feat(corpus): add save-to-corpus + private corpus search scope

- POST /api/save-to-corpus.php — saves tool output text to user's default CaveauAI corpus via ClientRagPipeline
- api/case/upload.php — dual-writes uploaded PDFs to CaveauAI client_documents (best-effort)
- assets/js/corpus-save.js — shared <dialog> handler for .js-save-corpus buttons on all tool pages
- includes/layout_footer.php — injects corpus-save.js + shared save dialog markup
- korrespond/deep-research/barnevernet/discrepancy JS — save-to-corpus buttons on output sections
- api/search.php + LegalTools::search() — corpus_scope param ('shared'|'private'|'both'), merges personal CaveauAI corpus with shared legal library when 'both'
- includes/tool_form.php + assets/js/tools.js — corpus scope radio toggle shown on search tab
- api/user-docs.php — add POST upload method for non-SSO authenticated users

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

api/case/upload.php

@@ -40,11 +40,51 @@ if (strncmp($head, '%PDF-', 5) !== 0) {
 try {
     $doc = CaseStore::registerUpload($userId, $name, $tmp, $size);
     CaseStore::caseEnqueueIngest((int)$doc['doc_id'], $userId);
-    dbnToolsRespond([
-        'ok' => true,
-        'doc_id' => $doc['doc_id'],
-        'filename' => $doc['filename'],
-    ]);
 } catch (Throwable $e) {
     dbnToolsError($e->getMessage(), 400, 'upload_failed');
 }
+
+// Dual-write to CaveauAI corpus (best-effort — never fails the upload)
+$caveauDocId = null;
+$clientId = (int)($_SESSION['dbn_tools_client_id'] ?? 0);
+if ($clientId > 0 && !empty($doc['storage_path'])) {
+    try {
+        dbnToolsBootCaveau();
+        $aiPortalRoot = dbnToolsAiPortalRoot();
+        $textExtractFile = $aiPortalRoot . '/platform/includes/text_extract.php';
+        if (is_file($textExtractFile)) {
+            require_once $textExtractFile;
+            $content = extractPdfText($doc['storage_path']);
+            if ($content !== '' && strlen($content) > 30) {
+                $caveauDb = getDb();
+                $corpusSt = $caveauDb->prepare(
+                    'SELECT id FROM client_corpora WHERE client_id = ? AND is_default = 1 LIMIT 1'
+                );
+                $corpusSt->execute([$clientId]);
+                $corpusId = (int)($corpusSt->fetchColumn() ?: 0);
+                if ($corpusId > 0) {
+                    $title = pathinfo($doc['filename'], PATHINFO_FILENAME);
+                    $caveauDb->prepare("
+                        INSERT INTO client_documents
+                          (client_id, corpus_id, title, source_type, content, category,
+                           import_method, word_count, status)
+                        VALUES (?, ?, ?, 'pdf', ?, 'user-upload', 'dbn_upload', ?, 'pending')
+                    ")->execute([$clientId, $corpusId, $title, $content, str_word_count($content)]);
+                    $caveauDocId = (int)$caveauDb->lastInsertId();
+                    $rag = new ClientRagPipeline($clientId);
+                    $rag->ingestDocument($caveauDocId);
+                }
+            }
+        }
+    } catch (Throwable $e) {
+        // Non-fatal: log and continue
+        error_log('[upload] CaveauAI dual-write failed for doc ' . ($doc['doc_id'] ?? '?') . ': ' . $e->getMessage());
+    }
+}
+
+dbnToolsRespond([
+    'ok' => true,
+    'doc_id' => $doc['doc_id'],
+    'filename' => $doc['filename'],
+    'caveau_doc_id' => $caveauDocId,
+]);

