 0) * 1000),
-      currentPeriodEnd: new Date(stripeSubscription.current_period_end * 1000),
+      currentPeriodEnd: new Date((item?.current_period_end ?? 0) * 1000),
     });
 
     // Log event
@@ -608,11 +610,13 @@ export class SubscriptionsService {
     const tier = this.determineTierFromStripeSubscription(stripeSubscription);
 
     // Update subscription
+    // Period dates moved from subscription to items in API 2025-03-31.basil
+    const item = stripeSubscription.items?.data?.[0];
     const updateData: UpdateSubscriptionData = {
       status: this.mapStripeStatus(stripeSubscription.status),
       tier,
-      currentPeriodStart: new Date(stripeSubscription.current_period_start * 1000),
+      currentPeriodStart: new Date((item?.current_period_start ?? 0) * 1000),
-      currentPeriodEnd: new Date(stripeSubscription.current_period_end * 1000),
+      currentPeriodEnd: new Date((item?.current_period_end ?? 0) * 1000),
       cancelAtPeriodEnd: stripeSubscription.cancel_at_period_end || false,
     };
 
@@ -849,6 +853,7 @@ export class SubscriptionsService {
     switch (stripeStatus) {
       case 'active':
       case 'trialing':
+      case 'incomplete':
         return 'active';
       case 'past_due':
         return 'past_due';

backend/src/features/subscriptions/external/stripe/stripe.client.ts

@@ -75,10 +75,18 @@ export class StripeClient {
     try {
       logger.info('Creating Stripe subscription', { customerId, priceId, paymentMethodId });
 
+      // Attach payment method to customer before creating subscription
+      if (paymentMethodId) {
+        await this.stripe.paymentMethods.attach(paymentMethodId, {
+          customer: customerId,
+        });
+        logger.info('Payment method attached to customer', { customerId, paymentMethodId });
+      }
+
       const subscriptionParams: Stripe.SubscriptionCreateParams = {
         customer: customerId,
         items: [{ price: priceId }],
-        payment_behavior: 'default_incomplete',
+        payment_behavior: 'error_if_incomplete',
         payment_settings: {
           save_default_payment_method: 'on_subscription',
         },
@@ -93,13 +101,16 @@ export class StripeClient {
 
       logger.info('Stripe subscription created', { subscriptionId: subscription.id });
 
+      // Period dates moved from subscription to items in API 2025-03-31.basil
+      const item = subscription.items?.data?.[0];
+
       return {
         id: subscription.id,
         customer: subscription.customer as string,
         status: subscription.status as StripeSubscription['status'],
         items: subscription.items,
-        currentPeriodStart: (subscription as any).current_period_start || 0,
+        currentPeriodStart: item?.current_period_start ?? 0,
-        currentPeriodEnd: (subscription as any).current_period_end || 0,
+        currentPeriodEnd: item?.current_period_end ?? 0,
         cancelAtPeriodEnd: subscription.cancel_at_period_end,
         canceledAt: subscription.canceled_at || undefined,
         created: subscription.created,
@@ -140,13 +151,15 @@ export class StripeClient {
         logger.info('Stripe subscription canceled immediately', { subscriptionId });
       }
 
+      const item = subscription.items?.data?.[0];
+
       return {
         id: subscription.id,
         customer: subscription.customer as string,
         status: subscription.status as StripeSubscription['status'],
         items: subscription.items,
-        currentPeriodStart: (subscription as any).current_period_start || 0,
+        currentPeriodStart: item?.current_period_start ?? 0,
-        currentPeriodEnd: (subscription as any).current_period_end || 0,
+        currentPeriodEnd: item?.current_period_end ?? 0,
         cancelAtPeriodEnd: subscription.cancel_at_period_end,
         canceledAt: subscription.canceled_at || undefined,
         created: subscription.created,
@@ -286,14 +299,15 @@ export class StripeClient {
       logger.info('Retrieving Stripe subscription', { subscriptionId });
 
       const subscription = await this.stripe.subscriptions.retrieve(subscriptionId);
+      const item = subscription.items?.data?.[0];
 
       return {
         id: subscription.id,
         customer: subscription.customer as string,
         status: subscription.status as StripeSubscription['status'],
         items: subscription.items,
-        currentPeriodStart: (subscription as any).current_period_start || 0,
+        currentPeriodStart: item?.current_period_start ?? 0,
-        currentPeriodEnd: (subscription as any).current_period_end || 0,
+        currentPeriodEnd: item?.current_period_end ?? 0,
         cancelAtPeriodEnd: subscription.cancel_at_period_end,
         canceledAt: subscription.canceled_at || undefined,
         created: subscription.created,

backend/src/features/vehicles/CLAUDE.md

@@ -16,6 +16,6 @@
 | `data/` | Repository, database queries | Database operations |
 | `docs/` | Feature-specific documentation | Vehicle design details |
 | `events/` | Event handlers and emitters | Cross-feature event integration |
-| `external/` | External service integrations (NHTSA) | VIN decoding, third-party APIs |
+| `external/` | External service integrations | VIN decoding, third-party APIs |
 | `migrations/` | Database schema | Schema changes |
 | `tests/` | Unit and integration tests | Adding or modifying tests |

backend/src/features/vehicles/README.md

@@ -13,7 +13,7 @@ Primary entity for vehicle management consuming MVP Platform Vehicles Service. H
 - `DELETE /api/vehicles/:id` - Soft delete vehicle
 
 ### VIN Decoding (Pro/Enterprise Only)
-- `POST /api/vehicles/decode-vin` - Decode VIN using NHTSA vPIC API
+- `POST /api/vehicles/decode-vin` - Decode VIN using Gemini via OCR service
 
 ### Hierarchical Vehicle Dropdowns
 **Status**: Vehicles service now proxies the platform vehicle catalog to provide fully dynamic dropdowns. Each selection step filters the next list, ensuring only valid combinations are shown.
@@ -104,11 +104,7 @@ vehicles/
 ├── data/                  # Database layer
 │   └── vehicles.repository.ts
 ├── external/              # External service integrations
-│   ├── CLAUDE.md          # Integration pattern docs
+│   └── CLAUDE.md          # Integration pattern docs
-│   └── nhtsa/             # NHTSA vPIC API client
-│       ├── nhtsa.client.ts
-│       ├── nhtsa.types.ts
-│       └── index.ts
 ├── migrations/            # Feature schema
 │   └── 001_create_vehicles_tables.sql
 ├── tests/                 # All tests
@@ -121,14 +117,14 @@ vehicles/
 
 ## Key Features
 
-### 🔍 VIN Decoding (NHTSA vPIC API)
+### VIN Decoding (Gemini via OCR Service)
 - **Tier Gating**: Pro and Enterprise users only (`vehicle.vinDecode` feature key)
-- **NHTSA API**: Calls official NHTSA vPIC API for authoritative vehicle data
+- **Gemini**: Calls OCR service Gemini VIN decode for authoritative vehicle data
 - **Caching**: Results cached in `vin_cache` table (1-year TTL, VIN data is static)
 - **Validation**: 17-character VIN format, excludes I/O/Q characters
 - **Matching**: Case-insensitive exact match against dropdown options
 - **Confidence Levels**: High (exact match), Medium (normalized match), None (hint only)
-- **Timeout**: 5-second timeout for NHTSA API calls
+- **Timeout**: 5-second timeout for OCR service calls
 
 #### Decode VIN Request
 ```json
@@ -140,15 +136,15 @@ Authorization: Bearer <jwt>
 
 Response (200):
 {
-  "year": { "value": 2021, "nhtsaValue": "2021", "confidence": "high" },
+  "year": { "value": 2021, "decodedValue": "2021", "confidence": "high" },
-  "make": { "value": "Honda", "nhtsaValue": "HONDA", "confidence": "high" },
+  "make": { "value": "Honda", "decodedValue": "HONDA", "confidence": "high" },
-  "model": { "value": "Civic", "nhtsaValue": "Civic", "confidence": "high" },
+  "model": { "value": "Civic", "decodedValue": "Civic", "confidence": "high" },
-  "trimLevel": { "value": "EX", "nhtsaValue": "EX", "confidence": "high" },
+  "trimLevel": { "value": "EX", "decodedValue": "EX", "confidence": "high" },
-  "engine": { "value": null, "nhtsaValue": "2.0L L4 DOHC 16V", "confidence": "none" },
+  "engine": { "value": null, "decodedValue": "2.0L L4 DOHC 16V", "confidence": "none" },
-  "transmission": { "value": null, "nhtsaValue": "CVT", "confidence": "none" },
+  "transmission": { "value": null, "decodedValue": "CVT", "confidence": "none" },
-  "bodyType": { "value": null, "nhtsaValue": "Sedan", "confidence": "none" },
+  "bodyType": { "value": null, "decodedValue": "Sedan", "confidence": "none" },
-  "driveType": { "value": null, "nhtsaValue": "FWD", "confidence": "none" },
+  "driveType": { "value": null, "decodedValue": "FWD", "confidence": "none" },
-  "fuelType": { "value": null, "nhtsaValue": "Gasoline", "confidence": "none" }
+  "fuelType": { "value": null, "decodedValue": "Gasoline", "confidence": "none" }
 }
 
 Error (400 - Invalid VIN):
@@ -157,7 +153,7 @@ Error (400 - Invalid VIN):
 Error (403 - Tier Required):
 { "error": "TIER_REQUIRED", "requiredTier": "pro", "currentTier": "free", ... }
 
-Error (502 - NHTSA Failure):
+Error (502 - OCR Service Failure):
 { "error": "VIN_DECODE_FAILED", "message": "Unable to decode VIN from external service" }
 ```
 
@@ -230,7 +226,7 @@ Error (502 - NHTSA Failure):
 ## Testing
 
 ### Unit Tests
-- `vehicles.service.test.ts` - Business logic with mocked dependencies (VIN decode, caching, CRUD operations)
+- `vehicles.service.test.ts` - Business logic with mocked dependencies (VIN decode via OCR service mock, caching, CRUD operations)
 
 ### Integration Tests
 - `vehicles.integration.test.ts` - Complete API workflow with test database (create, read, update, delete vehicles)

backend/src/features/vehicles/api/vehicles.controller.ts

@@ -10,19 +10,18 @@ import { pool } from '../../../core/config/database';
 import { logger } from '../../../core/logging/logger';
 import { CreateVehicleBody, UpdateVehicleBody, VehicleParams } from '../domain/vehicles.types';
 import { getStorageService } from '../../../core/storage/storage.service';
-import { NHTSAClient, DecodeVinRequest } from '../external/nhtsa';
+import { ocrClient } from '../../ocr/external/ocr-client';
+import type { DecodeVinRequest } from '../domain/vehicles.types';
 import crypto from 'crypto';
 import FileType from 'file-type';
 import path from 'path';
 
 export class VehiclesController {
   private vehiclesService: VehiclesService;
-  private nhtsaClient: NHTSAClient;
 
   constructor() {
     const repository = new VehiclesRepository(pool);
     this.vehiclesService = new VehiclesService(repository, pool);
-    this.nhtsaClient = new NHTSAClient(pool);
   }
 
   async getUserVehicles(request: FastifyRequest, reply: FastifyReply) {
@@ -378,7 +377,7 @@ export class VehiclesController {
   }
 
   /**
-   * Decode VIN using NHTSA vPIC API
+   * Decode VIN using OCR service (Gemini)
    * POST /api/vehicles/decode-vin
    * Requires Pro or Enterprise tier
    */
@@ -395,26 +394,39 @@ export class VehiclesController {
         });
       }
 
-      logger.info('VIN decode requested', { userId, vin: vin.substring(0, 6) + '...' });
+      // Validate VIN format
+      const sanitizedVin = vin.trim().toUpperCase();
+      const VIN_REGEX = /^[A-HJ-NPR-Z0-9]{17}$/;
+      if (!VIN_REGEX.test(sanitizedVin)) {
+        return reply.code(400).send({
+          error: 'INVALID_VIN',
+          message: 'Invalid VIN format. VIN must be exactly 17 characters and contain only letters (except I, O, Q) and numbers.'
+        });
+      }
 
-      // Validate and decode VIN
+      logger.info('VIN decode requested', { userId, vin: sanitizedVin.substring(0, 6) + '...' });
-      const response = await this.nhtsaClient.decodeVin(vin);
 
-      // Extract and map fields from NHTSA response
+      // Call OCR service for VIN decode
-      const decodedData = await this.vehiclesService.mapNHTSAResponse(response);
+      const response = await ocrClient.decodeVin(sanitizedVin);
+
+      // Map response to decoded vehicle data with dropdown matching
+      const decodedData = await this.vehiclesService.mapVinDecodeResponse(response);
 
       logger.info('VIN decode successful', {
         userId,
         hasYear: !!decodedData.year.value,
         hasMake: !!decodedData.make.value,
-        hasModel: !!decodedData.model.value
+        hasModel: !!decodedData.model.value,
+        hasTrim: !!decodedData.trimLevel.value,
+        hasEngine: !!decodedData.engine.value,
+        hasTransmission: !!decodedData.transmission.value,
       });
 
       return reply.code(200).send(decodedData);
     } catch (error: any) {
       logger.error('VIN decode failed', { error, userId });
 
-      // Handle validation errors
+      // Handle VIN validation errors
       if (error.message?.includes('Invalid VIN')) {
         return reply.code(400).send({
           error: 'INVALID_VIN',
@@ -422,16 +434,25 @@ export class VehiclesController {
         });
       }
 
-      // Handle timeout
+      // Handle OCR service errors by status code
-      if (error.message?.includes('timed out')) {
+      if (error.statusCode === 503 || error.statusCode === 422) {
-        return reply.code(504).send({
+        return reply.code(502).send({
-          error: 'VIN_DECODE_TIMEOUT',
+          error: 'VIN_DECODE_FAILED',
-          message: 'NHTSA API request timed out. Please try again.'
+          message: 'VIN decode service unavailable',
+          details: error.message
         });
       }
 
-      // Handle NHTSA API errors
+      // Handle timeout
-      if (error.message?.includes('NHTSA')) {
+      if (error.message?.includes('timed out') || error.message?.includes('aborted')) {
+        return reply.code(504).send({
+          error: 'VIN_DECODE_TIMEOUT',
+          message: 'VIN decode service timed out. Please try again.'
+        });
+      }
+
+      // Handle OCR service errors
+      if (error.message?.includes('OCR service error')) {
         return reply.code(502).send({
           error: 'VIN_DECODE_FAILED',
           message: 'Unable to decode VIN from external service',

backend/src/features/vehicles/api/vehicles.routes.ts

@@ -75,7 +75,7 @@ export const vehiclesRoutes: FastifyPluginAsync = async (
     handler: vehiclesController.getDropdownOptions.bind(vehiclesController)
   });
 
-  // POST /api/vehicles/decode-vin - Decode VIN using NHTSA vPIC API (Pro/Enterprise only)
+  // POST /api/vehicles/decode-vin - Decode VIN via OCR service (Pro/Enterprise only)
   fastify.post<{ Body: { vin: string } }>('/vehicles/decode-vin', {
     preHandler: [fastify.authenticate, fastify.requireTier({ featureKey: 'vehicle.vinDecode' })],
     handler: vehiclesController.decodeVin.bind(vehiclesController)

backend/src/features/vehicles/domain/vehicles.service.ts

@@ -1,6 +1,6 @@
 /**
  * @ai-summary Business logic for vehicles feature
- * @ai-context Handles VIN decoding, caching, and business rules
+ * @ai-context Handles VIN decoding and business rules
  */
 
 import { Pool } from 'pg';
@@ -24,7 +24,8 @@ import { isValidVIN, isValidPreModernVIN } from '../../../shared-minimal/utils/v
 import { normalizeMakeName, normalizeModelName } from './name-normalizer';
 import { getVehicleDataService, getPool } from '../../platform';
 import { auditLogService } from '../../audit-log';
-import { NHTSAClient, NHTSADecodeResponse, DecodedVehicleData, MatchedField } from '../external/nhtsa';
+import type { VinDecodeResponse } from '../../ocr/domain/ocr.types';
+import type { DecodedVehicleData, MatchedField } from './vehicles.types';
 import { canAddVehicle, getVehicleLimitConfig } from '../../../core/config/feature-tiers';
 import { UserProfileRepository } from '../../user-profile/data/user-profile.repository';
 import { SubscriptionTier } from '../../user-profile/domain/user-profile.types';
@@ -592,7 +593,7 @@ export class VehiclesService {
     const cacheKey = `${this.cachePrefix}:user:${userId}`;
     await cacheService.del(cacheKey);
   }
-  
+
   async getDropdownMakes(year: number): Promise<string[]> {
     const vehicleDataService = getVehicleDataService();
     const pool = getPool();
@@ -657,82 +658,89 @@ export class VehiclesService {
   }
 
