Fix AI provider reset, improve OCR model detection, and UI toggle

config/changelog.js

@@ -51,6 +51,15 @@ const RELEASES = [
       'Removed: Legacy data/.env migration notice on the settings page',
     ],
   },
+  {
+    version: 'v2026.07.04',
+    entries: [
+      'Fix: Quickstart OCR detection now suggests and lists dedicated OCR models (e.g. Mistral\'s mistral-ocr-latest) instead of requiring vision-capable naming, in both the Setup Wizard and Settings page',
+      'Fix: Setup wizard no longer leaves a stale AI provider selected when switching from Quickstart to manual AI configuration',
+      'Fix: AI response/prompt log files resolve relative to the working directory on native (non-Docker) installs',
+      'Improvement: Quickstart\'s "use this service for OCR" option is now a proper ON/OFF switch, matching the rest of the settings UI',
+    ],
+  },
 ];
 
 const latestRelease = RELEASES[RELEASES.length - 1];

config/config.js

@@ -352,7 +352,7 @@ startupLog(logLevel, 'info', 'Configuration loaded:', {
 });
 
 module.exports = {
-  PAPERLESS_AI_VERSION: 'v2026.07.03',
+  PAPERLESS_AI_VERSION: 'v2026.07.04',
   CONFIGURED: false,
   configSourceMode: CONFIG_SOURCE_MODE,
   getApiKey,

public/js/settings.js

@@ -1193,9 +1193,13 @@ function initializeFormHandlers() {
         const textModels = Array.isArray(quickstartDetection?.textModels)
           ? quickstartDetection.textModels
           : [];
-        const visionModels = Array.isArray(quickstartDetection?.visionModels)
-          ? quickstartDetection.visionModels
-          : [];
+        // Every detected model that isn't embedding-only is a valid OCR
+        // candidate (textModels already excludes embedding-only models).
+        // Gating on the vision name-heuristic instead misses real
+        // OCR-capable models with unfamiliar names, e.g. Mistral's
+        // dedicated mistral-ocr-* models. Mirrors the Quickstart fix in
+        // public/js/setup.js.
+        const ocrCandidateModels = textModels;
 
         populateModelSelect(
           quickstartAiModelSelect,
@@ -1210,10 +1214,10 @@ function initializeFormHandlers() {
 
         populateModelSelect(
           quickstartOcrModelSelect,
-          visionModels,
-          visionModels.length > 0
+          ocrCandidateModels,
+          ocrCandidateModels.length > 0
             ? 'Select OCR model'
-            : 'No vision-capable models found'
+            : 'No models found'
         );
         if (
           quickstartDetection?.suggestedOcrModel &&
@@ -1224,8 +1228,8 @@ function initializeFormHandlers() {
         }
 
         if (quickstartEnableOcrCheckbox) {
-          quickstartEnableOcrCheckbox.disabled = visionModels.length === 0;
-          quickstartEnableOcrCheckbox.checked = visionModels.length > 0;
+          quickstartEnableOcrCheckbox.disabled = ocrCandidateModels.length === 0;
+          quickstartEnableOcrCheckbox.checked = ocrCandidateModels.length > 0;
         }
 
         if (quickstartHint) {
@@ -1329,7 +1333,9 @@ function initializeFormHandlers() {
 
     if (applyOcr && selectedOcrModel) {
       setSwitchValue(ocrEnabledSelect, 'yes');
-      if (ocrProviderSelect) ocrProviderSelect.value = 'custom';
+      if (ocrProviderSelect) {
+        ocrProviderSelect.value = quickstartDetection.ocrProvider || 'custom';
+      }
       if (ocrApiUrlInput)
         ocrApiUrlInput.value = String(
           quickstartDetection.resolvedOcrApiUrl || ''

public/js/setup.js

@@ -1,3 +1,4 @@
+/* global Swal */
 class SetupWizard {
     constructor() {
         this.bootstrap = window.__SETUP_BOOTSTRAP__ || {};
@@ -926,6 +927,7 @@ class SetupWizard {
 