api/save-to-corpus.php

@@ -0,0 +1,95 @@
+<?php
+/**
+ * POST /api/save-to-corpus.php
+ *
+ * Save tool output text into the user's CaveauAI corpus.
+ * Uses dbnToolsBootCaveau() to call ClientRagPipeline directly via filesystem include.
+ *
+ * Request body (JSON, max 500 KB):
+ *   title       string (required)
+ *   content     string (required, min 30 chars)
+ *   source_tool string (optional, slug)
+ *   tags        string (optional, comma-separated)
+ */
+
+declare(strict_types=1);
+
+require_once dirname(__DIR__) . '/includes/bootstrap.php';
+
+dbnToolsRequireMethod('POST');
+dbnToolsRequireAuth();
+
+$user     = dbnToolsAuthenticatedUser();
+$clientId = (int)($user['client_id'] ?? 0);
+if ($clientId <= 0) {
+    dbnToolsError('No linked CaveauAI workspace. Log in via the CaveauAI portal first.', 403, 'no_workspace');
+}
+
+$input      = dbnToolsJsonInput(500_000);
+$title      = trim($input['title'] ?? '');
+$content    = trim($input['content'] ?? '');
+$sourceTool = substr(preg_replace('/[^a-z0-9\-_]/', '', strtolower($input['source_tool'] ?? '')), 0, 64) ?: null;
+$rawTags    = trim($input['tags'] ?? '');
+$tags       = json_encode(
+    array_values(array_filter(array_map('trim', explode(',', $rawTags)))),
+    JSON_UNESCAPED_UNICODE
+);
+
+if ($title === '') {
+    dbnToolsError('title is required.', 400, 'bad_request');
+}
+if (strlen($content) < 30) {
+    dbnToolsError('content too short (min 30 chars).', 400, 'bad_request');
+}
+if (strlen($content) > 2_000_000) {
+    dbnToolsError('content exceeds 2 MB limit.', 400, 'too_large');
+}
+
+// Load CaveauAI platform (getDb, ClientRagPipeline, etc.)
+dbnToolsBootCaveau();
+
+try {
+    $db = getDb();
+} catch (Throwable $e) {
+    dbnToolsError('CaveauAI database unavailable: ' . $e->getMessage(), 503, 'db_unavailable');
+}
+
+// Resolve default corpus for this client
+$stmt = $db->prepare('SELECT id FROM client_corpora WHERE client_id = ? AND is_default = 1 LIMIT 1');
+$stmt->execute([$clientId]);
+$corpusId = (int)($stmt->fetchColumn() ?: 0);
+if ($corpusId === 0) {
+    dbnToolsError(
+        'No default corpus found for your account. Set one up in the CaveauAI portal.',
+        409,
+        'no_corpus'
+    );
+}
+
+$wordCount = str_word_count($content);
+
+$ins = $db->prepare("
+    INSERT INTO client_documents
+      (client_id, corpus_id, title, source_type, content, category,
+       tags, import_method, source_tool, word_count, status)
+    VALUES (?, ?, ?, 'text', ?, 'tool-output', ?, 'tool_output', ?, ?, 'pending')
+");
+$ins->execute([$clientId, $corpusId, $title, $content, $tags, $sourceTool, $wordCount]);
+$docId = (int)$db->lastInsertId();
+
+try {
+    $rag    = new ClientRagPipeline($clientId);
+    $chunks = $rag->ingestDocument($docId);
+} catch (Throwable $e) {
+    // Document is saved but not indexed — mark error and return partial success
+    $db->prepare("UPDATE client_documents SET status='error', error_message=? WHERE id=?")
+       ->execute([$e->getMessage(), $docId]);
+    dbnToolsError(
+        'Saved to corpus but indexing failed: ' . $e->getMessage(),
+        500,
+        'index_failed',
+        ['document_id' => $docId]
+    );
+}
+
+dbnToolsRespond(['ok' => true, 'document_id' => $docId, 'chunks' => $chunks], 201);

api/search.php

@@ -17,5 +17,8 @@ dbnToolsWithTelemetry('search', $language, function () use ($input, $language):
     $asOfDate = isset($input['as_of_date']) && preg_match('/^\d{4}(-\d{2}(-\d{2})?)?$/', $input['as_of_date'])
         ? $input['as_of_date']
         : null;
-    return (new DbnLegalToolsService())->search($query, $language, $limit, $temporalMode, $asOfDate);
+    $scope = in_array($input['corpus_scope'] ?? '', ['shared', 'private', 'both'], true)
+        ? $input['corpus_scope']
+        : 'both';
+    return (new DbnLegalToolsService())->search($query, $language, $limit, $temporalMode, $asOfDate, $scope);
 });

api/user-docs.php

@@ -2,16 +2,18 @@
 declare(strict_types=1);
 