   /**
-   * Map NHTSA decode response to internal decoded vehicle data format
+   * Map VIN decode response to internal decoded vehicle data format
    * with dropdown matching and confidence levels
    */
-  async mapNHTSAResponse(response: NHTSADecodeResponse): Promise<DecodedVehicleData> {
+  async mapVinDecodeResponse(response: VinDecodeResponse): Promise<DecodedVehicleData> {
     const vehicleDataService = getVehicleDataService();
     const pool = getPool();
 
-    // Extract raw values from NHTSA response
+    // Read flat fields directly from Gemini response
-    const nhtsaYear = NHTSAClient.extractYear(response);
+    const sourceYear = response.year;
-    const nhtsaMake = NHTSAClient.extractValue(response, 'Make');
+    const sourceMake = response.make;
-    const nhtsaModel = NHTSAClient.extractValue(response, 'Model');
+    const sourceModel = response.model;
-    const nhtsaTrim = NHTSAClient.extractValue(response, 'Trim');
+    const sourceTrim = response.trimLevel;
-    const nhtsaBodyType = NHTSAClient.extractValue(response, 'Body Class');
+    const sourceBodyType = response.bodyType;
-    const nhtsaDriveType = NHTSAClient.extractValue(response, 'Drive Type');
+    const sourceDriveType = response.driveType;
-    const nhtsaFuelType = NHTSAClient.extractValue(response, 'Fuel Type - Primary');
+    const sourceFuelType = response.fuelType;
-    const nhtsaEngine = NHTSAClient.extractEngine(response);
+    const sourceEngine = response.engine;
-    const nhtsaTransmission = NHTSAClient.extractValue(response, 'Transmission Style');
+    const sourceTransmission = response.transmission;
+
+    logger.debug('VIN decode raw values', {
+      vin: response.vin,
+      year: sourceYear, make: sourceMake, model: sourceModel,
+      trim: sourceTrim, engine: sourceEngine, transmission: sourceTransmission,
+      confidence: response.confidence
+    });
 
     // Year is always high confidence if present (exact numeric match)
     const year: MatchedField<number> = {
-      value: nhtsaYear,
+      value: sourceYear,
-      nhtsaValue: nhtsaYear?.toString() || null,
+      sourceValue: sourceYear?.toString() || null,
-      confidence: nhtsaYear ? 'high' : 'none'
+      confidence: sourceYear ? 'high' : 'none'
     };
 
     // Match make against dropdown options
-    let make: MatchedField<string> = { value: null, nhtsaValue: nhtsaMake, confidence: 'none' };
+    let make: MatchedField<string> = { value: null, sourceValue: sourceMake, confidence: 'none' };
-    if (nhtsaYear && nhtsaMake) {
+    if (sourceYear && sourceMake) {
-      const makes = await vehicleDataService.getMakes(pool, nhtsaYear);
+      const makes = await vehicleDataService.getMakes(pool, sourceYear);
-      make = this.matchField(nhtsaMake, makes);
+      make = this.matchField(sourceMake, makes);
     }
 
     // Match model against dropdown options
-    let model: MatchedField<string> = { value: null, nhtsaValue: nhtsaModel, confidence: 'none' };
+    let model: MatchedField<string> = { value: null, sourceValue: sourceModel, confidence: 'none' };
-    if (nhtsaYear && make.value && nhtsaModel) {
+    if (sourceYear && make.value && sourceModel) {
-      const models = await vehicleDataService.getModels(pool, nhtsaYear, make.value);
+      const models = await vehicleDataService.getModels(pool, sourceYear, make.value);
-      model = this.matchField(nhtsaModel, models);
+      model = this.matchField(sourceModel, models);
     }
 
     // Match trim against dropdown options
-    let trimLevel: MatchedField<string> = { value: null, nhtsaValue: nhtsaTrim, confidence: 'none' };
+    let trimLevel: MatchedField<string> = { value: null, sourceValue: sourceTrim, confidence: 'none' };
-    if (nhtsaYear && make.value && model.value && nhtsaTrim) {
+    if (sourceYear && make.value && model.value && sourceTrim) {
-      const trims = await vehicleDataService.getTrims(pool, nhtsaYear, make.value, model.value);
+      const trims = await vehicleDataService.getTrims(pool, sourceYear, make.value, model.value);
-      trimLevel = this.matchField(nhtsaTrim, trims);
+      trimLevel = this.matchField(sourceTrim, trims);
     }
 
     // Match engine against dropdown options
-    let engine: MatchedField<string> = { value: null, nhtsaValue: nhtsaEngine, confidence: 'none' };
+    let engine: MatchedField<string> = { value: null, sourceValue: sourceEngine, confidence: 'none' };
-    if (nhtsaYear && make.value && model.value && trimLevel.value && nhtsaEngine) {
+    if (sourceYear && make.value && model.value && trimLevel.value && sourceEngine) {
-      const engines = await vehicleDataService.getEngines(pool, nhtsaYear, make.value, model.value, trimLevel.value);
+      const engines = await vehicleDataService.getEngines(pool, sourceYear, make.value, model.value, trimLevel.value);
-      engine = this.matchField(nhtsaEngine, engines);
+      engine = this.matchField(sourceEngine, engines);
     }
 
     // Match transmission against dropdown options
-    let transmission: MatchedField<string> = { value: null, nhtsaValue: nhtsaTransmission, confidence: 'none' };
+    let transmission: MatchedField<string> = { value: null, sourceValue: sourceTransmission, confidence: 'none' };
-    if (nhtsaYear && make.value && model.value && trimLevel.value && nhtsaTransmission) {
+    if (sourceYear && make.value && model.value && trimLevel.value && sourceTransmission) {
-      const transmissions = await vehicleDataService.getTransmissionsForTrim(pool, nhtsaYear, make.value, model.value, trimLevel.value);
+      const transmissions = await vehicleDataService.getTransmissionsForTrim(pool, sourceYear, make.value, model.value, trimLevel.value);
-      transmission = this.matchField(nhtsaTransmission, transmissions);
+      transmission = this.matchField(sourceTransmission, transmissions);
     }
 
     // Body type, drive type, and fuel type are display-only (no dropdown matching)
     const bodyType: MatchedField<string> = {
       value: null,
-      nhtsaValue: nhtsaBodyType,
+      sourceValue: sourceBodyType,
       confidence: 'none'
     };
 
     const driveType: MatchedField<string> = {
       value: null,
-      nhtsaValue: nhtsaDriveType,
+      sourceValue: sourceDriveType,
       confidence: 'none'
     };
 
     const fuelType: MatchedField<string> = {
       value: null,
-      nhtsaValue: nhtsaFuelType,
+      sourceValue: sourceFuelType,
       confidence: 'none'
     };
 
@@ -754,42 +762,62 @@ export class VehiclesService {
    * Returns the matched dropdown value with confidence level
    * Matching order: exact -> normalized -> prefix -> contains
    */
-  private matchField(nhtsaValue: string, options: string[]): MatchedField<string> {
+  private matchField(sourceValue: string, options: string[]): MatchedField<string> {
-    if (!nhtsaValue || options.length === 0) {
+    if (!sourceValue || options.length === 0) {
-      return { value: null, nhtsaValue, confidence: 'none' };
+      return { value: null, sourceValue, confidence: 'none' };
     }
 
-    const normalizedNhtsa = nhtsaValue.toLowerCase().trim();
+    const normalizedSource = sourceValue.toLowerCase().trim();
 
     // Try exact case-insensitive match
-    const exactMatch = options.find(opt => opt.toLowerCase().trim() === normalizedNhtsa);
+    const exactMatch = options.find(opt => opt.toLowerCase().trim() === normalizedSource);
     if (exactMatch) {
-      return { value: exactMatch, nhtsaValue, confidence: 'high' };
+      return { value: exactMatch, sourceValue, confidence: 'high' };
     }
 
     // Try normalized comparison (remove special chars)
     const normalizeForCompare = (s: string) => s.toLowerCase().replace(/[^a-z0-9]/g, '');
-    const normalizedNhtsaClean = normalizeForCompare(nhtsaValue);
+    const normalizedSourceClean = normalizeForCompare(sourceValue);
 
-    const normalizedMatch = options.find(opt => normalizeForCompare(opt) === normalizedNhtsaClean);
+    const normalizedMatch = options.find(opt => normalizeForCompare(opt) === normalizedSourceClean);
     if (normalizedMatch) {
-      return { value: normalizedMatch, nhtsaValue, confidence: 'medium' };
+      return { value: normalizedMatch, sourceValue, confidence: 'medium' };
     }
 
-    // Try prefix match - option starts with NHTSA value
+    // Try prefix match - option starts with source value
-    const prefixMatch = options.find(opt => opt.toLowerCase().trim().startsWith(normalizedNhtsa));
+    const prefixMatch = options.find(opt => opt.toLowerCase().trim().startsWith(normalizedSource));
     if (prefixMatch) {
-      return { value: prefixMatch, nhtsaValue, confidence: 'medium' };
+      return { value: prefixMatch, sourceValue, confidence: 'medium' };
     }
 
-    // Try contains match - option contains NHTSA value
+    // Try contains match - option contains source value
-    const containsMatch = options.find(opt => opt.toLowerCase().trim().includes(normalizedNhtsa));
+    const containsMatch = options.find(opt => opt.toLowerCase().trim().includes(normalizedSource));
     if (containsMatch) {
-      return { value: containsMatch, nhtsaValue, confidence: 'medium' };
+      return { value: containsMatch, sourceValue, confidence: 'medium' };
     }
 
-    // No match found - return NHTSA value as hint with no match
+    // Try reverse contains - source value contains option (e.g., source "X5 xDrive35i" contains option "X5")
-    return { value: null, nhtsaValue, confidence: 'none' };
+    // Prefer the longest matching option to avoid false positives (e.g., "X5 M" over "X5")
+    const reverseMatches = options.filter(opt => {
+      const normalizedOpt = opt.toLowerCase().trim();
+      return normalizedSource.includes(normalizedOpt) && normalizedOpt.length > 0;
+    });
+    if (reverseMatches.length > 0) {
+      const bestMatch = reverseMatches.reduce((a, b) => a.length >= b.length ? a : b);
+      return { value: bestMatch, sourceValue, confidence: 'medium' };
+    }
+
+    // Try word-start match - source starts with option + separator (e.g., "X5 xDrive" starts with "X5 ")
+    const wordStartMatch = options.find(opt => {
+      const normalizedOpt = opt.toLowerCase().trim();
+      return normalizedSource.startsWith(normalizedOpt + ' ') || normalizedSource.startsWith(normalizedOpt + '-');
+    });
+    if (wordStartMatch) {
+      return { value: wordStartMatch, sourceValue, confidence: 'medium' };
+    }
+
+    // No match found - return source value as hint with no match
+    return { value: null, sourceValue, confidence: 'none' };
   }
 
   private toResponse(vehicle: Vehicle): VehicleResponse {

backend/src/features/vehicles/domain/vehicles.types.ts

@@ -215,3 +215,41 @@ export interface TCOResponse {
   distanceUnit: string;
   currencyCode: string;
 }
+
+/** Confidence level for matched dropdown values */
+export type MatchConfidence = 'high' | 'medium' | 'none';
+
+/** Matched field with confidence indicator */
+export interface MatchedField<T> {
+  value: T | null;
+  sourceValue: string | null;
+  confidence: MatchConfidence;
+}
+
+/**
+ * Decoded vehicle data with match confidence per field.
+ * Maps VIN decode response fields to internal field names.
+ */
+export interface DecodedVehicleData {
+  year: MatchedField<number>;
+  make: MatchedField<string>;
+  model: MatchedField<string>;
+  trimLevel: MatchedField<string>;
+  bodyType: MatchedField<string>;
+  driveType: MatchedField<string>;
+  fuelType: MatchedField<string>;
+  engine: MatchedField<string>;
+  transmission: MatchedField<string>;
+}
+
+/** VIN decode request body */
+export interface DecodeVinRequest {
+  vin: string;
+}
+
+/** VIN decode error response */
+export interface VinDecodeError {
+  error: 'INVALID_VIN' | 'VIN_DECODE_FAILED' | 'TIER_REQUIRED';
+  message: string;
+  details?: string;
+}

backend/src/features/vehicles/external/CLAUDE.md

@@ -5,9 +5,3 @@
 | File | What | When to read |
 | ---- | ---- | ------------ |
 | `README.md` | Integration patterns, adding new services | Understanding external service conventions |
-
-## Subdirectories
-
-| Directory | What | When to read |
-| --------- | ---- | ------------ |
-| `nhtsa/` | NHTSA vPIC API client for VIN decoding | VIN decode feature work |

backend/src/features/vehicles/external/README.md

@@ -15,7 +15,7 @@ Each integration follows this structure:
 ## Adding New Integrations
 
 1. Create subdirectory: `external/{service}/`
-2. Add client: `{service}.client.ts` following NHTSAClient pattern
+2. Add client: `{service}.client.ts` following the axios-based client pattern
 3. Add types: `{service}.types.ts`
 4. Update `CLAUDE.md` with new directory
 5. Add tests in `tests/unit/{service}.client.test.ts`

backend/src/features/vehicles/external/nhtsa/index.ts

@@ -1,16 +0,0 @@
-/**
- * @ai-summary NHTSA vPIC integration exports
- * @ai-context Public API for VIN decoding functionality
- */
-
-export { NHTSAClient } from './nhtsa.client';
-export type {
-  NHTSADecodeResponse,
-  NHTSAResult,
-  DecodedVehicleData,
-  MatchedField,
-  MatchConfidence,
-  VinCacheEntry,
-  DecodeVinRequest,
-  VinDecodeError,
-} from './nhtsa.types';