     applyPreset(preset) {
         if (!preset) {
+            this.aiProvider.value = 'custom';
             this.aiPresetHint.textContent = 'Manual mode: choose provider and enter values yourself. Token is optional for custom endpoints.';
             return;
         }
@@ -979,28 +981,35 @@ class SetupWizard {
             this.quickstartState.detection = detection;
 
             const textModels = Array.isArray(detection.textModels) ? detection.textModels : [];
-            const visionModels = Array.isArray(detection.visionModels) ? detection.visionModels : [];
+            // Every detected model that isn't embedding-only is a valid OCR
+            // candidate (textModels already excludes embedding-only models -
+            // see quickstartService.classifyModelName). Rather than also
+            // gating on the vision name-heuristic (which can miss real
+            // OCR-capable models with unfamiliar names, e.g. dedicated OCR
+            // models), offer the full list here and let the user pick; a
+            // heuristic "suggested" default is still pre-selected below.
+            const ocrCandidateModels = textModels;
 
             this.setModelSelectOptions(this.quickstartAiModel, textModels, textModels.length > 0 ? 'Select AI model' : 'No text-capable models found');
             this.quickstartAiModel.disabled = textModels.length === 0;
             if (detection.suggestedAiModel) {
                 this.quickstartAiModel.value = detection.suggestedAiModel;
             }
 
-            this.setModelSelectOptions(this.quickstartOcrModel, visionModels, visionModels.length > 0 ? 'Select OCR model' : 'No vision-capable models found');
-            this.quickstartOcrModel.disabled = visionModels.length === 0;
+            this.setModelSelectOptions(this.quickstartOcrModel, ocrCandidateModels, ocrCandidateModels.length > 0 ? 'Select OCR model' : 'No models found');
+            this.quickstartOcrModel.disabled = ocrCandidateModels.length === 0;
             if (detection.suggestedOcrModel) {
                 this.quickstartOcrModel.value = detection.suggestedOcrModel;
             }
 
-            const hasVisionModels = visionModels.length > 0;
-            this.quickstartEnableOcr.disabled = !hasVisionModels;
-            this.quickstartEnableOcr.checked = hasVisionModels;
+            const hasOcrCandidateModels = ocrCandidateModels.length > 0;
+            this.quickstartEnableOcr.disabled = !hasOcrCandidateModels;
+            this.quickstartEnableOcr.checked = hasOcrCandidateModels;
             if (this.quickstartOcrHint) {
-                this.quickstartOcrHint.classList.toggle('hidden', hasVisionModels);
-                this.quickstartOcrHint.textContent = hasVisionModels
+                this.quickstartOcrHint.classList.toggle('hidden', hasOcrCandidateModels);
+                this.quickstartOcrHint.textContent = hasOcrCandidateModels
                     ? ''
-                    : 'No vision-capable model found — OCR fallback stays disabled.';
+                    : 'No models found — OCR fallback stays disabled.';
             }
 
             if (this.quickstartHint) {
@@ -1066,7 +1075,7 @@ class SetupWizard {
 
         if (enableOcr && selectedOcrModel) {
             this.mistralOcrEnabled.value = 'yes';
-            this.ocrProvider.value = 'custom';
+            this.ocrProvider.value = detection.ocrProvider || 'custom';
             this.ocrApiUrl.value = detection.resolvedOcrApiUrl || '';
             this.ocrApiKey.value = quickstartKey;
             this.setModelSelectOptions(this.mistralOcrModel, [selectedOcrModel], 'Select OCR model');
@@ -1665,7 +1674,7 @@ class SetupWizard {
         try {
             await navigator.clipboard.writeText(this.envPreview.value);
             await this.showPopup({ icon: 'success', title: 'Copied', text: 'Environment keys copied to clipboard.' });
-        } catch (_error) {
+        } catch {
             this.envPreview.select();
             document.execCommand('copy');
             await this.showPopup({ icon: 'success', title: 'Copied', text: 'Environment keys copied to clipboard.' });
@@ -1707,7 +1716,6 @@ class SetupWizard {
     async finalizeSetup(options = {}) {
         const validations = [];
         for (let index = 0; index <= 5; index += 1) {
-            // eslint-disable-next-line no-await-in-loop
             const valid = await this.validateStepBeforeContinue(index);
             if (!valid) {
                 validations.push(index);