 /**
- * GET    /api/user-docs.php        — list SSO user's uploaded documents
+ * GET    /api/user-docs.php        — list uploaded documents for current user
  * DELETE /api/user-docs.php?id=X  — remove a document
+ * POST   /api/user-docs.php        — upload a document (file field = 'file')
  *
- * Only available for SSO users (dbn_tools_sso_uid set in session).
+ * SSO users (dbn_tools_sso_uid) are keyed by their SSO uid.
- * Reads from the shared dobetternorge.dbn_user_docs table, keyed by sso_uid.
+ * Other authenticated users are keyed by session_id() as a fallback.
+ * Reads/writes the shared dobetternorge.dbn_user_docs table.
  * Requires DBN_DB_* env vars pointing at the dobetternorge database.
  */
 require_once __DIR__ . '/../includes/bootstrap.php';
 
-dbnToolsRequireMethod('GET', 'DELETE');
+dbnToolsRequireMethod('GET', 'DELETE', 'POST');
 
 if (!dbnToolsIsAuthenticated()) {
     http_response_code(401);
@@ -20,13 +22,9 @@ if (!dbnToolsIsAuthenticated()) {
     exit;
 }
 
-// Only SSO users have shared docs
+// SSO uid for SSO users; session id as stable key for client sessions
 $ssoUid = (string)($_SESSION['dbn_tools_sso_uid'] ?? '');
-if ($ssoUid === '') {
+$userKey = $ssoUid !== '' ? $ssoUid : 'sess_' . session_id();
-    header('Content-Type: application/json');
-    echo json_encode(['ok' => true, 'docs' => [], 'reason' => 'sso_only']);
-    exit;
-}
 
 header('Content-Type: application/json; charset=utf-8');
 
@@ -51,6 +49,50 @@ function dbnSharedDb(): ?PDO
 
 $method = $_SERVER['REQUEST_METHOD'];
 
+// ── POST — upload a document ──────────────────────────────────────────────────
+if ($method === 'POST') {
+    if (empty($_FILES['file']) || !is_array($_FILES['file'])) {
+        http_response_code(422);
+        echo json_encode(['ok' => false, 'error' => 'No file uploaded.']);
+        exit;
+    }
+
+    try {
+        $extracted = dbnToolsExtractUploadedFile($_FILES['file']);
+    } catch (Throwable $e) {
+        http_response_code(422);
+        echo json_encode(['ok' => false, 'error' => $e->getMessage()]);
+        exit;
+    }
+
+    $docId    = uniqid('wbd_', true);
+    $filename = basename((string)($_FILES['file']['name'] ?? 'document'));
+    $fileType = strtolower(pathinfo($filename, PATHINFO_EXTENSION));
+    $chunks   = isset($extracted['text']) ? max(1, (int)ceil(mb_strlen($extracted['text']) / 1000)) : 0;
+    $now      = gmdate('Y-m-d H:i:s');
+
+    $db = dbnSharedDb();
+    if ($db) {
+        $db->prepare(
+            'INSERT INTO dbn_user_docs (id, user_id, filename, file_type, chunk_count, source, status, created_at)
+             VALUES (?, ?, ?, ?, ?, ?, ?, ?)'
+        )->execute([$docId, $userKey, $filename, $fileType, $chunks, 'workbench', 'ready', $now]);
+    }
+
+    echo json_encode([
+        'ok'  => true,
+        'doc' => [
+            'doc_id'      => $docId,
+            'filename'    => $filename,
+            'file_type'   => $fileType,
+            'chunk_count' => $chunks,
+            'source'      => 'workbench',
+            'created_at'  => $now,
+        ],
+    ]);
+    exit;
+}
+
 // ── DELETE ────────────────────────────────────────────────────────────────────
 if ($method === 'DELETE') {
     $docId = trim($_GET['id'] ?? '');
@@ -63,10 +105,10 @@ if ($method === 'DELETE') {
     $db = dbnSharedDb();
     if ($db) {
         $stmt = $db->prepare('SELECT id FROM dbn_user_docs WHERE id = ? AND user_id = ?');
-        $stmt->execute([$docId, $ssoUid]);
+        $stmt->execute([$docId, $userKey]);
         if ($stmt->fetch()) {
             $db->prepare('DELETE FROM dbn_user_docs WHERE id = ? AND user_id = ?')
-               ->execute([$docId, $ssoUid]);
+               ->execute([$docId, $userKey]);
 