backend/src/features/vehicles/external/nhtsa/nhtsa.client.ts

@@ -1,235 +0,0 @@
-/**
- * @ai-summary NHTSA vPIC API client for VIN decoding
- * @ai-context Fetches vehicle data from NHTSA and caches results
- */
-
-import axios, { AxiosError } from 'axios';
-import { logger } from '../../../../core/logging/logger';
-import { NHTSADecodeResponse, VinCacheEntry } from './nhtsa.types';
-import { Pool } from 'pg';
-
-/**
- * VIN validation regex
- * - 17 characters
- * - Excludes I, O, Q (not used in VINs)
- * - Alphanumeric only
- */
-const VIN_REGEX = /^[A-HJ-NPR-Z0-9]{17}$/;
-
-/**
- * Cache TTL: 1 year (VIN data is static - vehicle specs don't change)
- */
-const CACHE_TTL_SECONDS = 365 * 24 * 60 * 60;
-
-export class NHTSAClient {
-  private readonly baseURL = 'https://vpic.nhtsa.dot.gov/api';
-  private readonly timeout = 5000; // 5 seconds
-
-  constructor(private readonly pool: Pool) {}
-
-  /**
-   * Validate VIN format
-   * @throws Error if VIN format is invalid
-   */
-  validateVin(vin: string): string {
-    const sanitized = vin.trim().toUpperCase();
-
-    if (!sanitized) {
-      throw new Error('VIN is required');
-    }
-
-    if (!VIN_REGEX.test(sanitized)) {
-      throw new Error('Invalid VIN format. VIN must be exactly 17 characters and contain only letters (except I, O, Q) and numbers.');
-    }
-
-    return sanitized;
-  }
-
-  /**
-   * Check cache for existing VIN data
-   */
-  async getCached(vin: string): Promise<VinCacheEntry | null> {
-    try {
-      const result = await this.pool.query<{
-        vin: string;
-        make: string | null;
-        model: string | null;
-        year: number | null;
-        engine_type: string | null;
-        body_type: string | null;
-        raw_data: NHTSADecodeResponse;
-        cached_at: Date;
-      }>(
-        `SELECT vin, make, model, year, engine_type, body_type, raw_data, cached_at
-         FROM vin_cache
-         WHERE vin = $1
-         AND cached_at > NOW() - INTERVAL '${CACHE_TTL_SECONDS} seconds'`,
-        [vin]
-      );
-
-      if (result.rows.length === 0) {
-        return null;
-      }
-
-      const row = result.rows[0];
-      return {
-        vin: row.vin,
-        make: row.make,
-        model: row.model,
-        year: row.year,
-        engineType: row.engine_type,
-        bodyType: row.body_type,
-        rawData: row.raw_data,
-        cachedAt: row.cached_at,
-      };
-    } catch (error) {
-      logger.error('Failed to check VIN cache', { vin, error });
-      return null;
-    }
-  }
-
-  /**
-   * Save VIN data to cache
-   */
-  async saveToCache(vin: string, response: NHTSADecodeResponse): Promise<void> {
-    try {
-      const findValue = (variable: string): string | null => {
-        const result = response.Results.find(r => r.Variable === variable);
-        return result?.Value || null;
-      };
-
-      const year = findValue('Model Year');
-      const make = findValue('Make');
-      const model = findValue('Model');
-      const engineType = findValue('Engine Model');
-      const bodyType = findValue('Body Class');
-
-      await this.pool.query(
-        `INSERT INTO vin_cache (vin, make, model, year, engine_type, body_type, raw_data, cached_at)
-         VALUES ($1, $2, $3, $4, $5, $6, $7, NOW())
-         ON CONFLICT (vin) DO UPDATE SET
-           make = EXCLUDED.make,
-           model = EXCLUDED.model,
-           year = EXCLUDED.year,
-           engine_type = EXCLUDED.engine_type,
-           body_type = EXCLUDED.body_type,
-           raw_data = EXCLUDED.raw_data,
-           cached_at = NOW()`,
-        [vin, make, model, year ? parseInt(year) : null, engineType, bodyType, JSON.stringify(response)]
-      );
-
-      logger.debug('VIN cached', { vin });
-    } catch (error) {
-      logger.error('Failed to cache VIN data', { vin, error });
-      // Don't throw - caching failure shouldn't break the decode flow
-    }
-  }
-
-  /**
-   * Decode VIN using NHTSA vPIC API
-   * @param vin - 17-character VIN
-   * @returns Raw NHTSA decode response
-   * @throws Error if VIN is invalid or API call fails
-   */
-  async decodeVin(vin: string): Promise<NHTSADecodeResponse> {
-    // Validate and sanitize VIN
-    const sanitizedVin = this.validateVin(vin);
-
-    // Check cache first
-    const cached = await this.getCached(sanitizedVin);
-    if (cached) {
-      logger.debug('VIN cache hit', { vin: sanitizedVin });
-      return cached.rawData;
-    }
-
-    // Call NHTSA API
-    logger.info('Calling NHTSA vPIC API', { vin: sanitizedVin });
-
-    try {
-      const response = await axios.get<NHTSADecodeResponse>(
-        `${this.baseURL}/vehicles/decodevin/${sanitizedVin}`,
-        {
-          params: { format: 'json' },
-          timeout: this.timeout,
-        }
-      );
-
-      // Check for NHTSA-level errors
-      if (response.data.Count === 0) {
-        throw new Error('NHTSA returned no results for this VIN');
-      }
-
-      // Check for error messages in results
-      const errorResult = response.data.Results.find(
-        r => r.Variable === 'Error Code' && r.Value && r.Value !== '0'
-      );
-      if (errorResult) {
-        const errorText = response.data.Results.find(r => r.Variable === 'Error Text');
-        throw new Error(`NHTSA error: ${errorText?.Value || 'Unknown error'}`);
-      }
-
-      // Cache the successful response
-      await this.saveToCache(sanitizedVin, response.data);
-
-      return response.data;
-    } catch (error) {
-      if (axios.isAxiosError(error)) {
-        const axiosError = error as AxiosError;
-        if (axiosError.code === 'ECONNABORTED') {
-          logger.error('NHTSA API timeout', { vin: sanitizedVin });
-          throw new Error('NHTSA API request timed out. Please try again.');
-        }
-        if (axiosError.response) {
-          logger.error('NHTSA API error response', {
-            vin: sanitizedVin,
-            status: axiosError.response.status,
-            data: axiosError.response.data,
-          });
-          throw new Error(`NHTSA API error: ${axiosError.response.status}`);
-        }
-        logger.error('NHTSA API network error', { vin: sanitizedVin, message: axiosError.message });
-        throw new Error('Unable to connect to NHTSA API. Please try again later.');
-      }
-      throw error;
-    }
-  }
-
-  /**
-   * Extract a specific value from NHTSA response
-   */
-  static extractValue(response: NHTSADecodeResponse, variable: string): string | null {
-    const result = response.Results.find(r => r.Variable === variable);
-    return result?.Value?.trim() || null;
-  }
-
-  /**
-   * Extract year from NHTSA response
-   */
-  static extractYear(response: NHTSADecodeResponse): number | null {
-    const value = NHTSAClient.extractValue(response, 'Model Year');
-    if (!value) return null;
-    const parsed = parseInt(value, 10);
-    return isNaN(parsed) ? null : parsed;
-  }
-
-  /**
-   * Extract engine description from NHTSA response
-   * Combines multiple engine-related fields
-   */
-  static extractEngine(response: NHTSADecodeResponse): string | null {
-    const engineModel = NHTSAClient.extractValue(response, 'Engine Model');
-    if (engineModel) return engineModel;
-
-    // Build engine description from components
-    const cylinders = NHTSAClient.extractValue(response, 'Engine Number of Cylinders');
-    const displacement = NHTSAClient.extractValue(response, 'Displacement (L)');
-    const fuelType = NHTSAClient.extractValue(response, 'Fuel Type - Primary');
-
-    const parts: string[] = [];
-    if (cylinders) parts.push(`${cylinders}-Cylinder`);
-    if (displacement) parts.push(`${displacement}L`);
-    if (fuelType && fuelType !== 'Gasoline') parts.push(fuelType);
-
-    return parts.length > 0 ? parts.join(' ') : null;
-  }
-}

backend/src/features/vehicles/external/nhtsa/nhtsa.types.ts

@@ -1,96 +0,0 @@
-/**
- * @ai-summary Type definitions for NHTSA vPIC API
- * @ai-context Defines request/response types for VIN decoding
- */
-
-/**
- * Individual result from NHTSA DecodeVin API
- */
-export interface NHTSAResult {
-  Value: string | null;
-  ValueId: string | null;
-  Variable: string;
-  VariableId: number;
-}
-
-/**
- * Raw response from NHTSA DecodeVin API
- * GET https://vpic.nhtsa.dot.gov/api/vehicles/decodevin/{VIN}?format=json
- */
-export interface NHTSADecodeResponse {
-  Count: number;
-  Message: string;
-  SearchCriteria: string;
-  Results: NHTSAResult[];
-}
-
-/**
- * Confidence level for matched dropdown values
- */
-export type MatchConfidence = 'high' | 'medium' | 'none';
-
-/**
- * Matched field with confidence indicator
- */
-export interface MatchedField<T> {
-  value: T | null;
-  nhtsaValue: string | null;
-  confidence: MatchConfidence;
-}
-
-/**
- * Decoded vehicle data with match confidence per field
- * Maps NHTSA response fields to internal field names (camelCase)
- *
- * NHTSA Field Mappings:
- * - ModelYear -> year
- * - Make -> make
- * - Model -> model
- * - Trim -> trimLevel
- * - BodyClass -> bodyType
- * - DriveType -> driveType
- * - FuelTypePrimary -> fuelType
- * - EngineModel / EngineCylinders + EngineDisplacementL -> engine
- * - TransmissionStyle -> transmission
- */
-export interface DecodedVehicleData {
-  year: MatchedField<number>;
-  make: MatchedField<string>;
-  model: MatchedField<string>;
-  trimLevel: MatchedField<string>;
-  bodyType: MatchedField<string>;
-  driveType: MatchedField<string>;
-  fuelType: MatchedField<string>;
-  engine: MatchedField<string>;
-  transmission: MatchedField<string>;
-}
-
-/**
- * Cached VIN data from vin_cache table
- */
-export interface VinCacheEntry {
-  vin: string;
-  make: string | null;
-  model: string | null;
-  year: number | null;
-  engineType: string | null;
-  bodyType: string | null;
-  rawData: NHTSADecodeResponse;
-  cachedAt: Date;
-}
-
-/**
- * VIN decode request body
- */
-export interface DecodeVinRequest {
-  vin: string;
-}
-
-/**
- * VIN decode error response
- */
-export interface VinDecodeError {
-  error: 'INVALID_VIN' | 'VIN_DECODE_FAILED' | 'TIER_REQUIRED';
-  message: string;
-  details?: string;
-}

backend/src/features/vehicles/tests/integration/vehicles.integration.test.ts

@@ -36,15 +36,13 @@ describe('Vehicles Integration Tests', () => {
   afterAll(async () => {
     // Clean up test database
     await pool.query('DROP TABLE IF EXISTS vehicles CASCADE');
-    await pool.query('DROP TABLE IF EXISTS vin_cache CASCADE');
     await pool.query('DROP FUNCTION IF EXISTS update_updated_at_column() CASCADE');
     await pool.end();
   });
 
   beforeEach(async () => {
-    // Clean up test data before each test - more thorough cleanup
+    // Clean up test data before each test
     await pool.query('DELETE FROM vehicles WHERE user_id = $1', ['test-user-123']);
-    await pool.query('DELETE FROM vin_cache WHERE vin LIKE $1', ['1HGBH41JXMN%']);
     
     // Clear Redis cache for the test user
     try {

config/app/ci.yml

@@ -22,11 +22,6 @@ platform:
       url: http://mvp-platform-vehicles-api:8000
       timeout: 5s
 
-external:
-  vpic:
-    url: https://vpic.nhtsa.dot.gov/api/vehicles
-    timeout: 10s
-
 service:
   name: mvp-backend
 

config/app/production.yml.example

@@ -21,5 +21,3 @@ auth0:
   domain: motovaultpro.us.auth0.com
   audience: https://api.motovaultpro.com
 
-external:
-  vpic_api_url: https://vpic.nhtsa.dot.gov/api/vehicles

config/shared/production.yml

@@ -107,9 +107,6 @@ external_services:
   google_maps:
     base_url: https://maps.googleapis.com/maps/api
 
-  vpic:
-    base_url: https://vpic.nhtsa.dot.gov/api/vehicles
-
 # Development Configuration
 development:
   debug_enabled: false

docker-compose.blue-green.yml

@@ -11,6 +11,63 @@
 # Shared services (from base compose):
 #   mvp-traefik, mvp-postgres, mvp-redis
 
+# ========================================
+# Extension fields (YAML anchors for DRY)
+# ========================================
+x-frontend-env: &frontend-env
+  VITE_API_BASE_URL: /api
+  VITE_AUTH0_DOMAIN: ${VITE_AUTH0_DOMAIN:-motovaultpro.us.auth0.com}
+  VITE_AUTH0_CLIENT_ID: ${VITE_AUTH0_CLIENT_ID:-yspR8zdnSxmV8wFIghHynQ08iXAPoQJ3}
+  VITE_AUTH0_AUDIENCE: ${VITE_AUTH0_AUDIENCE:-https://api.motovaultpro.com}
+  SECRETS_DIR: /run/secrets
+
+x-frontend-volumes: &frontend-volumes
+  - ./secrets/app/google-maps-api-key.txt:/run/secrets/google-maps-api-key:ro
+  - ./secrets/app/google-maps-map-id.txt:/run/secrets/google-maps-map-id:ro
+
+x-frontend-healthcheck: &frontend-healthcheck
+  test: ["CMD-SHELL", "curl -sf http://localhost:3000 || exit 1"]
+  interval: 5s
+  timeout: 5s
+  retries: 3
+  start_period: 10s
+
+x-backend-env: &backend-env
+  NODE_ENV: production
+  CONFIG_PATH: /app/config/production.yml
+  SECRETS_DIR: /run/secrets
+  LOG_LEVEL: ${BACKEND_LOG_LEVEL:-debug}
+  DATABASE_HOST: mvp-postgres
+  REDIS_HOST: mvp-redis
+  STRIPE_PRO_MONTHLY_PRICE_ID: ${STRIPE_PRO_MONTHLY_PRICE_ID:-price_1T1ZHMJXoKkh5RcKwKSSGIlR}
+  STRIPE_PRO_YEARLY_PRICE_ID: ${STRIPE_PRO_YEARLY_PRICE_ID:-price_1T1ZHnJXoKkh5RcKWlG2MPpX}
+  STRIPE_ENTERPRISE_MONTHLY_PRICE_ID: ${STRIPE_ENTERPRISE_MONTHLY_PRICE_ID:-price_1T1ZIBJXoKkh5RcKu2jyhqBN}
+  STRIPE_ENTERPRISE_YEARLY_PRICE_ID: ${STRIPE_ENTERPRISE_YEARLY_PRICE_ID:-price_1T1ZIQJXoKkh5RcK34YXiJQm}
+
+x-backend-volumes: &backend-volumes
+  - ./config/app/production.yml:/app/config/production.yml:ro
+  - ./config/shared/production.yml:/app/config/shared.yml:ro
+  - ./secrets/app/postgres-password.txt:/run/secrets/postgres-password:ro
+  - ./secrets/app/auth0-client-secret.txt:/run/secrets/auth0-client-secret:ro
+  - ./secrets/app/google-maps-api-key.txt:/run/secrets/google-maps-api-key:ro
+  - ./secrets/app/google-maps-map-id.txt:/run/secrets/google-maps-map-id:ro
+  - ./secrets/app/resend-api-key.txt:/run/secrets/resend-api-key:ro
+  - ./secrets/app/auth0-management-client-id.txt:/run/secrets/auth0-management-client-id:ro
+  - ./secrets/app/auth0-management-client-secret.txt:/run/secrets/auth0-management-client-secret:ro
+  - ./secrets/app/stripe-secret-key.txt:/run/secrets/stripe-secret-key:ro
+  - ./secrets/app/stripe-webhook-secret.txt:/run/secrets/stripe-webhook-secret:ro
+  - ./data/documents:/app/data/documents
+  - ./data/backups:/app/data/backups
+
+x-backend-healthcheck: &backend-healthcheck
+  test:
+    - CMD-SHELL
+    - node -e "require('http').get('http://localhost:3001/health', r => process.exit(r.statusCode===200?0:1)).on('error', () => process.exit(1))"
+  interval: 5s
+  timeout: 5s
+  retries: 5
+  start_period: 180s
+
 services:
   # ========================================
   # BLUE Stack - Frontend
@@ -19,25 +76,13 @@ services:
     image: ${FRONTEND_IMAGE:-git.motovaultpro.com/egullickson/frontend:latest}
     container_name: mvp-frontend-blue
     restart: unless-stopped
-    environment:
+    environment: *frontend-env
-      VITE_API_BASE_URL: /api
+    volumes: *frontend-volumes
-      VITE_AUTH0_DOMAIN: ${VITE_AUTH0_DOMAIN:-motovaultpro.us.auth0.com}
-      VITE_AUTH0_CLIENT_ID: ${VITE_AUTH0_CLIENT_ID:-yspR8zdnSxmV8wFIghHynQ08iXAPoQJ3}
-      VITE_AUTH0_AUDIENCE: ${VITE_AUTH0_AUDIENCE:-https://api.motovaultpro.com}
-      SECRETS_DIR: /run/secrets
-    volumes:
-      - ./secrets/app/google-maps-api-key.txt:/run/secrets/google-maps-api-key:ro
-      - ./secrets/app/google-maps-map-id.txt:/run/secrets/google-maps-map-id:ro
     networks:
       - frontend
     depends_on:
       - mvp-backend-blue
-    healthcheck:
+    healthcheck: *frontend-healthcheck
-      test: ["CMD-SHELL", "curl -sf http://localhost:3000 || exit 1"]
-      interval: 5s
-      timeout: 5s
-      retries: 3
-      start_period: 10s
     deploy:
       resources:
         limits:
@@ -55,50 +100,15 @@ services:
     image: ${BACKEND_IMAGE:-git.motovaultpro.com/egullickson/backend:latest}
     container_name: mvp-backend-blue
     restart: unless-stopped
-    environment:
+    environment: *backend-env
-      NODE_ENV: production
+    volumes: *backend-volumes
-      CONFIG_PATH: /app/config/production.yml
-      SECRETS_DIR: /run/secrets
-      DATABASE_HOST: mvp-postgres
-      REDIS_HOST: mvp-redis
-      # Production Variables
-      #STRIPE_PRO_MONTHLY_PRICE_ID: prod_Toj6BG9Z9JwREl
-      #STRIPE_PRO_YEARLY_PRICE_ID: prod_Toj8oo0RpVBQmB
-      #STRIPE_ENTERPRISE_MONTHLY_PRICE_ID: prod_Toj8xGEui9jl6j
-      #STRIPE_ENTERPRISE_YEARLY_PRICE_ID: prod_Toj9A7A773xrdn
-      # Sandbox Variables
-      STRIPE_PRO_MONTHLY_PRICE_ID: price_1T1ZHMJXoKkh5RcKwKSSGIlR
-      STRIPE_PRO_YEARLY_PRICE_ID: price_1T1ZHnJXoKkh5RcKWlG2MPpX
-      STRIPE_ENTERPRISE_MONTHLY_PRICE_ID: price_1T1ZIBJXoKkh5RcKu2jyhqBN
-      STRIPE_ENTERPRISE_YEARLY_PRICE_ID: price_1T1ZIQJXoKkh5RcK34YXiJQm
-    volumes:
-      - ./config/app/production.yml:/app/config/production.yml:ro
-      - ./config/shared/production.yml:/app/config/shared.yml:ro
-      - ./secrets/app/postgres-password.txt:/run/secrets/postgres-password:ro
-      - ./secrets/app/auth0-client-secret.txt:/run/secrets/auth0-client-secret:ro
-      - ./secrets/app/google-maps-api-key.txt:/run/secrets/google-maps-api-key:ro
-      - ./secrets/app/google-maps-map-id.txt:/run/secrets/google-maps-map-id:ro
-      - ./secrets/app/resend-api-key.txt:/run/secrets/resend-api-key:ro
-      - ./secrets/app/auth0-management-client-id.txt:/run/secrets/auth0-management-client-id:ro
-      - ./secrets/app/auth0-management-client-secret.txt:/run/secrets/auth0-management-client-secret:ro
-      - ./secrets/app/stripe-secret-key.txt:/run/secrets/stripe-secret-key:ro
-      - ./secrets/app/stripe-webhook-secret.txt:/run/secrets/stripe-webhook-secret:ro
-      - ./data/documents:/app/data/documents
-      - ./data/backups:/app/data/backups
     networks:
       - backend
       - database
     depends_on:
       - mvp-postgres
       - mvp-redis
-    healthcheck:
+    healthcheck: *backend-healthcheck
-      test:
-        - CMD-SHELL
-        - node -e "require('http').get('http://localhost:3001/health', r => process.exit(r.statusCode===200?0:1)).on('error', () => process.exit(1))"
-      interval: 5s
-      timeout: 5s
-      retries: 5
-      start_period: 180s
     deploy:
       resources:
         limits:
@@ -116,25 +126,13 @@ services:
     image: ${FRONTEND_IMAGE:-git.motovaultpro.com/egullickson/frontend:latest}
     container_name: mvp-frontend-green
     restart: unless-stopped
-    environment:
+    environment: *frontend-env
-      VITE_API_BASE_URL: /api
+    volumes: *frontend-volumes
-      VITE_AUTH0_DOMAIN: ${VITE_AUTH0_DOMAIN:-motovaultpro.us.auth0.com}
-      VITE_AUTH0_CLIENT_ID: ${VITE_AUTH0_CLIENT_ID:-yspR8zdnSxmV8wFIghHynQ08iXAPoQJ3}
-      VITE_AUTH0_AUDIENCE: ${VITE_AUTH0_AUDIENCE:-https://api.motovaultpro.com}
-      SECRETS_DIR: /run/secrets
-    volumes:
-      - ./secrets/app/google-maps-api-key.txt:/run/secrets/google-maps-api-key:ro
-      - ./secrets/app/google-maps-map-id.txt:/run/secrets/google-maps-map-id:ro
     networks:
       - frontend
     depends_on:
       - mvp-backend-green
-    healthcheck:
+    healthcheck: *frontend-healthcheck
-      test: ["CMD-SHELL", "curl -sf http://localhost:3000 || exit 1"]
-      interval: 5s
-      timeout: 5s
-      retries: 3
-      start_period: 10s
     deploy:
       resources:
         limits:
@@ -152,44 +150,15 @@ services:
     image: ${BACKEND_IMAGE:-git.motovaultpro.com/egullickson/backend:latest}
     container_name: mvp-backend-green
     restart: unless-stopped
-    environment:
+    environment: *backend-env
-      NODE_ENV: production
+    volumes: *backend-volumes
-      CONFIG_PATH: /app/config/production.yml
-      SECRETS_DIR: /run/secrets
-      DATABASE_HOST: mvp-postgres
-      REDIS_HOST: mvp-redis
-      STRIPE_PRO_MONTHLY_PRICE_ID: prod_Toj6BG9Z9JwREl
-      STRIPE_PRO_YEARLY_PRICE_ID: prod_Toj8oo0RpVBQmB
-      STRIPE_ENTERPRISE_MONTHLY_PRICE_ID: prod_Toj8xGEui9jl6j
-      STRIPE_ENTERPRISE_YEARLY_PRICE_ID: prod_Toj9A7A773xrdn
-    volumes:
-      - ./config/app/production.yml:/app/config/production.yml:ro
-      - ./config/shared/production.yml:/app/config/shared.yml:ro
-      - ./secrets/app/postgres-password.txt:/run/secrets/postgres-password:ro
-      - ./secrets/app/auth0-client-secret.txt:/run/secrets/auth0-client-secret:ro
-      - ./secrets/app/google-maps-api-key.txt:/run/secrets/google-maps-api-key:ro
-      - ./secrets/app/google-maps-map-id.txt:/run/secrets/google-maps-map-id:ro
-      - ./secrets/app/resend-api-key.txt:/run/secrets/resend-api-key:ro
-      - ./secrets/app/auth0-management-client-id.txt:/run/secrets/auth0-management-client-id:ro
-      - ./secrets/app/auth0-management-client-secret.txt:/run/secrets/auth0-management-client-secret:ro
-      - ./secrets/app/stripe-secret-key.txt:/run/secrets/stripe-secret-key:ro
-      - ./secrets/app/stripe-webhook-secret.txt:/run/secrets/stripe-webhook-secret:ro
-      - ./data/documents:/app/data/documents
-      - ./data/backups:/app/data/backups
     networks:
       - backend
       - database
     depends_on:
       - mvp-postgres
       - mvp-redis
-    healthcheck:
+    healthcheck: *backend-healthcheck
-      test:
-        - CMD-SHELL
-        - node -e "require('http').get('http://localhost:3001/health', r => process.exit(r.statusCode===200?0:1)).on('error', () => process.exit(1))"
-      interval: 5s
-      timeout: 5s
-      retries: 5
-      start_period: 180s
     deploy:
       resources:
         limits:

docker-compose.prod.yml

@@ -6,18 +6,14 @@
 #
 # This file removes development-only configurations:
 #   - Database port exposure (PostgreSQL, Redis)
-#   - Development-specific settings
+#   - Traefik dashboard auth middleware
+#
+# Environment-specific values (log levels, Stripe IDs) are driven by .env
+# generated by CI/CD from Gitea variables + scripts/ci/generate-log-config.sh
 
 services:
-  # Traefik - Production log level and dashboard auth
+  # Traefik - Dashboard auth middleware
   mvp-traefik:
-    environment:
-      # Traefik log levels: TRACE | DEBUG | INFO | WARN | ERROR
-      LOG_LEVEL: error
-    command:
-      - --configFile=/etc/traefik/traefik.yml
-      # Traefik log levels: TRACE | DEBUG | INFO | WARN | ERROR
-      - --log.level=ERROR
     labels:
       - "traefik.enable=true"
       - "traefik.http.routers.traefik-dashboard.rule=Host(`traefik.motovaultpro.local`)"
@@ -26,64 +22,10 @@ services:
       - "traefik.http.services.traefik-dashboard.loadbalancer.server.port=8080"
       - "traefik.http.middlewares.dashboard-auth.basicauth.users=admin:$$2y$$10$$foobar"
 
-  # Backend - Production log level
+  # PostgreSQL - Remove dev ports
-  mvp-backend:
-    environment:
-      NODE_ENV: production
-      CONFIG_PATH: /app/config/production.yml
-      SECRETS_DIR: /run/secrets
-      # Pino log levels: trace | debug | info | warn | error | fatal
-      LOG_LEVEL: error
-      DATABASE_HOST: mvp-postgres
-      REDIS_HOST: mvp-redis
-      # Production Variables
-      #STRIPE_PRO_MONTHLY_PRICE_ID: prod_Toj6BG9Z9JwREl
-      #STRIPE_PRO_YEARLY_PRICE_ID: prod_Toj8oo0RpVBQmB
-      #STRIPE_ENTERPRISE_MONTHLY_PRICE_ID: prod_Toj8xGEui9jl6j
-      #STRIPE_ENTERPRISE_YEARLY_PRICE_ID: prod_Toj9A7A773xrdn
-      # Sandbox Variables
-      STRIPE_PRO_MONTHLY_PRICE_ID: price_1T1ZHMJXoKkh5RcKwKSSGIlR
-      STRIPE_PRO_YEARLY_PRICE_ID: price_1T1ZHnJXoKkh5RcKWlG2MPpX
-      STRIPE_ENTERPRISE_MONTHLY_PRICE_ID: price_1T1ZIBJXoKkh5RcKu2jyhqBN
-      STRIPE_ENTERPRISE_YEARLY_PRICE_ID: price_1T1ZIQJXoKkh5RcK34YXiJQm
-
-  # OCR - Production log level + engine config
-  mvp-ocr:
-    environment:
-      # Python log levels: DEBUG | INFO | WARNING | ERROR | CRITICAL
-      LOG_LEVEL: error
-      REDIS_HOST: mvp-redis
-      REDIS_PORT: 6379
-      REDIS_DB: 1
-      # OCR engine configuration (Google Vision primary, PaddleOCR fallback)
-      OCR_PRIMARY_ENGINE: google_vision
-      OCR_FALLBACK_ENGINE: paddleocr
-      OCR_CONFIDENCE_THRESHOLD: "0.6"
-      OCR_FALLBACK_THRESHOLD: "0.6"
-      GOOGLE_VISION_KEY_PATH: /run/secrets/google-wif-config.json
-      VISION_MONTHLY_LIMIT: "1000"
-      # Vertex AI / Gemini configuration (maintenance schedule extraction)
-      VERTEX_AI_PROJECT: motovaultpro
-      VERTEX_AI_LOCATION: us-central1
-      GEMINI_MODEL: gemini-2.5-flash
-
-  # PostgreSQL - Remove dev ports, production log level
   mvp-postgres:
     ports: []
-    environment:
-      POSTGRES_DB: motovaultpro
-      POSTGRES_USER: postgres
-      POSTGRES_PASSWORD_FILE: /run/secrets/postgres-password
-      POSTGRES_INITDB_ARGS: --encoding=UTF8
-      LOG_LEVEL: error
-      # PostgreSQL log statements: none | ddl | mod | all
-      POSTGRES_LOG_STATEMENT: none
-      # Minimum query duration to log: -1 (disabled) | 0 (all) | N (ms threshold)
-      POSTGRES_LOG_MIN_DURATION_STATEMENT: -1
-      PGDATA: /var/lib/postgresql/data/pgdata
 
-  # Redis - Remove dev ports, production log level
+  # Redis - Remove dev ports
   mvp-redis:
     ports: []
-    # Redis log levels: debug | verbose | notice | warning
-    command: redis-server --appendonly yes --loglevel warning

docker-compose.staging.yml

@@ -63,27 +63,6 @@ services:
   mvp-ocr:
     image: ${OCR_IMAGE:-git.motovaultpro.com/egullickson/ocr:latest}
     container_name: mvp-ocr-staging
-    environment:
-      # Python log levels: DEBUG | INFO | WARNING | ERROR | CRITICAL
-      LOG_LEVEL: debug
-      REDIS_HOST: mvp-redis
-      REDIS_PORT: 6379
-      REDIS_DB: 1
-      # OCR engine configuration (Google Vision primary, PaddleOCR fallback)
-      OCR_PRIMARY_ENGINE: google_vision
-      OCR_FALLBACK_ENGINE: paddleocr
-      OCR_CONFIDENCE_THRESHOLD: "0.6"
-      OCR_FALLBACK_THRESHOLD: "0.6"
-      GOOGLE_VISION_KEY_PATH: /run/secrets/google-wif-config.json
-      VISION_MONTHLY_LIMIT: "1000"
-      # Vertex AI / Gemini configuration (maintenance schedule extraction)
-      VERTEX_AI_PROJECT: motovaultpro
-      VERTEX_AI_LOCATION: us-central1
-      GEMINI_MODEL: gemini-2.5-flash
-    volumes:
-      - ./secrets/app/auth0-ocr-client-id.txt:/run/secrets/auth0-ocr-client-id:ro
-      - ./secrets/app/auth0-ocr-client-secret.txt:/run/secrets/auth0-ocr-client-secret:ro
-      - ./secrets/app/google-wif-config.json:/run/secrets/google-wif-config.json:ro
 
   # ========================================
   # PostgreSQL (Staging - Separate Database)

docker-compose.yml

@@ -11,8 +11,9 @@ services:
     command:
       - --configFile=/etc/traefik/traefik.yml
     environment:
-      # Traefik log levels: TRACE | DEBUG | INFO | WARN | ERROR
+      # Traefik natively reads TRAEFIK_LOG_LEVEL (maps to --log.level)
-      LOG_LEVEL: debug
+      # Levels: TRACE | DEBUG | INFO | WARN | ERROR
+      TRAEFIK_LOG_LEVEL: ${TRAEFIK_LOG_LEVEL:-DEBUG}
       CLOUDFLARE_DNS_API_TOKEN_FILE: /run/secrets/cloudflare-dns-token
     ports:
       - "80:80"
@@ -60,7 +61,7 @@ services:
         VITE_AUTH0_CLIENT_ID: ${VITE_AUTH0_CLIENT_ID:-yspR8zdnSxmV8wFIghHynQ08iXAPoQJ3}
         VITE_AUTH0_AUDIENCE: ${VITE_AUTH0_AUDIENCE:-https://api.motovaultpro.com}
         VITE_API_BASE_URL: ${VITE_API_BASE_URL:-/api}
-        VITE_STRIPE_PUBLISHABLE_KEY: ${VITE_STRIPE_PUBLISHABLE_KEY:-pk_live_51Sr2yQJk87CpWj04YNBIaUWUtnJjeVTgk5NqHdpjqxgsbjy3dMKkIsqhjcpSkCzp3KvLi23BGgxhwV021EnEW3H400HhPYVyfN}
+        VITE_STRIPE_PUBLISHABLE_KEY: ${VITE_STRIPE_PUBLISHABLE_KEY}
     container_name: mvp-frontend
     restart: unless-stopped
     environment:
@@ -115,15 +116,15 @@ services:
       CONFIG_PATH: /app/config/production.yml
       SECRETS_DIR: /run/secrets
       # Pino log levels: trace | debug | info | warn | error | fatal
-      LOG_LEVEL: debug
+      LOG_LEVEL: ${BACKEND_LOG_LEVEL:-debug}
       # Service references
       DATABASE_HOST: mvp-postgres
       REDIS_HOST: mvp-redis
-      #Stripe Variables
+      # Stripe Price IDs (override via .env for staging/production)
-      STRIPE_PRO_MONTHLY_PRICE_ID: prod_Toj6BG9Z9JwREl
+      STRIPE_PRO_MONTHLY_PRICE_ID: ${STRIPE_PRO_MONTHLY_PRICE_ID:-price_1T1ZHMJXoKkh5RcKwKSSGIlR}
-      STRIPE_PRO_YEARLY_PRICE_ID: prod_Toj8oo0RpVBQmB
+      STRIPE_PRO_YEARLY_PRICE_ID: ${STRIPE_PRO_YEARLY_PRICE_ID:-price_1T1ZHnJXoKkh5RcKWlG2MPpX}
-      STRIPE_ENTERPRISE_MONTHLY_PRICE_ID: prod_Toj8xGEui9jl6j
+      STRIPE_ENTERPRISE_MONTHLY_PRICE_ID: ${STRIPE_ENTERPRISE_MONTHLY_PRICE_ID:-price_1T1ZIBJXoKkh5RcKu2jyhqBN}
-      STRIPE_ENTERPRISE_YEARLY_PRICE_ID: prod_Toj9A7A773xrdn
+      STRIPE_ENTERPRISE_YEARLY_PRICE_ID: ${STRIPE_ENTERPRISE_YEARLY_PRICE_ID:-price_1T1ZIQJXoKkh5RcK34YXiJQm}
     volumes:
       # Configuration files (K8s ConfigMap equivalent)
       - ./config/app/production.yml:/app/config/production.yml:ro
@@ -192,7 +193,7 @@ services:
     restart: unless-stopped
     environment:
       # Python log levels: DEBUG | INFO | WARNING | ERROR | CRITICAL
-      LOG_LEVEL: debug
+      LOG_LEVEL: ${BACKEND_LOG_LEVEL:-debug}
       REDIS_HOST: mvp-redis
       REDIS_PORT: 6379
       REDIS_DB: 1
@@ -205,8 +206,8 @@ services:
       VISION_MONTHLY_LIMIT: "1000"
       # Vertex AI / Gemini configuration (maintenance schedule extraction)
       VERTEX_AI_PROJECT: motovaultpro
-      VERTEX_AI_LOCATION: us-central1
+      VERTEX_AI_LOCATION: global
-      GEMINI_MODEL: gemini-2.5-flash
+      GEMINI_MODEL: gemini-3-flash-preview
     volumes:
       - /tmp/vin-debug:/tmp/vin-debug
       - ./secrets/app/auth0-ocr-client-id.txt:/run/secrets/auth0-ocr-client-id:ro
@@ -239,11 +240,11 @@ services:
       POSTGRES_USER: postgres
       POSTGRES_PASSWORD_FILE: /run/secrets/postgres-password
       POSTGRES_INITDB_ARGS: --encoding=UTF8
-      LOG_LEVEL: debug
+      LOG_LEVEL: ${BACKEND_LOG_LEVEL:-debug}
       # PostgreSQL log statements: none | ddl | mod | all
-      POSTGRES_LOG_STATEMENT: all
+      POSTGRES_LOG_STATEMENT: ${POSTGRES_LOG_STATEMENT:-all}
       # Minimum query duration to log: -1 (disabled) | 0 (all) | N (ms threshold)
-      POSTGRES_LOG_MIN_DURATION_STATEMENT: 0
+      POSTGRES_LOG_MIN_DURATION_STATEMENT: ${POSTGRES_LOG_MIN_DURATION:-0}
       PGDATA: /var/lib/postgresql/data/pgdata
     volumes:
       - mvp_postgres_data:/var/lib/postgresql/data/pgdata
@@ -271,7 +272,7 @@ services:
     container_name: mvp-redis
     restart: unless-stopped
     # Redis log levels: debug | verbose | notice | warning
-    command: redis-server --appendonly yes --loglevel debug
+    command: redis-server --appendonly yes --loglevel ${REDIS_LOGLEVEL:-debug}
     volumes:
       - mvp_redis_data:/data
     networks:

docs/PLATFORM-SERVICES.md

@@ -35,7 +35,7 @@ The platform provides vehicle hierarchical data lookups:
 VIN decoding is planned but not yet implemented. Future capabilities will include:
 - `GET /api/platform/vehicle?vin={vin}` - Decode VIN to vehicle details
 - PostgreSQL-based VIN decode function
-- NHTSA vPIC API fallback with circuit breaker
+- Gemini VIN decode via OCR service
 - Redis caching (7-day TTL for successful decodes)
 