scripts/run-tests.js

@@ -38,6 +38,7 @@ const TESTS = {
   'injected-env-priority': 'test-injected-env-priority.js',
   'log-level-config': 'test-log-level-config.js',
   'log-level-logger': 'test-log-level-logger.js',
+  'native-install-log-paths': 'test-native-install-log-paths.js',
   'login-mfa-flow': 'test-login-mfa-flow.js',
   'ocr-fallback-ai-errors': 'test-ocr-fallback-ai-errors.js',
   'ocr-startup-recovery': 'test-ocr-startup-recovery.js',
@@ -46,6 +47,7 @@ const TESTS = {
   'quickstart-model-classification': 'test-quickstart-model-classification.js',
   'quickstart-endpoint-protection': 'test-quickstart-endpoint-protection.js',
   'setup-wizard-quickstart': 'test-setup-wizard-quickstart.js',
+  'setup-preset-manual-reset': 'test-setup-preset-manual-reset.js',
   'rate-limiting': 'test-rate-limiting.js',
   'scan-stop-flow': 'test-scan-stop-flow.js',
   'setup-auth-endpoint-protection': 'test-setup-auth-endpoint-protection.js',
@@ -102,7 +104,11 @@ const AREAS = {
     'setup-ocr-disabled-skip',
     'setupservice-ocr-validation',
   ],
-  observability: ['log-level-config', 'log-level-logger'],
+  observability: [
+    'log-level-config',
+    'log-level-logger',
+    'native-install-log-paths',
+  ],
   processing: [
     'document-type-restriction',
     'ignore-tags-filter',
@@ -126,6 +132,7 @@ const AREAS = {
     'quickstart-endpoint-protection',
     'runtime-first-setup-state',
     'setup-wizard-tag-default',
+    'setup-preset-manual-reset',
   ],
   security: [
     'setup-remote-guard',

services/azureService.js

@@ -6,7 +6,7 @@
   extractChatMessageContent
 } = require('./serviceUtils');
 const axios = require('axios');
 const OpenAI = require('openai');
 const AzureOpenAI = require('openai').AzureOpenAI;
 const config = require('../config/config');
 const paperlessService = require('./paperlessService');
@@ -14,7 +14,7 @@
 const path = require('path');
 const { THUMBNAIL_CACHE_DIR, getThumbnailCachePath } = require('./thumbnailCachePaths');
 const RestrictionPromptService = require('./restrictionPromptService');
-const responseLogPath = path.join('/app', 'data', 'logs', 'response.txt');
+const responseLogPath = path.join(process.cwd(), 'data', 'logs', 'response.txt');
 
 class AzureOpenAIService {
   constructor() {
@@ -47,7 +47,7 @@
       try {
         await fs.access(cachePath);
         console.log('[DEBUG] Thumbnail already cached');
       } catch (err) {
         console.log('Thumbnail not cached, fetching from Paperless');
 
         const thumbnailData = await paperlessService.getThumbnailImage(id);
@@ -219,7 +219,7 @@
         parsedResponse = JSON.parse(jsonContent);
       } catch (error) {
         console.error(`Failed to parse JSON response: ${error.message}`); console.debug(error);
         throw new Error('Invalid JSON response from API');
       }
 
       try {
@@ -352,7 +352,7 @@
         parsedResponse = JSON.parse(jsonContent);
       } catch (error) {
         console.error(`Failed to parse JSON response: ${error.message}`); console.debug(error);
         throw new Error('Invalid JSON response from API');
       }
 