             // Delete Qdrant points for this doc
             $qdrantUrl = 'http://10.0.2.10:6333';
@@ -74,7 +116,7 @@ if ($method === 'DELETE') {
                 'filter' => [
                     'must' => [
                         ['key' => 'doc_id',  'match' => ['value' => $docId]],
-                        ['key' => 'user_id', 'match' => ['value' => $ssoUid]],
+                        ['key' => 'user_id', 'match' => ['value' => $userKey]],
                     ],
                 ],
             ];
@@ -108,7 +150,7 @@ $stmt = $db->prepare(
      ORDER BY created_at DESC
      LIMIT 50'
 );
-$stmt->execute([$ssoUid, 'ready']);
+$stmt->execute([$userKey, 'ready']);
 $rows = $stmt->fetchAll();
 
 $docs = array_map(static fn($r) => [

assets/js/barnevernet.js

@@ -813,6 +813,21 @@
       els.results.appendChild(finalContainer.firstChild);
     }
 
+    // Save-to-corpus button
+    const briefEl = els.results.querySelector('.dr-brief');
+    if (briefEl) {
+      briefEl.id = 'bvjBriefText';
+      const saveBtn = document.createElement('button');
+      saveBtn.type = 'button';
+      saveBtn.className = 'js-save-corpus secondary-button';
+      saveBtn.dataset.tool = 'barnevernet';
+      saveBtn.dataset.contentId = 'bvjBriefText';
+      saveBtn.dataset.suggestedTitle = 'BVJ analyse: ' + (document.getElementById('bvjQuestion')?.value?.slice(0, 80) ?? 'Svar');
+      saveBtn.textContent = 'Save to corpus';
+      saveBtn.style.marginTop = '12px';
+      briefEl.insertAdjacentElement('afterend', saveBtn);
+    }
+
     // Bind source card clicks
     els.results.querySelectorAll('.dr-source-card[data-source-n]').forEach((node) => {
       node.addEventListener('click', (e) => {

assets/js/corpus-save.js

@@ -0,0 +1,106 @@
+/**
+ * corpus-save.js — "Save to corpus" shared handler for all DBN tool pages.
+ *
+ * Buttons that trigger a save must have:
+ *   class="js-save-corpus"
+ *   data-content-id="<id of element containing the text to save>"
+ *   data-tool="<source_tool slug, e.g. korrespond>"
+ *   data-suggested-title="<pre-filled title string>"  (optional)
+ */
+
+(function () {
+  'use strict';
+
+  const dlg      = document.getElementById('save-corpus-dialog');
+  const form     = document.getElementById('save-corpus-form');
+  const titleIn  = document.getElementById('save-corpus-title');
+  const tagsIn   = document.getElementById('save-corpus-tags');
+  const cancelBtn = document.getElementById('save-corpus-cancel');
+
+  if (!dlg || !form) return; // dialog not present (e.g. not logged in)
+
+  cancelBtn?.addEventListener('click', () => dlg.close());
+
+  let _pendingBtn  = null;
+  let _pendingContent = '';
+  let _pendingTool = '';
+
+  // Delegated click — catches buttons added dynamically by tool JS
+  document.addEventListener('click', (e) => {
+    const btn = e.target.closest('.js-save-corpus');
+    if (!btn) return;
+
+    const contentId = btn.dataset.contentId;
+    const el = contentId ? document.getElementById(contentId) : null;
+    const content = (el ? (el.value ?? el.textContent) : '').trim();
+
+    if (!content || content.length < 30) {
+      btn.textContent = 'Nothing to save';
+      setTimeout(() => { btn.textContent = 'Save to corpus'; }, 2000);
+      return;
+    }
+
+    _pendingBtn     = btn;
+    _pendingContent = content;
+    _pendingTool    = btn.dataset.tool ?? '';
+
+    titleIn.value = btn.dataset.suggestedTitle ?? '';
+    tagsIn.value  = '';
+    dlg.showModal();
+    titleIn.focus();
+    titleIn.select();
+  });
+
+  // Form submit inside dialog
+  form.addEventListener('submit', async (e) => {
+    e.preventDefault();
+    dlg.close();
+
+    const btn     = _pendingBtn;
+    const content = _pendingContent;
+    const title   = titleIn.value.trim();
+    const tags    = tagsIn.value.trim();
+    const tool    = _pendingTool;
+
+    if (!title || !content) return;
+
+    if (btn) {
+      btn.disabled    = true;
+      btn.textContent = 'Saving…';
+    }
+
+    try {
+      const resp = await fetch('api/save-to-corpus.php', {
+        method:  'POST',
+        headers: { 'Content-Type': 'application/json' },
+        body:    JSON.stringify({ title, content, source_tool: tool, tags }),
+      });
+
+      const data = await resp.json().catch(() => ({}));
+
+      if (resp.ok && data.ok) {
+        if (btn) {
+          btn.textContent = '✓ Saved to corpus';
+          btn.classList.add('js-save-corpus--saved');
+        }
+      } else {
+        const msg = data.error ?? `Error ${resp.status}`;
+        if (btn) {
+          btn.textContent = 'Save failed';
+          btn.disabled    = false;
+          btn.title       = msg;
+        }
+        console.error('[corpus-save] Save failed:', msg);
+      }
+    } catch (err) {
+      if (btn) {
+        btn.textContent = 'Network error';
+        btn.disabled    = false;
+      }
+      console.error('[corpus-save] Network error:', err);
+    }
+
+    _pendingBtn     = null;
+    _pendingContent = '';
+  });
+}());