 **Data Source**: Vehicle data from standardized sources

docs/TESTING.md

@@ -74,7 +74,7 @@ docker compose exec mvp-frontend npm test -- --coverage
 
 Example: `vehicles.service.test.ts`
 - Tests VIN validation logic
-- Tests vehicle creation with mocked vPIC responses
+- Tests vehicle creation with mocked OCR service responses
 - Tests caching behavior with mocked Redis
 - Tests error handling paths
 
@@ -194,7 +194,7 @@ All 15 features have test suites with unit and/or integration tests:
 - `vehicles` - Unit + integration tests
 
 ### Mock Strategy
-- **External APIs**: Completely mocked (vPIC, Google Maps)
+- **External APIs**: Completely mocked (OCR service, Google Maps)
 - **Database**: Real database with transactions
 - **Redis**: Mocked for unit tests, real for integration
 - **Auth**: Mocked JWT tokens for protected endpoints
@@ -319,9 +319,9 @@ describe('Error Handling', () => {
     ).rejects.toThrow('Invalid VIN format');
   });
   
-  it('should handle vPIC API failure', async () => {
+  it('should handle OCR service failure', async () => {
-    mockVpicClient.decode.mockRejectedValue(new Error('API down'));
+    mockOcrClient.decodeVin.mockRejectedValue(new Error('API down'));
-    
+
     const result = await vehicleService.create(validVehicle, 'user123');
     expect(result.make).toBeNull(); // Graceful degradation
   });

docs/USER-GUIDE.md

@@ -644,7 +644,7 @@ When you attempt to use a Pro feature on the Free tier, an **Upgrade Required**
 
 ### VIN Camera Scanning and Decode (Pro)
 
-**What it does:** Use your device camera to photograph your vehicle's VIN plate, and the system automatically reads the VIN using OCR (Optical Character Recognition) and decodes it from the NHTSA database.
+**What it does:** Use your device camera to photograph your vehicle's VIN plate, and the system automatically reads the VIN using OCR (Optical Character Recognition) and decodes it from the vehicle database.
 
 **How to use it:**
 
@@ -655,7 +655,7 @@ When you attempt to use a Pro feature on the Free tier, an **Upgrade Required**
 5. A **VIN OCR Review modal** appears showing the detected VIN with confidence indicators
 6. Confirm or correct the VIN, then click **Accept**
 7. Click the **Decode VIN** button
-8. The system queries the NHTSA database and auto-populates: Year, Make, Model, Engine, Transmission, and Trim
+8. The system queries the vehicle database and auto-populates: Year, Make, Model, Engine, Transmission, and Trim
 9. Review the pre-filled fields and complete the remaining details
 
 This eliminates manual data entry errors and ensures accurate vehicle specifications.

frontend/src/features/vehicles/api/vehicles.api.ts

@@ -82,11 +82,13 @@ export const vehiclesApi = {
   },
 
   /**
-   * Decode VIN using NHTSA vPIC API
+   * Decode VIN using VIN decode service
    * Requires Pro or Enterprise tier
    */
   decodeVin: async (vin: string): Promise<DecodedVehicleData> => {
-    const response = await apiClient.post('/vehicles/decode-vin', { vin });
+    const response = await apiClient.post('/vehicles/decode-vin', { vin }, {
+      timeout: 120000 // 120 seconds for Gemini + Google Search grounding
+    });
     return response.data;
   }
 };

frontend/src/features/vehicles/components/VehicleForm.tsx

@@ -114,6 +114,7 @@ export const VehicleForm: React.FC<VehicleFormProps> = ({
   const [isDecoding, setIsDecoding] = useState(false);
   const [showUpgradeDialog, setShowUpgradeDialog] = useState(false);
   const [decodeError, setDecodeError] = useState<string | null>(null);
+  const [decodeHint, setDecodeHint] = useState<string | null>(null);
 
   // VIN OCR capture hook
   const vinOcr = useVinOcr();
@@ -507,7 +508,7 @@ export const VehicleForm: React.FC<VehicleFormProps> = ({
 
   /**
    * Handle VIN decode button click
-   * Calls NHTSA API and populates empty form fields
+   * Calls VIN decode service and populates empty form fields
    */
   const handleDecodeVin = async () => {
     // Check tier access first
@@ -524,6 +525,7 @@ export const VehicleForm: React.FC<VehicleFormProps> = ({
 
     setIsDecoding(true);
     setDecodeError(null);
+    setDecodeHint(null);
 
     try {
       const decoded = await vehiclesApi.decodeVin(vin);
@@ -588,6 +590,21 @@ export const VehicleForm: React.FC<VehicleFormProps> = ({
         setValue('transmission', decoded.transmission.value);
       }
 
+      // Check if decode returned data but matching failed for key fields
+      const hasMatchedValue = decoded.year.value || decoded.make.value || decoded.model.value;
+      const hasSourceValue = decoded.year.sourceValue || decoded.make.sourceValue || decoded.model.sourceValue;
+      if (!hasMatchedValue && hasSourceValue) {
+        const parts = [
+          decoded.year.sourceValue,
+          decoded.make.sourceValue,
+          decoded.model.sourceValue,
+          decoded.trimLevel.sourceValue
+        ].filter(Boolean);
+        setDecodeHint(
+          `Could not match VIN data to dropdowns. Decoded as: ${parts.join(' ')}. Please select values manually.`
+        );
+      }
+
       setLoadingDropdowns(false);
       isVinDecoding.current = false;
     } catch (error: any) {
@@ -671,6 +688,9 @@ export const VehicleForm: React.FC<VehicleFormProps> = ({
         {decodeError && (
           <p className="mt-1 text-sm text-red-600 dark:text-red-400">{decodeError}</p>
         )}
+        {decodeHint && (
+          <p className="mt-1 text-sm text-amber-600 dark:text-amber-400">{decodeHint}</p>
+        )}
         {vinOcr.error && (
           <p className="mt-1 text-sm text-red-600 dark:text-red-400">{vinOcr.error}</p>
         )}

frontend/src/features/vehicles/components/VinOcrReviewModal.tsx

@@ -95,8 +95,8 @@ const ReviewContent: React.FC<{
   const [selectedEngine, setSelectedEngine] = useState('');
   const [selectedTransmission, setSelectedTransmission] = useState('');
 
-  // NHTSA reference values for unmatched fields
+  // Source reference values for unmatched fields
-  const [nhtsaRefs, setNhtsaRefs] = useState<Record<string, string | null>>({});
+  const [sourceRefs, setSourceRefs] = useState<Record<string, string | null>>({});
 
   // Initialize dropdown options and pre-select decoded values
   useEffect(() => {
@@ -109,13 +109,13 @@ const ReviewContent: React.FC<{
 
         if (!decodedVehicle) return;
 
-        // Store NHTSA reference values for unmatched fields
+        // Store source reference values for unmatched fields
-        setNhtsaRefs({
+        setSourceRefs({
-          make: decodedVehicle.make.confidence === 'none' ? decodedVehicle.make.nhtsaValue : null,
+          make: decodedVehicle.make.confidence === 'none' ? decodedVehicle.make.sourceValue : null,
-          model: decodedVehicle.model.confidence === 'none' ? decodedVehicle.model.nhtsaValue : null,
+          model: decodedVehicle.model.confidence === 'none' ? decodedVehicle.model.sourceValue : null,
-          trim: decodedVehicle.trimLevel.confidence === 'none' ? decodedVehicle.trimLevel.nhtsaValue : null,
+          trim: decodedVehicle.trimLevel.confidence === 'none' ? decodedVehicle.trimLevel.sourceValue : null,
-          engine: decodedVehicle.engine.confidence === 'none' ? decodedVehicle.engine.nhtsaValue : null,
+          engine: decodedVehicle.engine.confidence === 'none' ? decodedVehicle.engine.sourceValue : null,
-          transmission: decodedVehicle.transmission.confidence === 'none' ? decodedVehicle.transmission.nhtsaValue : null,
+          transmission: decodedVehicle.transmission.confidence === 'none' ? decodedVehicle.transmission.sourceValue : null,
         });
 
         const yearValue = decodedVehicle.year.value;
@@ -277,9 +277,9 @@ const ReviewContent: React.FC<{
     });
   };
 
-  /** Show NHTSA reference when field had no dropdown match */
+  /** Show source reference when field had no dropdown match */
-  const nhtsaHint = (field: string) => {
+  const sourceHint = (field: string) => {
-    const ref = nhtsaRefs[field];
+    const ref = sourceRefs[field];
     if (!ref) return null;
     // Only show hint when no value is currently selected
     const selected: Record<string, string> = {
@@ -292,7 +292,7 @@ const ReviewContent: React.FC<{
     if (selected[field]) return null;
     return (
       <p className="mt-1 text-xs text-gray-500 dark:text-titanio">
-        NHTSA returned: {ref}
+        Decoded value: {ref}
       </p>
     );
   };
@@ -409,7 +409,7 @@ const ReviewContent: React.FC<{
                 </option>
               ))}
             </select>
-            {nhtsaHint('make')}
+            {sourceHint('make')}
           </div>
 
           {/* Model */}
@@ -439,7 +439,7 @@ const ReviewContent: React.FC<{
                 </option>
               ))}
             </select>
-            {nhtsaHint('model')}
+            {sourceHint('model')}
           </div>
 
           {/* Trim */}
@@ -469,7 +469,7 @@ const ReviewContent: React.FC<{
                 </option>
               ))}
             </select>
-            {nhtsaHint('trim')}
+            {sourceHint('trim')}
           </div>
 
           {/* Engine */}
@@ -499,7 +499,7 @@ const ReviewContent: React.FC<{
                 </option>
               ))}
             </select>
-            {nhtsaHint('engine')}
+            {sourceHint('engine')}
           </div>
 
           {/* Transmission */}
@@ -529,7 +529,7 @@ const ReviewContent: React.FC<{
                 </option>
               ))}
             </select>
-            {nhtsaHint('transmission')}
+            {sourceHint('transmission')}
           </div>
         </div>
       </Box>

frontend/src/features/vehicles/hooks/useVinOcr.ts

@@ -1,5 +1,5 @@
 /**
- * @ai-summary Hook to orchestrate VIN OCR extraction and NHTSA decode
+ * @ai-summary Hook to orchestrate VIN OCR extraction and VIN decode
  * @ai-context Handles camera capture -> OCR extraction -> VIN decode flow
  */
 
@@ -109,7 +109,7 @@ export function useVinOcr(): UseVinOcrReturn {
         );
       }
 
-      // Step 2: Decode VIN using NHTSA
+      // Step 2: Decode VIN
       setProcessingStep('decoding');
       let decodedVehicle: DecodedVehicleData | null = null;
       let decodeError: string | null = null;
@@ -121,7 +121,7 @@ export function useVinOcr(): UseVinOcrReturn {
         if (err.response?.data?.error === 'TIER_REQUIRED') {
           decodeError = 'VIN decode requires Pro or Enterprise subscription';
         } else if (err.response?.data?.error === 'INVALID_VIN') {
-          decodeError = 'VIN format is not recognized by NHTSA';
+          decodeError = 'VIN format is not recognized';
         } else {
           decodeError = 'Unable to decode vehicle information';
         }

frontend/src/features/vehicles/types/vehicles.types.ts

@@ -72,12 +72,12 @@ export type MatchConfidence = 'high' | 'medium' | 'none';
  */
 export interface MatchedField<T> {
   value: T | null;
-  nhtsaValue: string | null;
+  sourceValue: string | null;
   confidence: MatchConfidence;
 }
 
 /**
- * Decoded vehicle data from NHTSA vPIC API
+ * Decoded vehicle data from VIN decode
  * with match confidence per field
  */
 export interface DecodedVehicleData {

frontend/src/pages/GuidePage/sections/SubscriptionSection.tsx

@@ -43,7 +43,7 @@ export const SubscriptionSection = () => {
       </h3>
 
       <p className="text-titanio/70 leading-relaxed mb-4">
-        <strong className="text-avus">What it does:</strong> Use your device camera to photograph your vehicle's VIN plate, and the system automatically reads the VIN using OCR (Optical Character Recognition) and decodes it from the NHTSA database.
+        <strong className="text-avus">What it does:</strong> Use your device camera to photograph your vehicle's VIN plate, and the system automatically reads the VIN using OCR (Optical Character Recognition) and decodes it from the vehicle database.
       </p>
 
       <p className="text-titanio/70 leading-relaxed mb-4">
@@ -58,7 +58,7 @@ export const SubscriptionSection = () => {
         <li>A <strong className="text-avus">VIN OCR Review modal</strong> appears showing the detected VIN with confidence indicators</li>
         <li>Confirm or correct the VIN, then click <strong className="text-avus">Accept</strong></li>
         <li>Click the <strong className="text-avus">Decode VIN</strong> button</li>
-        <li>The system queries the NHTSA database and auto-populates: Year, Make, Model, Engine, Transmission, and Trim</li>
+        <li>The system queries the vehicle database and auto-populates: Year, Make, Model, Engine, Transmission, and Trim</li>
         <li>Review the pre-filled fields and complete the remaining details</li>
       </ol>
 

frontend/src/pages/GuidePage/sections/VehiclesSection.tsx

@@ -141,7 +141,7 @@ export const VehiclesSection = () => {
         <GuideScreenshot
           src="/guide/vin-decode-desktop.png"
           alt="VIN Decode feature showing auto-populated vehicle specifications"
-          caption="The VIN Decode feature automatically fills in vehicle details from the NHTSA database"
+          caption="The VIN Decode feature automatically fills in vehicle details from the vehicle database"
         />
       </div>
 

ocr/CLAUDE.md

@@ -1,6 +1,6 @@
 # ocr/
 
-Python OCR microservice. Primary engine: PaddleOCR PP-OCRv4 with optional Google Vision cloud fallback. Gemini 2.5 Flash for maintenance manual PDF extraction. Pluggable engine abstraction in `app/engines/`.
+Python OCR microservice. Primary engine: PaddleOCR PP-OCRv4 with optional Google Vision cloud fallback. Gemini 2.5 Flash for maintenance manual PDF extraction and VIN decode. Pluggable engine abstraction in `app/engines/`.
 
 ## Files
 

ocr/app/CLAUDE.md

@@ -7,7 +7,7 @@ Python OCR microservice (FastAPI). Primary engine: PaddleOCR PP-OCRv4 with optio
 | File | What | When to read |
 | ---- | ---- | ------------ |
 | `main.py` | FastAPI application entry point | Route registration, app setup |
-| `config.py` | Configuration settings (OCR engines, Vertex AI, Redis, Vision API limits) | Environment variables, settings |
+| `config.py` | Configuration settings (OCR engines, Google GenAI, Redis, Vision API limits) | Environment variables, settings |
 | `__init__.py` | Package init | Package structure |
 
 ## Subdirectories
@@ -19,7 +19,7 @@ Python OCR microservice (FastAPI). Primary engine: PaddleOCR PP-OCRv4 with optio
 | `models/` | Data models and schemas | Request/response types |
 | `patterns/` | Regex patterns and service name mapping (27 maintenance subtypes) | Pattern matching rules, service categorization |
 | `preprocessors/` | Image preprocessing pipeline | Image preparation before OCR |
-| `routers/` | FastAPI route handlers (/extract, /extract/receipt, /extract/manual, /jobs) | API endpoint changes |
+| `routers/` | FastAPI route handlers (/extract, /extract/receipt, /extract/manual, /decode, /jobs) | API endpoint changes |
 | `services/` | Business logic services (job queue with Redis) | Core OCR processing, async job management |
 | `table_extraction/` | Table detection and parsing | Structured data extraction from images |
 | `validators/` | Input validation | Validation rules |

ocr/app/config.py

@@ -29,10 +29,10 @@ class Settings:
             os.getenv("VISION_MONTHLY_LIMIT", "1000")
         )
 
-        # Vertex AI / Gemini configuration
+        # Google GenAI / Gemini configuration
         self.vertex_ai_project: str = os.getenv("VERTEX_AI_PROJECT", "")
         self.vertex_ai_location: str = os.getenv(
-            "VERTEX_AI_LOCATION", "us-central1"
+            "VERTEX_AI_LOCATION", "global"
         )
         self.gemini_model: str = os.getenv("GEMINI_MODEL", "gemini-2.5-flash")
 

ocr/app/engines/CLAUDE.md

@@ -3,7 +3,7 @@
 OCR engine abstraction layer. Two categories of engines:
 
 1. **OcrEngine subclasses** (image-to-text): PaddleOCR, Google Vision, Hybrid. Accept image bytes, return text + confidence + word boxes.
-2. **GeminiEngine** (PDF-to-structured-data): Standalone module for maintenance schedule extraction via Vertex AI. Accepts PDF bytes, returns structured JSON. Not an OcrEngine subclass because the interface signatures differ.
+2. **GeminiEngine** (PDF-to-structured-data and VIN decode): Standalone module for maintenance schedule extraction and VIN decoding via google-genai SDK. Accepts PDF bytes or VIN strings, returns structured JSON. Not an OcrEngine subclass because the interface signatures differ.
 