       // Validate response structure

services/customService.js

@@ -2,20 +2,20 @@
   calculateTokens,
   calculateTotalPromptTokens,
   truncateToTokenLimit,
   writePromptToFile,
   extractChatMessageContent,
   isTimeoutError,
   buildTimeoutErrorMessage
 } = require('./serviceUtils');
 const OpenAI = require('openai');
 const config = require('../config/config');
 const tiktoken = require('tiktoken');
 const paperlessService = require('./paperlessService');
 const fs = require('fs').promises;
 const path = require('path');
 const { THUMBNAIL_CACHE_DIR, getThumbnailCachePath } = require('./thumbnailCachePaths');
 const RestrictionPromptService = require('./restrictionPromptService');
-const responseLogPath = path.join('/app', 'data', 'logs', 'response.txt');
+const responseLogPath = path.join(process.cwd(), 'data', 'logs', 'response.txt');
 const CUSTOM_PROVIDER_FALLBACK_API_KEY = 'no-auth-required';
 
 class CustomOpenAIService {
@@ -96,10 +96,10 @@
         parsed: JSON.parse(sanitizedContent),
         normalized: sanitizedContent
       };
     } catch (_directParseError) {
       const extractedJson = this._extractFirstJsonValue(sanitizedContent);
       if (!extractedJson) {
         throw new Error('Invalid JSON response from API');
       }
 
       try {
@@ -107,8 +107,8 @@
           parsed: JSON.parse(extractedJson),
           normalized: extractedJson
         };
       } catch (_extractedParseError) {
         throw new Error('Invalid JSON response from API');
       }
     }
   }

services/openaiService.js

@@ -13,7 +13,7 @@ const path = require('path');
 const { THUMBNAIL_CACHE_DIR, getThumbnailCachePath } = require('./thumbnailCachePaths');
 const { model } = require('./ollamaService');
 const RestrictionPromptService = require('./restrictionPromptService');
-const responseLogPath = path.join('/app', 'data', 'logs', 'response.txt');
+const responseLogPath = path.join(process.cwd(), 'data', 'logs', 'response.txt');
 
 class OpenAIService {
   constructor() {

services/quickstartService.js

@@ -36,6 +36,13 @@ class QuickstartService {
     // classified as vision. The suggestion is a dropdown default the user
     // can always override.
     this.visionNameHints = ['llava', 'bakllava', '-vl', 'vl-', 'vlm', 'vision', 'minicpm-v', 'moondream', 'pixtral', 'gemma3', 'internvl', 'smolvlm'];
+    // Dedicated OCR models (e.g. Mistral's mistral-ocr-latest) classify as
+    // plain ['text'] under classifyModelName - none of the vision hints
+    // above match "ocr" naming. suggestModels() needs its own hint list so
+    // these still get suggested as the default OCR model instead of being
+    // invisible to the suggestion logic (see classifyModelName's untouched
+    // hint lists; this only affects which text-capable model is suggested).
+    this.ocrNameHints = ['ocr'];
   }
 
   // ── Classification helpers (pure, offline-testable) ──────────────────────
@@ -112,6 +119,9 @@ class QuickstartService {
   suggestModels(classifiedModels = []) {
     const textCandidates = classifiedModels.filter((model) => model.capabilities.includes('text'));
     const visionCandidates = classifiedModels.filter((model) => model.capabilities.includes('vision'));
+    const ocrNameCandidates = textCandidates.filter(
+      (model) => this.ocrNameHints.some((hint) => model.id.toLowerCase().includes(hint))
+    );
 
     let suggestedAiModel = null;
     let bestAiScore = -1;
@@ -139,10 +149,14 @@ class QuickstartService {
       suggestedAiModel = visionCandidates[0].id;
     }
 
+    // Dedicated OCR models (named "*ocr*") are purpose-built for this and
+    // take priority over generic vision models when present.
+    const ocrCandidates = ocrNameCandidates.length > 0 ? ocrNameCandidates : visionCandidates;
+
     let suggestedOcrModel = null;
     let bestOcrScore = -1;
     let bestOcrSize = Number.POSITIVE_INFINITY;
-    visionCandidates.forEach((model) => {
+    ocrCandidates.forEach((model) => {
       const score = model.state === 'loaded' ? 2 : 0;
       const size = this.parseParameterSizeBillions(model.parameterSize);
       const effectiveSize = size == null ? Number.POSITIVE_INFINITY : size;
@@ -181,6 +195,17 @@ class QuickstartService {
     return headers;
   }
 