assets/js/deep-research.js

@@ -563,6 +563,21 @@
       ${nextHtml}
     `;
 
+    // Save-to-corpus button (inject after brief block)
+    const briefEl = els.results.querySelector('.dr-brief');
+    if (briefEl) {
+      briefEl.id = 'drBriefText';
+      const saveBtn = document.createElement('button');
+      saveBtn.type = 'button';
+      saveBtn.className = 'js-save-corpus secondary-button';
+      saveBtn.dataset.tool = 'deep-research';
+      saveBtn.dataset.contentId = 'drBriefText';
+      saveBtn.dataset.suggestedTitle = 'Research: ' + (document.getElementById('drQuery')?.value?.slice(0, 80) ?? 'Report');
+      saveBtn.textContent = 'Save to corpus';
+      saveBtn.style.marginTop = '12px';
+      briefEl.insertAdjacentElement('afterend', saveBtn);
+    }
+
     // Bind source-card click handlers (open modal) — but ignore clicks on inner <a>
     els.results.querySelectorAll('.dr-source-card[data-source-n]').forEach((node) => {
       node.addEventListener('click', (e) => {

assets/js/discrepancy.js

@@ -551,6 +551,17 @@
       els.results.appendChild(finalContainer.firstChild);
     }
 
+    // Save-to-corpus button (appended after final results)
+    const saveBtn = document.createElement('button');
+    saveBtn.type = 'button';
+    saveBtn.className = 'js-save-corpus secondary-button';
+    saveBtn.dataset.tool = 'discrepancy';
+    saveBtn.dataset.contentId = 'dcResults';
+    saveBtn.dataset.suggestedTitle = 'Discrepancy report';
+    saveBtn.textContent = 'Save to corpus';
+    saveBtn.style.marginTop = '16px';
+    els.results.appendChild(saveBtn);
+
     // Bind tabs
     els.results.querySelectorAll('.dc-tab').forEach((btn) => {
       btn.addEventListener('click', () => {

assets/js/korrespond.js

@@ -543,6 +543,12 @@
             </div>
           </div>
           <pre class="korr-draft-body" id="korrDraftNo">${esc(draftNo)}</pre>
+          <button type="button" class="js-save-corpus secondary-button"
+            data-tool="korrespond"
+            data-content-id="korrDraftNo"
+            data-suggested-title="${esc((data.output_type || 'Brev') + ' — ' + (data.recipient_body || ''))}">
+            Save to corpus
+          </button>
         </div>
         ${isSameLang ? '' : `
         <div class="korr-draft-col">
@@ -554,6 +560,12 @@
             </div>
           </div>
           <pre class="korr-draft-body" id="korrDraftUser">${esc(draftUser)}</pre>
+          <button type="button" class="js-save-corpus secondary-button"
+            data-tool="korrespond"
+            data-content-id="korrDraftUser"
+            data-suggested-title="${esc((data.output_type || 'Brev') + ' — ' + (data.recipient_body || '') + ' (translation)')}">
+            Save to corpus
+          </button>
         </div>`}
       </div>
 