 ## Files
 
@@ -15,7 +15,7 @@ OCR engine abstraction layer. Two categories of engines:
 | `cloud_engine.py` | Google Vision TEXT_DETECTION fallback engine (WIF authentication) | Cloud OCR configuration, API quota |
 | `hybrid_engine.py` | Combines primary + fallback engine with confidence threshold switching | Engine selection logic, fallback behavior |
 | `engine_factory.py` | Factory function and engine registry for instantiation | Adding new engine types |
-| `gemini_engine.py` | Gemini 2.5 Flash integration for maintenance schedule extraction (Vertex AI SDK, 20MB PDF limit, structured JSON output) | Manual extraction debugging, Gemini configuration |
+| `gemini_engine.py` | Gemini 2.5 Flash integration for maintenance schedule extraction and VIN decoding (google-genai SDK, 20MB PDF limit, structured JSON output, Google Search grounding for VIN decode) | Manual extraction debugging, VIN decode, Gemini configuration |
 
 ## Engine Selection
 
@@ -30,4 +30,4 @@ create_engine(config)
 HybridEngine (tries primary, falls back if confidence < threshold)
 ```
 
-GeminiEngine is created independently by ManualExtractor, not through the engine factory.
+GeminiEngine is created independently by ManualExtractor and the VIN decode router, not through the engine factory.

ocr/app/engines/gemini_engine.py

@@ -1,14 +1,15 @@
-"""Gemini 2.5 Flash engine for maintenance schedule extraction from PDFs.
+"""Gemini 2.5 Flash engine for document understanding and VIN decode.
 
 Standalone module (does NOT extend OcrEngine) because Gemini performs
 semantic document understanding, not traditional OCR word-box extraction.
-Uses Vertex AI SDK with structured JSON output enforcement.
+Uses google-genai SDK with structured JSON output enforcement.
 """
 
 import json
 import logging
 import os
 from dataclasses import dataclass
+from datetime import datetime
 from typing import Any
 
 from app.config import settings
@@ -37,18 +38,114 @@ Do not include one-time procedures, troubleshooting steps, or warranty informati
 Return the results as a JSON object with a single "maintenanceSchedule" array.\
 """
 
+# VIN year code lookup: position 10 character -> base year (first cycle, 1980-2009).
+# The 30-year cycle repeats: +30 for 2010-2039, +60 for 2040-2069.
+# Disambiguation uses position 7: alphabetic -> 2010+ cycle, numeric -> 1980s cycle.
+# Per NHTSA FMVSS No. 115: MY2010+ vehicles must use alphabetic position 7.
+# For the 2040+ cycle (when position 7 is numeric again), we pick the most
+# recent plausible year (not more than 2 years in the future).
+_VIN_YEAR_CODES: dict[str, int] = {
+    "A": 1980, "B": 1981, "C": 1982, "D": 1983, "E": 1984,
+    "F": 1985, "G": 1986, "H": 1987, "J": 1988, "K": 1989,
+    "L": 1990, "M": 1991, "N": 1992, "P": 1993, "R": 1994,
+    "S": 1995, "T": 1996, "V": 1997, "W": 1998, "X": 1999,
+    "Y": 2000,
+    "1": 2001, "2": 2002, "3": 2003, "4": 2004, "5": 2005,
+    "6": 2006, "7": 2007, "8": 2008, "9": 2009,
+}
+
+
+def resolve_vin_year(vin: str) -> int | None:
+    """Deterministically resolve model year from VIN positions 7 and 10.
+
+    VIN year codes repeat on a 30-year cycle. Position 7 disambiguates:
+      - Alphabetic position 7 -> 2010-2039 cycle (NHTSA MY2010+ requirement)
+      - Numeric position 7 -> 1980-2009 or 2040-2069 cycle
+
+    For the numeric case with two possible cycles, picks the most recent
+    year that is not more than 2 years in the future.
+
+    Returns None if the VIN is too short or position 10 is not a valid year code.
+    """
+    if len(vin) < 17:
+        return None
+
+    code = vin[9].upper()   # position 10 (0-indexed)
+    pos7 = vin[6].upper()   # position 7 (0-indexed)
+
+    base_year = _VIN_YEAR_CODES.get(code)
+    if base_year is None:
+        return None
+
+    if pos7.isalpha():
+        # Alphabetic position 7 -> second cycle (2010-2039)
+        return base_year + 30
+
+    # Numeric position 7 -> first cycle (1980-2009) or third cycle (2040-2069)
+    # Pick the most recent plausible year
+    max_plausible = datetime.now().year + 2
+
+    third_cycle = base_year + 60  # 2040-2069
+    if third_cycle <= max_plausible:
+        return third_cycle
+
+    return base_year
+
+
+_VIN_DECODE_PROMPT = """\
+Decode the following VIN (Vehicle Identification Number) using standard VIN structure rules.
+
+VIN: {vin}
+Model year: {year} (determined from position 10 code '{year_code}')
+
+The model year has already been resolved deterministically. Use {year} as the year.
+
+VIN position reference:
+- Positions 1-3 (WMI): World Manufacturer Identifier (country + manufacturer)
+- Positions 4-8 (VDS): Vehicle attributes (model, body, engine, etc.)
+- Position 9: Check digit
+- Position 10: Model year code (30-year cycle, extended through 2050):
+  A=1980/2010/2040 B=1981/2011/2041 C=1982/2012/2042 D=1983/2013/2043 E=1984/2014/2044
+  F=1985/2015/2045 G=1986/2016/2046 H=1987/2017/2047 J=1988/2018/2048 K=1989/2019/2049
+  L=1990/2020/2050 M=1991/2021       N=1992/2022       P=1993/2023       R=1994/2024
+  S=1995/2025       T=1996/2026       V=1997/2027       W=1998/2028       X=1999/2029
+  Y=2000/2030       1=2001/2031       2=2002/2032       3=2003/2033       4=2004/2034
+  5=2005/2035       6=2006/2036       7=2007/2037       8=2008/2038       9=2009/2039
+- Position 11: Assembly plant
+- Positions 12-17: Sequential production number
+
+Return the vehicle's make, model, trim level, body type, drive type, fuel type, engine description, and transmission type. If a field cannot be determined from the VIN, return null for that field. Return a confidence score (0.0-1.0) indicating overall decode reliability.\
+"""
+
+_VIN_DECODE_SCHEMA: dict[str, Any] = {
+    "type": "OBJECT",
+    "properties": {
+        "year": {"type": "INTEGER", "nullable": True},
+        "make": {"type": "STRING", "nullable": True},
+        "model": {"type": "STRING", "nullable": True},
+        "trimLevel": {"type": "STRING", "nullable": True},
+        "bodyType": {"type": "STRING", "nullable": True},
+        "driveType": {"type": "STRING", "nullable": True},
+        "fuelType": {"type": "STRING", "nullable": True},
+        "engine": {"type": "STRING", "nullable": True},
+        "transmission": {"type": "STRING", "nullable": True},
+        "confidence": {"type": "NUMBER"},
+    },
+    "required": ["confidence"],
+}
+
 _RESPONSE_SCHEMA: dict[str, Any] = {
-    "type": "object",
+    "type": "OBJECT",
     "properties": {
         "maintenanceSchedule": {
-            "type": "array",
+            "type": "ARRAY",
             "items": {
-                "type": "object",
+                "type": "OBJECT",
                 "properties": {
-                    "serviceName": {"type": "string"},
+                    "serviceName": {"type": "STRING"},
-                    "intervalMiles": {"type": "number", "nullable": True},
+                    "intervalMiles": {"type": "NUMBER", "nullable": True},
-                    "intervalMonths": {"type": "number", "nullable": True},
+                    "intervalMonths": {"type": "NUMBER", "nullable": True},
-                    "details": {"type": "string", "nullable": True},
+                    "details": {"type": "STRING", "nullable": True},
                 },
                 "required": ["serviceName"],
             },
@@ -70,6 +167,22 @@ class GeminiProcessingError(GeminiEngineError):
     """Raised when Gemini fails to process a document."""
 
 
+@dataclass
+class VinDecodeResult:
+    """Result from Gemini VIN decode."""
+
+    year: int | None = None
+    make: str | None = None
+    model: str | None = None
+    trim_level: str | None = None
+    body_type: str | None = None
+    drive_type: str | None = None
+    fuel_type: str | None = None
+    engine: str | None = None
+    transmission: str | None = None
+    confidence: float = 0.0
+
+
 @dataclass
 class MaintenanceItem:
     """A single extracted maintenance schedule item."""
@@ -89,25 +202,26 @@ class MaintenanceExtractionResult:
 
 
 class GeminiEngine:
-    """Gemini 2.5 Flash wrapper for maintenance schedule extraction.
+    """Gemini 2.5 Flash wrapper for maintenance schedule extraction and VIN decode.
 
     Standalone class (not an OcrEngine subclass) because Gemini performs
     semantic document understanding rather than traditional OCR.
 
-    Uses lazy initialization: the Vertex AI client is not created until
+    Uses lazy initialization: the Gemini client is not created until
-    the first ``extract_maintenance()`` call.
+    the first call to ``extract_maintenance()`` or ``decode_vin()``.
     """
 
     def __init__(self) -> None:
-        self._model: Any | None = None
+        self._client: Any | None = None
+        self._model_name: str = ""
 
-    def _get_model(self) -> Any:
+    def _get_client(self) -> Any:
-        """Create the GenerativeModel on first use.
+        """Create the genai.Client on first use.
 
         Authentication uses the same WIF credential path as Google Vision.
         """
-        if self._model is not None:
+        if self._client is not None:
-            return self._model
+            return self._client
 
         key_path = settings.google_vision_key_path
         if not os.path.isfile(key_path):
@@ -117,46 +231,37 @@ class GeminiEngine:
             )
 
         try:
-            from google.cloud import aiplatform  # type: ignore[import-untyped]
+            from google import genai  # type: ignore[import-untyped]
-            from vertexai.generative_models import (  # type: ignore[import-untyped]
-                GenerationConfig,
-                GenerativeModel,
-            )
 
-            # Point ADC at the WIF credential config
+            # Point ADC at the WIF credential config (must be set BEFORE Client construction)
             os.environ["GOOGLE_APPLICATION_CREDENTIALS"] = key_path
             os.environ["GOOGLE_EXTERNAL_ACCOUNT_ALLOW_EXECUTABLES"] = "1"
 
-            aiplatform.init(
+            self._client = genai.Client(
+                vertexai=True,
                 project=settings.vertex_ai_project,
                 location=settings.vertex_ai_location,
             )
-
+            self._model_name = settings.gemini_model
-            model_name = settings.gemini_model
-            self._model = GenerativeModel(model_name)
-            self._generation_config = GenerationConfig(
-                response_mime_type="application/json",
-                response_schema=_RESPONSE_SCHEMA,
-            )
 
             logger.info(
                 "Gemini engine initialized (model=%s, project=%s, location=%s)",
-                model_name,
+                self._model_name,
                 settings.vertex_ai_project,
                 settings.vertex_ai_location,
             )
-            return self._model
+            return self._client
 
         except ImportError as exc:
-            logger.exception("Vertex AI SDK import failed")
+            logger.exception("google-genai SDK import failed")
             raise GeminiUnavailableError(
-                "google-cloud-aiplatform is not installed. "
+                "google-genai is not installed. "
-                "Install with: pip install google-cloud-aiplatform"
+                "Install with: pip install google-genai"
             ) from exc
         except Exception as exc:
-            logger.exception("Vertex AI authentication failed")
+            logger.exception("Gemini authentication failed: %s", type(exc).__name__)
             raise GeminiUnavailableError(
-                f"Vertex AI authentication failed: {exc}"
+                f"Gemini authentication failed: {exc}"
             ) from exc
 
     def extract_maintenance(
@@ -181,19 +286,23 @@ class GeminiEngine:
                 "inline processing. Upload to GCS and use a gs:// URI instead."
             )
 
-        model = self._get_model()
+        client = self._get_client()
 
         try:
-            from vertexai.generative_models import Part  # type: ignore[import-untyped]
+            from google.genai import types  # type: ignore[import-untyped]
 
-            pdf_part = Part.from_data(
+            pdf_part = types.Part.from_bytes(
                 data=pdf_bytes,
                 mime_type="application/pdf",
             )
 
-            response = model.generate_content(
+            response = client.models.generate_content(
-                [pdf_part, _EXTRACTION_PROMPT],
+                model=self._model_name,
-                generation_config=self._generation_config,
+                contents=[pdf_part, _EXTRACTION_PROMPT],
+                config=types.GenerateContentConfig(
+                    response_mime_type="application/json",
+                    response_schema=_RESPONSE_SCHEMA,
+                ),
             )
 
             raw = json.loads(response.text)
@@ -228,3 +337,94 @@ class GeminiEngine:
             raise GeminiProcessingError(
                 f"Gemini maintenance extraction failed: {exc}"
             ) from exc
+
+    def decode_vin(self, vin: str) -> VinDecodeResult:
+        """Decode a VIN string into structured vehicle data via Gemini.
+
+        The model year is resolved deterministically from VIN positions 7
+        and 10 -- never delegated to the LLM. Gemini handles make, model,
+        trim, and other fields that require manufacturer knowledge.
+
+        Args:
+            vin: A 17-character Vehicle Identification Number.
+
+        Returns:
+            Structured vehicle specification result.
+
+        Raises:
+            GeminiProcessingError: If Gemini fails to decode the VIN.
+            GeminiUnavailableError: If the engine cannot be initialized.
+        """
+        client = self._get_client()
+
+        # Resolve year deterministically from VIN structure
+        resolved_year = resolve_vin_year(vin)
+        year_code = vin[9].upper() if len(vin) >= 10 else "?"
+        logger.info(
+            "VIN year resolved: code=%s pos7=%s -> year=%s",
+            year_code,
+            vin[6] if len(vin) >= 7 else "?",
+            resolved_year,
+        )
+
+        try:
+            from google.genai import types  # type: ignore[import-untyped]
+
+            prompt = _VIN_DECODE_PROMPT.format(
+                vin=vin,
+                year=resolved_year or "unknown",
+                year_code=year_code,
+            )
+            response = client.models.generate_content(
+                model=self._model_name,
+                contents=[prompt],
+                config=types.GenerateContentConfig(
+                    response_mime_type="application/json",
+                    response_schema=_VIN_DECODE_SCHEMA,
+                    tools=[types.Tool(google_search=types.GoogleSearch())],
+                ),
+            )
+
+            raw = json.loads(response.text)
+
+            # Override year with deterministic value -- never trust the LLM
+            # for a mechanical lookup
+            gemini_year = raw.get("year")
+            if resolved_year and gemini_year != resolved_year:
+                logger.warning(
+                    "Gemini returned year %s but resolved year is %s for VIN %s -- overriding",
+                    gemini_year,
+                    resolved_year,
+                    vin,
+                )
+
+            logger.info(
+                "Gemini decoded VIN %s (confidence=%.2f) raw=%s",
+                vin,
+                raw.get("confidence", 0),
+                json.dumps(raw, default=str),
+            )
+
+            return VinDecodeResult(
+                year=resolved_year if resolved_year else raw.get("year"),
+                make=raw.get("make"),
+                model=raw.get("model"),
+                trim_level=raw.get("trimLevel"),
+                body_type=raw.get("bodyType"),
+                drive_type=raw.get("driveType"),
+                fuel_type=raw.get("fuelType"),
+                engine=raw.get("engine"),
+                transmission=raw.get("transmission"),
+                confidence=raw.get("confidence", 0.0),
+            )
+
+        except (GeminiEngineError,):
+            raise
+        except json.JSONDecodeError as exc:
+            raise GeminiProcessingError(
+                f"Gemini returned invalid JSON for VIN decode: {exc}"
+            ) from exc
+        except Exception as exc:
+            raise GeminiProcessingError(
+                f"Gemini VIN decode failed: {exc}"
+            ) from exc

ocr/app/extractors/maintenance_receipt_extractor.py

@@ -14,6 +14,7 @@ import time
 from typing import Any, Optional
 
 from app.config import settings
+from app.engines.gemini_engine import GeminiUnavailableError
 from app.extractors.receipt_extractor import (
     ExtractedField,
     ReceiptExtractionResult,
@@ -54,16 +55,16 @@ OCR Text:
 """
 
 _RECEIPT_RESPONSE_SCHEMA: dict[str, Any] = {
-    "type": "object",
+    "type": "OBJECT",
     "properties": {
-        "serviceName": {"type": "string", "nullable": True},
+        "serviceName": {"type": "STRING", "nullable": True},
-        "serviceDate": {"type": "string", "nullable": True},
+        "serviceDate": {"type": "STRING", "nullable": True},
-        "totalCost": {"type": "number", "nullable": True},
+        "totalCost": {"type": "NUMBER", "nullable": True},
-        "shopName": {"type": "string", "nullable": True},
+        "shopName": {"type": "STRING", "nullable": True},
-        "laborCost": {"type": "number", "nullable": True},
+        "laborCost": {"type": "NUMBER", "nullable": True},
-        "partsCost": {"type": "number", "nullable": True},
+        "partsCost": {"type": "NUMBER", "nullable": True},
-        "odometerReading": {"type": "number", "nullable": True},
+        "odometerReading": {"type": "NUMBER", "nullable": True},
-        "vehicleInfo": {"type": "string", "nullable": True},
+        "vehicleInfo": {"type": "STRING", "nullable": True},
     },
     "required": [
         "serviceName",
@@ -87,8 +88,8 @@ class MaintenanceReceiptExtractor:
     """
 
     def __init__(self) -> None:
-        self._model: Any | None = None
+        self._client: Any | None = None
-        self._generation_config: Any | None = None
+        self._model_name: str = ""
 
     def extract(
         self,
@@ -169,47 +170,52 @@ class MaintenanceReceiptExtractor:
             processing_time_ms=processing_time_ms,
         )
 
-    def _get_model(self) -> Any:
+    def _get_client(self) -> Any:
-        """Lazy-initialize Vertex AI Gemini model.
+        """Lazy-initialize google-genai Gemini client.
 
         Uses the same authentication pattern as GeminiEngine.
         """
-        if self._model is not None:
+        if self._client is not None:
-            return self._model
+            return self._client
 
         key_path = settings.google_vision_key_path
         if not os.path.isfile(key_path):
-            raise RuntimeError(
+            raise GeminiUnavailableError(
                 f"Google credential config not found at {key_path}. "
                 "Set GOOGLE_VISION_KEY_PATH or mount the secret."
             )
 
-        from google.cloud import aiplatform  # type: ignore[import-untyped]
+        try:
-        from vertexai.generative_models import (  # type: ignore[import-untyped]
+            from google import genai  # type: ignore[import-untyped]
-            GenerationConfig,
-            GenerativeModel,
-        )
 
-        os.environ["GOOGLE_APPLICATION_CREDENTIALS"] = key_path
+            # Point ADC at the WIF credential config (must be set BEFORE Client construction)
-        os.environ["GOOGLE_EXTERNAL_ACCOUNT_ALLOW_EXECUTABLES"] = "1"
+            os.environ["GOOGLE_APPLICATION_CREDENTIALS"] = key_path
+            os.environ["GOOGLE_EXTERNAL_ACCOUNT_ALLOW_EXECUTABLES"] = "1"
 
-        aiplatform.init(
+            self._client = genai.Client(
-            project=settings.vertex_ai_project,
+                vertexai=True,
-            location=settings.vertex_ai_location,
+                project=settings.vertex_ai_project,
-        )
+                location=settings.vertex_ai_location,
+            )
+            self._model_name = settings.gemini_model
 
-        model_name = settings.gemini_model
+            logger.info(
-        self._model = GenerativeModel(model_name)
+                "Maintenance receipt Gemini client initialized (model=%s)",
-        self._generation_config = GenerationConfig(
+                self._model_name,
-            response_mime_type="application/json",
+            )
-            response_schema=_RECEIPT_RESPONSE_SCHEMA,
+            return self._client
-        )
 
-        logger.info(
+        except ImportError as exc:
-            "Maintenance receipt Gemini model initialized (model=%s)",
+            logger.exception("google-genai SDK import failed")
-            model_name,
+            raise GeminiUnavailableError(
-        )
+                "google-genai is not installed. "
-        return self._model
+                "Install with: pip install google-genai"
+            ) from exc
+        except Exception as exc:
+            logger.exception("Gemini authentication failed: %s", type(exc).__name__)
+            raise GeminiUnavailableError(
+                f"Gemini authentication failed: {exc}"
+            ) from exc
 
     def _extract_with_gemini(self, ocr_text: str) -> dict:
         """Send OCR text to Gemini for semantic field extraction.
@@ -220,13 +226,19 @@ class MaintenanceReceiptExtractor:
         Returns:
             Dictionary of field_name -> extracted_value from Gemini.
         """
-        model = self._get_model()
+        client = self._get_client()
+
+        from google.genai import types  # type: ignore[import-untyped]
 
         prompt = _RECEIPT_EXTRACTION_PROMPT.format(ocr_text=ocr_text)
 
-        response = model.generate_content(
+        response = client.models.generate_content(
-            [prompt],
+            model=self._model_name,
-            generation_config=self._generation_config,
+            contents=[prompt],
+            config=types.GenerateContentConfig(
+                response_mime_type="application/json",
+                response_schema=_RECEIPT_RESPONSE_SCHEMA,
+            ),
         )
 
         raw = json.loads(response.text)

ocr/app/main.py

@@ -6,7 +6,7 @@ from typing import AsyncIterator
 from fastapi import FastAPI
 
 from app.config import settings
-from app.routers import extract_router, jobs_router
+from app.routers import decode_router, extract_router, jobs_router
 from app.services import job_queue
 
 # Configure logging
@@ -36,6 +36,7 @@ app = FastAPI(
 )
 
 # Include routers
+app.include_router(decode_router)
 app.include_router(extract_router)
 app.include_router(jobs_router)
 
@@ -54,6 +55,7 @@ async def root() -> dict:
         "version": "1.0.0",
         "log_level": settings.log_level,
         "endpoints": [
+            "POST /decode/vin - VIN string decode via Gemini",
             "POST /extract - Synchronous OCR extraction",
             "POST /extract/vin - VIN-specific extraction with validation",
             "POST /extract/receipt - Receipt extraction (fuel, general)",

ocr/app/models/__init__.py

@@ -14,6 +14,8 @@ from .schemas import (
     ReceiptExtractedField,
     ReceiptExtractionResponse,
     VinAlternative,
+    VinDecodeRequest,
+    VinDecodeResponse,
     VinExtractionResponse,
 )
 
@@ -32,5 +34,7 @@ __all__ = [
     "ReceiptExtractedField",
     "ReceiptExtractionResponse",
     "VinAlternative",
+    "VinDecodeRequest",
+    "VinDecodeResponse",
     "VinExtractionResponse",
 ]

ocr/app/models/schemas.py

@@ -169,3 +169,30 @@ class ManualJobResponse(BaseModel):
     error: Optional[str] = None
 
     model_config = {"populate_by_name": True}
+
+
+class VinDecodeRequest(BaseModel):
+    """Request body for VIN decode endpoint."""
+
+    vin: str
+
+
+class VinDecodeResponse(BaseModel):
+    """Response from VIN decode endpoint."""
+
+    success: bool
+    vin: str
+    year: Optional[int] = None
+    make: Optional[str] = None
+    model: Optional[str] = None
+    trim_level: Optional[str] = Field(default=None, alias="trimLevel")
+    body_type: Optional[str] = Field(default=None, alias="bodyType")
+    drive_type: Optional[str] = Field(default=None, alias="driveType")
+    fuel_type: Optional[str] = Field(default=None, alias="fuelType")
+    engine: Optional[str] = None
+    transmission: Optional[str] = None
+    confidence: float = Field(ge=0.0, le=1.0)
+    processing_time_ms: int = Field(alias="processingTimeMs")
+    error: Optional[str] = None
+
+    model_config = {"populate_by_name": True}

ocr/app/routers/__init__.py

@@ -1,5 +1,6 @@
 """OCR API routers."""
+from .decode import router as decode_router
 from .extract import router as extract_router
 from .jobs import router as jobs_router
 
-__all__ = ["extract_router", "jobs_router"]
+__all__ = ["decode_router", "extract_router", "jobs_router"]

ocr/app/routers/decode.py

@@ -0,0 +1,67 @@
+"""VIN decode router - Gemini-powered VIN string decoding."""
+import logging
+import re
+import time
+
+from fastapi import APIRouter, HTTPException
+
+from app.engines.gemini_engine import (
+    GeminiEngine,
+    GeminiProcessingError,
+    GeminiUnavailableError,
+)
+from app.models import VinDecodeRequest, VinDecodeResponse
+
+logger = logging.getLogger(__name__)
+
+router = APIRouter(prefix="/decode", tags=["decode"])
+
+_VIN_REGEX = re.compile(r"^[A-HJ-NPR-Z0-9]{17}$")
+
+# Shared engine instance (lazy init on first request)
+_gemini_engine = GeminiEngine()
+
+
+@router.post("/vin", response_model=VinDecodeResponse)
+async def decode_vin(request: VinDecodeRequest) -> VinDecodeResponse:
+    """Decode a VIN string into structured vehicle data using Gemini.
+
+    Accepts a 17-character VIN and returns year, make, model, trim, etc.
+    """
+    vin = request.vin.upper().strip()
+
+    if not _VIN_REGEX.match(vin):
+        raise HTTPException(
+            status_code=400,
+            detail=f"Invalid VIN format: must be 17 alphanumeric characters (excluding I, O, Q). Got: {vin}",
+        )
+
+    start_ms = time.monotonic_ns() // 1_000_000
+
+    try:
+        result = _gemini_engine.decode_vin(vin)
+    except GeminiUnavailableError as exc:
+        logger.error("Gemini unavailable for VIN decode: %s", exc)
+        raise HTTPException(status_code=503, detail=str(exc)) from exc
+    except GeminiProcessingError as exc:
+        logger.error("Gemini processing error for VIN %s: %s", vin, exc)
+        raise HTTPException(status_code=422, detail=str(exc)) from exc
+
+    elapsed_ms = (time.monotonic_ns() // 1_000_000) - start_ms
+
+    return VinDecodeResponse(
+        success=True,
+        vin=vin,
+        year=result.year,
+        make=result.make,
+        model=result.model,
+        trimLevel=result.trim_level,
+        bodyType=result.body_type,
+        driveType=result.drive_type,
+        fuelType=result.fuel_type,
+        engine=result.engine,
+        transmission=result.transmission,
+        confidence=result.confidence,
+        processingTimeMs=elapsed_ms,
+        error=None,
+    )

ocr/requirements.txt

@@ -21,8 +21,8 @@ google-cloud-vision>=3.7.0
 # PDF Processing
 PyMuPDF>=1.23.0
 
-# Vertex AI / Gemini (maintenance schedule extraction)
+# Google GenAI / Gemini (maintenance schedule extraction, VIN decode)
-google-cloud-aiplatform>=1.40.0
+google-genai>=1.0.0
 
 # Redis for job queue
 redis>=5.0.0

ocr/tests/test_gemini_engine.py

@@ -2,11 +2,11 @@
 
 Covers: GeminiEngine initialization, PDF size validation,
 successful extraction, empty results, and error handling.
-All Vertex AI SDK calls are mocked.
+All google-genai SDK calls are mocked.
 """
 
 import json
-from unittest.mock import MagicMock, patch, PropertyMock
+from unittest.mock import MagicMock, patch
 
 import pytest
 
@@ -156,22 +156,16 @@ class TestExtractMaintenance:
             },
         ]
 
-        mock_model = MagicMock()
+        mock_client = MagicMock()
-        mock_model.generate_content.return_value = _make_gemini_response(schedule)
+        mock_client.models.generate_content.return_value = _make_gemini_response(schedule)
 
-        with (
+        with patch.dict("sys.modules", {
-            patch(
+            "google.genai": MagicMock(),
-                "app.engines.gemini_engine.importlib_vertex_ai"
+            "google.genai.types": MagicMock(),
-            ) if False else patch.dict("sys.modules", {
+        }):
-                "google.cloud": MagicMock(),
-                "google.cloud.aiplatform": MagicMock(),
-                "vertexai": MagicMock(),
-                "vertexai.generative_models": MagicMock(),
-            }),
-        ):
             engine = GeminiEngine()
-            engine._model = mock_model
+            engine._client = mock_client
-            engine._generation_config = MagicMock()
+            engine._model_name = "gemini-2.5-flash"
 
             result = engine.extract_maintenance(_make_pdf_bytes())
 
@@ -200,12 +194,12 @@ class TestExtractMaintenance:
         mock_settings.vertex_ai_location = "us-central1"
         mock_settings.gemini_model = "gemini-2.5-flash"
 
-        mock_model = MagicMock()
+        mock_client = MagicMock()
-        mock_model.generate_content.return_value = _make_gemini_response([])
+        mock_client.models.generate_content.return_value = _make_gemini_response([])
 
         engine = GeminiEngine()
-        engine._model = mock_model
+        engine._client = mock_client
-        engine._generation_config = MagicMock()
+        engine._model_name = "gemini-2.5-flash"
 
         result = engine.extract_maintenance(_make_pdf_bytes())
 
@@ -223,12 +217,12 @@ class TestExtractMaintenance:
 
         schedule = [{"serviceName": "Brake Fluid Replacement"}]
 
-        mock_model = MagicMock()
+        mock_client = MagicMock()
-        mock_model.generate_content.return_value = _make_gemini_response(schedule)
+        mock_client.models.generate_content.return_value = _make_gemini_response(schedule)
 
         engine = GeminiEngine()
-        engine._model = mock_model
+        engine._client = mock_client
-        engine._generation_config = MagicMock()
+        engine._model_name = "gemini-2.5-flash"
 
         result = engine.extract_maintenance(_make_pdf_bytes())
 
@@ -264,7 +258,8 @@ class TestErrorHandling:
         with (
             patch("app.engines.gemini_engine.settings") as mock_settings,
             patch.dict("sys.modules", {
-                "google.cloud.aiplatform": None,
+                "google": None,
+                "google.genai": None,
             }),
         ):
             mock_settings.google_vision_key_path = "/fake/creds.json"
@@ -283,12 +278,12 @@ class TestErrorHandling:
         mock_settings.vertex_ai_location = "us-central1"
         mock_settings.gemini_model = "gemini-2.5-flash"
 
-        mock_model = MagicMock()
+        mock_client = MagicMock()
-        mock_model.generate_content.side_effect = RuntimeError("API quota exceeded")
+        mock_client.models.generate_content.side_effect = RuntimeError("API quota exceeded")
 
         engine = GeminiEngine()
-        engine._model = mock_model
+        engine._client = mock_client
-        engine._generation_config = MagicMock()
+        engine._model_name = "gemini-2.5-flash"
 
         with pytest.raises(GeminiProcessingError, match="maintenance extraction failed"):
             engine.extract_maintenance(_make_pdf_bytes())
@@ -307,12 +302,12 @@ class TestErrorHandling:
         mock_response = MagicMock()
         mock_response.text = "not valid json {{"
 
-        mock_model = MagicMock()
+        mock_client = MagicMock()
-        mock_model.generate_content.return_value = mock_response
+        mock_client.models.generate_content.return_value = mock_response
 
         engine = GeminiEngine()
-        engine._model = mock_model
+        engine._client = mock_client
-        engine._generation_config = MagicMock()
+        engine._model_name = "gemini-2.5-flash"
 
         with pytest.raises(GeminiProcessingError, match="invalid JSON"):
             engine.extract_maintenance(_make_pdf_bytes())
@@ -322,32 +317,32 @@ class TestErrorHandling:
 
 
 class TestLazyInitialization:
-    """Verify the model is not created until first use."""