+  // Which OCR provider to default the wizard to. This is deliberately not a
+  // per-model capability guess (those can miss real OCR-capable models with
+  // unfamiliar names); it only checks the one endpoint this codebase already
+  // knows has a dedicated /ocr path (see
+  // setupService.getMistralUrlValidationOptions). The "OCR fallback" wizard
+  // step always lets the user override this before finishing setup.
+  resolveOcrProviderDefault(bareBaseUrl) {
+    const normalized = String(bareBaseUrl || '').trim().toLowerCase();
+    return normalized.includes('api.mistral.ai') ? 'mistral' : 'custom';
+  }
+
   buildLoopbackBlockedError(validationError) {
     return new Error(
       `URL validation failed: ${validationError} — localhost URLs are blocked by default. `
@@ -371,9 +396,7 @@ class QuickstartService {
       flavor: probeResult.flavor,
       aiProvider: isOllama ? 'ollama' : 'custom',
       resolvedAiApiUrl: isOllama ? urls.bareBaseUrl : urls.versionedBaseUrl,
-      // The wizard/settings OCR provider is always "custom" for local
-      // endpoints (the backend aliases custom -> ollama internally).
-      ocrProvider: 'custom',
+      ocrProvider: this.resolveOcrProviderDefault(urls.bareBaseUrl),
       resolvedOcrApiUrl: isOllama ? urls.bareBaseUrl : urls.versionedBaseUrl,
       models: models.map((m) => ({
         id: m.id,

services/serviceUtils.js

@@ -163,7 +163,7 @@ async function truncateToTokenLimit(text, maxTokens, model = process.env.OPENAI_
 }
 
 // Write prompt and content to a file with size management
-async function writePromptToFile(systemPrompt, truncatedContent, filePath = '/app/data/logs/prompt.txt', maxSize = 10 * 1024 * 1024) {
+async function writePromptToFile(systemPrompt, truncatedContent, filePath = path.join(process.cwd(), 'data', 'logs', 'prompt.txt'), maxSize = 10 * 1024 * 1024) {
     try {
         // Ensure the logs directory exists
         await fs.mkdir(path.dirname(filePath), { recursive: true });

tests/test-native-install-log-paths.js

@@ -0,0 +1,63 @@
+const fs = require('fs');
+const path = require('path');
+
+function assertIncludes(content, snippet, message) {
+  if (!content.includes(snippet)) {
+    throw new Error(message);
+  }
+}
+
+function assertNotIncludes(content, snippet, message) {
+  if (content.includes(snippet)) {
+    throw new Error(message);
+  }
+}
+
+function run() {
+  console.log('\n=== Native Install Log Path Checks ===');
+
+  const filesUsingResponseLogPath = [
+    'services/openaiService.js',
+    'services/customService.js',
+    'services/azureService.js',
+  ];
+
+  filesUsingResponseLogPath.forEach((relativePath) => {
+    const fullPath = path.join(process.cwd(), relativePath);
+    const content = fs.readFileSync(fullPath, 'utf8');
+
+    assertNotIncludes(
+      content,
+      "path.join('/app'",
+      `${relativePath} must not hardcode the Docker '/app' path for AI response logging`
+    );
+    assertIncludes(
+      content,
+      'path.join(process.cwd()',
+      `${relativePath} must resolve the AI response log path relative to process.cwd()`
+    );
+  });
+
+  const serviceUtilsPath = path.join(
+    process.cwd(),
+    'services',
+    'serviceUtils.js'
+  );
+  const serviceUtilsContent = fs.readFileSync(serviceUtilsPath, 'utf8');
+
+  assertNotIncludes(
+    serviceUtilsContent,
+    "'/app/data/logs/prompt.txt'",
+    "serviceUtils.js's writePromptToFile() must not default to a hardcoded '/app' path"
+  );
+  assertIncludes(
+    serviceUtilsContent,
+    "path.join(process.cwd(), 'data', 'logs', 'prompt.txt')",
+    "serviceUtils.js's writePromptToFile() must default to a process.cwd()-relative prompt log path"
+  );
+
+  console.log('✅ All native install log path checks passed');
+}
+
+run();
+console.log('✅ test-native-install-log-paths passed');