+    """Verify the client is not created until first use."""
 
-    def test_model_is_none_after_construction(self):
+    def test_client_is_none_after_construction(self):
-        """GeminiEngine should not initialize the model in __init__."""
+        """GeminiEngine should not initialize the client in __init__."""
         engine = GeminiEngine()
-        assert engine._model is None
+        assert engine._client is None
 
     @patch("app.engines.gemini_engine.settings")
     @patch("app.engines.gemini_engine.os.path.isfile", return_value=True)
-    def test_model_reused_on_second_call(self, mock_isfile, mock_settings):
+    def test_client_reused_on_second_call(self, mock_isfile, mock_settings):
-        """Once initialized, the same model instance is reused."""
+        """Once initialized, the same client instance is reused."""
         mock_settings.google_vision_key_path = "/fake/creds.json"
         mock_settings.vertex_ai_project = "test-project"
         mock_settings.vertex_ai_location = "us-central1"
         mock_settings.gemini_model = "gemini-2.5-flash"
 
         schedule = [{"serviceName": "Oil Change", "intervalMiles": 5000}]
-        mock_model = MagicMock()
+        mock_client = MagicMock()
-        mock_model.generate_content.return_value = _make_gemini_response(schedule)
+        mock_client.models.generate_content.return_value = _make_gemini_response(schedule)
 
         engine = GeminiEngine()
-        engine._model = mock_model
+        engine._client = mock_client
-        engine._generation_config = MagicMock()
+        engine._model_name = "gemini-2.5-flash"
 
         engine.extract_maintenance(_make_pdf_bytes())
         engine.extract_maintenance(_make_pdf_bytes())
 
-        # Model's generate_content should have been called twice
+        # Client's generate_content should have been called twice
-        assert mock_model.generate_content.call_count == 2
+        assert mock_client.models.generate_content.call_count == 2

ocr/tests/test_resolve_vin_year.py

@@ -0,0 +1,122 @@
+"""Tests for deterministic VIN year resolution.
+
+Covers: all three 30-year cycles (1980-2009, 2010-2039, 2040-2050),
+position 7 disambiguation, edge cases, and invalid inputs.
+"""
+
+import pytest
+from unittest.mock import patch
+from datetime import datetime
+
+from app.engines.gemini_engine import resolve_vin_year
+
+
+class TestSecondCycle:
+    """Position 7 alphabetic -> 2010-2039 cycle (NHTSA MY2010+ requirement)."""
+
+    def test_p_with_alpha_pos7_returns_2023(self):
+        """P=2023 when position 7 is alphabetic (the bug that triggered this fix)."""
+        # VIN: 1G1YE2D32P5602473 -- pos7='D' (alphabetic), pos10='P'
+        assert resolve_vin_year("1G1YE2D32P5602473") == 2023
+
+    def test_a_with_alpha_pos7_returns_2010(self):
+        """A=2010 when position 7 is alphabetic."""
+        assert resolve_vin_year("1G1YE2D12A5602473") == 2010
+
+    def test_l_with_alpha_pos7_returns_2020(self):
+        """L=2020 when position 7 is alphabetic."""
+        assert resolve_vin_year("1G1YE2D12L5602473") == 2020
+
+    def test_9_with_alpha_pos7_returns_2039(self):
+        """9=2039 when position 7 is alphabetic."""
+        assert resolve_vin_year("1G1YE2D1295602473") == 2039
+
+    def test_digit_1_with_alpha_pos7_returns_2031(self):
+        """1=2031 when position 7 is alphabetic."""
+        assert resolve_vin_year("1G1YE2D1215602473") == 2031
+
+    def test_s_with_alpha_pos7_returns_2025(self):
+        """S=2025 when position 7 is alphabetic."""
+        assert resolve_vin_year("1G1YE2D12S5602473") == 2025
+
+    def test_t_with_alpha_pos7_returns_2026(self):
+        """T=2026 when position 7 is alphabetic."""
+        assert resolve_vin_year("1G1YE2D12T5602473") == 2026
+
+
+class TestFirstCycle:
+    """Position 7 numeric -> 1980-2009 cycle."""
+
+    def test_m_with_numeric_pos7_returns_1991(self):
+        """M=1991 when position 7 is numeric."""
+        assert resolve_vin_year("1G1YE2132M5602473") == 1991
+
+    def test_n_with_numeric_pos7_returns_1992(self):
+        """N=1992 when position 7 is numeric."""
+        assert resolve_vin_year("1G1YE2132N5602473") == 1992
+
+    def test_p_with_numeric_pos7_returns_1993(self):
+        """P=1993 when position 7 is numeric."""
+        assert resolve_vin_year("1G1YE2132P5602473") == 1993
+
+    def test_y_with_numeric_pos7_returns_2000(self):
+        """Y=2000 when position 7 is numeric."""
+        assert resolve_vin_year("1G1YE2132Y5602473") == 2000
+
+
+class TestThirdCycle:
+    """Position 7 numeric + third cycle year (2040-2050) is plausible."""
+
+    @patch("app.engines.gemini_engine.datetime")
+    def test_a_with_numeric_pos7_returns_2040_when_plausible(self, mock_dt):
+        """A=2040 when position 7 is numeric and year 2040 is plausible."""
+        mock_dt.now.return_value = datetime(2039, 1, 1)
+        # 2039 + 2 = 2041 >= 2040, so third cycle is plausible
+        assert resolve_vin_year("1G1YE2132A5602473") == 2040
+
+    @patch("app.engines.gemini_engine.datetime")
+    def test_l_with_numeric_pos7_returns_2050_when_plausible(self, mock_dt):
+        """L=2050 when position 7 is numeric and year 2050 is plausible."""
+        mock_dt.now.return_value = datetime(2049, 6, 1)
+        assert resolve_vin_year("1G1YE2132L5602473") == 2050
+
+    @patch("app.engines.gemini_engine.datetime")
+    def test_a_with_numeric_pos7_returns_1980_when_2040_not_plausible(self, mock_dt):
+        """A=1980 when third cycle year (2040) exceeds max plausible."""
+        mock_dt.now.return_value = datetime(2026, 2, 20)
+        # 2026 + 2 = 2028 < 2040, so third cycle not plausible -> first cycle
+        assert resolve_vin_year("1G1YE2132A5602473") == 1980
+
+    @patch("app.engines.gemini_engine.datetime")
+    def test_k_with_numeric_pos7_returns_2049_when_plausible(self, mock_dt):
+        """K=2049 when position 7 is numeric and year is plausible."""
+        mock_dt.now.return_value = datetime(2048, 1, 1)
+        assert resolve_vin_year("1G1YE2132K5602473") == 2049
+
+
+class TestEdgeCases:
+    """Invalid inputs and boundary conditions."""
+
+    def test_short_vin_returns_none(self):
+        """VIN shorter than 17 chars returns None."""
+        assert resolve_vin_year("1G1YE2D32") is None
+
+    def test_empty_vin_returns_none(self):
+        """Empty string returns None."""
+        assert resolve_vin_year("") is None
+
+    def test_invalid_year_code_returns_none(self):
+        """Position 10 with invalid code (e.g., 'Z') returns None."""
+        # Z is not a valid year code
+        assert resolve_vin_year("1G1YE2D32Z5602473") is None
+
+    def test_lowercase_vin_handled(self):
+        """Lowercase VIN characters are handled correctly."""
+        assert resolve_vin_year("1g1ye2d32p5602473") == 2023
+
+    def test_i_o_q_not_valid_year_codes(self):
+        """Letters I, O, Q are not valid VIN year codes."""
+        # These are excluded from VINs entirely but test graceful handling
+        assert resolve_vin_year("1G1YE2D32I5602473") is None
+        assert resolve_vin_year("1G1YE2D32O5602473") is None
+        assert resolve_vin_year("1G1YE2D32Q5602473") is None

ocr/tests/test_vin_decode.py

@@ -0,0 +1,199 @@
+"""Tests for the VIN decode endpoint (POST /decode/vin).
+
+Covers: valid VIN returns 200 with correct response shape,
+invalid VIN format returns 400, Gemini unavailable returns 503,
+and Gemini processing error returns 422.
+All GeminiEngine calls are mocked.
+"""
+
+from unittest.mock import MagicMock, patch
+
+import pytest
+from fastapi.testclient import TestClient
+
+from app.engines.gemini_engine import (
+    GeminiProcessingError,
+    GeminiUnavailableError,
+    VinDecodeResult,
+)
+from app.main import app
+
+client = TestClient(app)
+
+# A valid 17-character VIN (no I, O, Q)
+_VALID_VIN = "1HGBH41JXMN109186"
+
+_FULL_RESULT = VinDecodeResult(
+    year=2021,
+    make="Honda",
+    model="Civic",
+    trim_level="EX",
+    body_type="Sedan",
+    drive_type="FWD",
+    fuel_type="Gasoline",
+    engine="2.0L I4",
+    transmission="CVT",
+    confidence=0.95,
+)
+
+
+# --- Valid VIN ---
+
+
+class TestDecodeVinSuccess:
+    """Verify successful VIN decode returns 200 with correct response shape."""
+
+    @patch("app.routers.decode._gemini_engine")
+    def test_valid_vin_returns_200(self, mock_engine):
+        """Normal: Valid VIN returns 200 with all vehicle fields populated."""
+        mock_engine.decode_vin.return_value = _FULL_RESULT
+
+        response = client.post("/decode/vin", json={"vin": _VALID_VIN})
+
+        assert response.status_code == 200
+        data = response.json()
+        assert data["success"] is True
+        assert data["vin"] == _VALID_VIN
+        assert data["year"] == 2021
+        assert data["make"] == "Honda"
+        assert data["model"] == "Civic"
+        assert data["trimLevel"] == "EX"
+        assert data["bodyType"] == "Sedan"
+        assert data["driveType"] == "FWD"
+        assert data["fuelType"] == "Gasoline"
+        assert data["engine"] == "2.0L I4"
+        assert data["transmission"] == "CVT"
+        assert data["confidence"] == 0.95
+        assert "processingTimeMs" in data
+        assert data["error"] is None
+
+    @patch("app.routers.decode._gemini_engine")
+    def test_vin_uppercased_before_decode(self, mock_engine):
+        """VIN submitted in lowercase is normalised to uppercase before decoding."""
+        mock_engine.decode_vin.return_value = _FULL_RESULT
+
+        response = client.post("/decode/vin", json={"vin": _VALID_VIN.lower()})
+
+        assert response.status_code == 200
+        data = response.json()
+        assert data["vin"] == _VALID_VIN
+        mock_engine.decode_vin.assert_called_once_with(_VALID_VIN)
+
+    @patch("app.routers.decode._gemini_engine")
+    def test_nullable_fields_allowed(self, mock_engine):
+        """Edge: VIN decode with only confidence set returns valid response."""
+        mock_engine.decode_vin.return_value = VinDecodeResult(confidence=0.3)
+
+        response = client.post("/decode/vin", json={"vin": _VALID_VIN})
+
+        assert response.status_code == 200
+        data = response.json()
+        assert data["success"] is True
+        assert data["year"] is None
+        assert data["make"] is None
+        assert data["confidence"] == 0.3
+
+
+# --- Invalid VIN format ---
+
+
+class TestDecodeVinValidation:
+    """Verify invalid VIN formats return 400."""
+
+    def test_too_short_vin_returns_400(self):
+        """VIN shorter than 17 characters is rejected."""
+        response = client.post("/decode/vin", json={"vin": "1HGBH41JXM"})
+
+        assert response.status_code == 400
+        assert "Invalid VIN format" in response.json()["detail"]
+
+    def test_too_long_vin_returns_400(self):
+        """VIN longer than 17 characters is rejected."""
+        response = client.post("/decode/vin", json={"vin": "1HGBH41JXMN109186X"})
+
+        assert response.status_code == 400
+
+    def test_vin_with_letter_i_returns_400(self):
+        """VIN containing the letter I (invalid character) is rejected."""
+        # Replace position 0 with I to create invalid VIN
+        invalid_vin = "IHGBH41JXMN109186"
+        response = client.post("/decode/vin", json={"vin": invalid_vin})
+
+        assert response.status_code == 400
+        assert "Invalid VIN format" in response.json()["detail"]
+
+    def test_vin_with_letter_o_returns_400(self):
+        """VIN containing the letter O (invalid character) is rejected."""
+        invalid_vin = "OHGBH41JXMN109186"
+        response = client.post("/decode/vin", json={"vin": invalid_vin})
+
+        assert response.status_code == 400
+
+    def test_vin_with_letter_q_returns_400(self):
+        """VIN containing the letter Q (invalid character) is rejected."""
+        invalid_vin = "QHGBH41JXMN109186"
+        response = client.post("/decode/vin", json={"vin": invalid_vin})
+
+        assert response.status_code == 400
+
+    def test_empty_vin_returns_400(self):
+        """Empty VIN string is rejected."""
+        response = client.post("/decode/vin", json={"vin": ""})
+
+        assert response.status_code == 400
+
+    def test_vin_with_special_chars_returns_400(self):
+        """VIN containing special characters is rejected."""
+        response = client.post("/decode/vin", json={"vin": "1HGBH41J-MN109186"})
+
+        assert response.status_code == 400
+
+
+# --- Gemini unavailable ---
+
+
+class TestDecodeVinGeminiUnavailable:
+    """Verify Gemini service unavailability returns 503."""
+
+    @patch("app.routers.decode._gemini_engine")
+    def test_gemini_unavailable_returns_503(self, mock_engine):
+        """When Gemini cannot be initialized, endpoint returns 503."""
+        mock_engine.decode_vin.side_effect = GeminiUnavailableError(
+            "Google credential config not found"
+        )
+
+        response = client.post("/decode/vin", json={"vin": _VALID_VIN})
+
+        assert response.status_code == 503
+        assert "Google credential config not found" in response.json()["detail"]
+
+
+# --- Gemini processing error ---
+
+
+class TestDecodeVinGeminiProcessingError:
+    """Verify Gemini processing failures return 422."""
+
+    @patch("app.routers.decode._gemini_engine")
+    def test_gemini_processing_error_returns_422(self, mock_engine):
+        """When Gemini returns invalid output, endpoint returns 422."""
+        mock_engine.decode_vin.side_effect = GeminiProcessingError(
+            "Gemini returned invalid JSON for VIN decode: ..."
+        )
+
+        response = client.post("/decode/vin", json={"vin": _VALID_VIN})
+
+        assert response.status_code == 422
+        assert "Gemini returned invalid JSON" in response.json()["detail"]
+
+    @patch("app.routers.decode._gemini_engine")
+    def test_gemini_api_failure_returns_422(self, mock_engine):
+        """When Gemini API call fails at runtime, endpoint returns 422."""
+        mock_engine.decode_vin.side_effect = GeminiProcessingError(
+            "Gemini VIN decode failed: API quota exceeded"
+        )
+
+        response = client.post("/decode/vin", json={"vin": _VALID_VIN})
+
+        assert response.status_code == 422
+        assert "Gemini VIN decode failed" in response.json()["detail"]

scripts/ci/generate-log-config.sh

@@ -1,11 +1,12 @@
 #!/bin/bash
-# generate-log-config.sh - Generate .env.logging from LOG_LEVEL
+# generate-log-config.sh - Generate log-level environment variables
-# Maps a single LOG_LEVEL environment variable to per-container settings
+# Maps a single LOG_LEVEL to per-container settings and writes to stdout
 #
 # Usage: ./generate-log-config.sh [LOG_LEVEL]
 #   LOG_LEVEL: DEBUG, INFO, WARN, or ERROR (default: INFO)
 #
-# Output: Creates .env.logging file with container-specific log settings
+# Output: Log configuration variables on stdout (append to .env)
+#   Example: ./generate-log-config.sh INFO >> .env
 #
 # Exit codes:
 #   0 - Configuration generated successfully
@@ -43,27 +44,13 @@ case "$LOG_LEVEL" in
   ERROR) REDIS_LOGLEVEL="warning" ;;
 esac
 
-# Generate .env.logging file
+# Output log configuration to stdout
-cat > .env.logging << EOF
+cat << EOF
-# Generated by generate-log-config.sh - DO NOT EDIT MANUALLY
-# Regenerate with: ./scripts/ci/generate-log-config.sh $LOG_LEVEL
-LOG_LEVEL=$LOG_LEVEL
 
-# Backend/OCR (Pino)
+# Log levels (generated by generate-log-config.sh $LOG_LEVEL)
 BACKEND_LOG_LEVEL=$LOG_LEVEL_LOWER
-
+TRAEFIK_LOG_LEVEL=$LOG_LEVEL
-# Frontend (Vite)
-VITE_LOG_LEVEL=$LOG_LEVEL_LOWER
-
-# PostgreSQL
 POSTGRES_LOG_STATEMENT=$POSTGRES_LOG_STATEMENT
 POSTGRES_LOG_MIN_DURATION=$POSTGRES_LOG_MIN_DURATION
-
-# Redis
 REDIS_LOGLEVEL=$REDIS_LOGLEVEL
-
-# Traefik
-TRAEFIK_LOG_LEVEL=$LOG_LEVEL
 EOF
-
-echo "Generated .env.logging with LOG_LEVEL=$LOG_LEVEL"

scripts/ci/notify.sh

@@ -174,7 +174,7 @@ echo "  Subject: $SUBJECT"
 # Build JSON payload
 JSON_PAYLOAD=$(cat <<EOF
 {
-  "from": "MotoVaultPro <deploy@motovaultpro.com>",
+  "from": "MotoVaultPro <hello@notify.motovaultpro.com>",
   "to": ["$NOTIFY_EMAIL"],
   "subject": "$SUBJECT",
   "html": $(echo "$HTML_BODY" | jq -Rs .)