assets/js/tools.js

@@ -910,6 +910,7 @@ document.addEventListener('DOMContentLoaded', () => {
     uploadFileList: document.querySelector('#uploadFileList'),
     uploadClear: document.querySelector('#uploadClear'),
     aliasSection: document.querySelector('#aliasSection'),
+    corpusScopeControl: document.querySelector('#corpusScopeControl'),
     addAliasRow: document.querySelector('#addAliasRow'),
     aliasRows: document.querySelector('#aliasRows'),
     audioZone: document.querySelector('#audioZone'),
@@ -1013,6 +1014,7 @@ function setTool(toolName) {
     els.input.placeholder = tool.placeholder;
   }
   els.languageControl.classList.toggle('is-hidden', !tool.usesLanguage);
+  els.corpusScopeControl?.classList.toggle('is-hidden', toolName !== 'search');
   els.redactionControl.classList.toggle('is-hidden', toolName !== 'redact');
   els.uploadZone.classList.toggle('is-hidden', toolName !== 'redact' && toolName !== 'timeline');
   els.aliasSection.classList.toggle('is-hidden', toolName !== 'redact');
@@ -1080,6 +1082,7 @@ async function runTool(event) {
   }
   if (state.activeTool === 'search') {
     payload.limit = 7;
+    payload.corpus_scope = currentCorpusScope();
   }
   if (state.activeTool === 'redact') {
     lastOriginalText = text;
@@ -1329,6 +1332,10 @@ function currentLanguage() {
   return document.querySelector('input[name="language"]:checked')?.value || 'en';
 }
 
+function currentCorpusScope() {
+  return document.querySelector('input[name="corpusScope"]:checked')?.value || 'both';
+}
+
 function currentRedactionMode() {
   return document.querySelector('input[name="redactionMode"]:checked')?.value || 'standard';
 }

includes/LegalTools.php

@@ -20,7 +20,8 @@ final class DbnLegalToolsService
         string  $language    = 'en',
         int     $limit       = 6,
         string  $temporalMode = 'disabled',
-        ?string $asOfDate    = null
+        ?string $asOfDate    = null,
+        string  $scope       = 'both'
     ): array {
         $query = trim($query);
         if (mb_strlen($query, 'UTF-8') < 3) {
@@ -28,15 +29,24 @@ final class DbnLegalToolsService
         }
         $limit = max(1, min(10, $limit));
         $temporalMode = in_array($temporalMode, ['legal_conservative', 'disabled'], true) ? $temporalMode : 'disabled';
+        $scope = in_array($scope, ['shared', 'private', 'both'], true) ? $scope : 'both';
 
+        $scopeLabel = match ($scope) {
+            'private' => 'personal corpus only',
+            'shared'  => 'Legal Library only',
+            default   => 'Legal Library + personal corpus',
+        };
         $trace = [
-            $this->trace('Query interpretation', 'Searching Do Better Norge private corpus plus the subscribed family-legal package.', 'complete'),
+            $this->trace('Query interpretation', "Searching Do Better Norge {$scopeLabel}.", 'complete'),
-            $this->trace('Search tools used', 'ClientRagPipeline::searchAll with keyword mode, private corpus enabled, shared package filter set to family-legal.', 'running'),
+            $this->trace('Search tools used', 'ClientRagPipeline::searchAll with keyword mode.', 'running'),
         ];
 
         $client = dbnToolsRequireClient();
         $package = $this->requireFamilyPackage((int)$client['id']);
 
+        // Personal corpus client_id from session (may be 0 if user has no linked workspace)
+        $personalClientId = (int)($_SESSION['dbn_tools_client_id'] ?? 0);
+
         $chunks = [];
         $retrievalNote = 'ClientRagPipeline keyword retrieval';
         try {
@@ -52,6 +62,20 @@ final class DbnLegalToolsService
                 // Retrieval still works in keyword mode without gateway config.
             }
 
+            if ($scope === 'private') {
+                // Search only the user's personal corpus
+                if ($personalClientId > 0) {
+                    $rag = new ClientRagPipeline($personalClientId, $gatewayUrl, 30);
+                    $chunks = $rag->searchAll($query, $limit, null, [
+                        'search_private' => true,
+                        'search_shared' => false,
+                        'chunk_limit' => $limit,
+                        'search_method' => 'keyword',
+                        'min_private' => 0,
+                    ]);
+                }
+            } elseif ($scope === 'shared') {
+                // Search only the shared legal library
                 $rag = new ClientRagPipeline((int)$client['id'], $gatewayUrl, 30);
                 $chunks = $rag->searchAll($query, $limit, null, [
                     'search_private' => true,
@@ -62,6 +86,40 @@ final class DbnLegalToolsService
                     'min_private' => 0,
                     'include_beta_website' => true,
                 ]);
+            } else {
+                // 'both': shared library + personal corpus merged and re-ranked by score
+                $rag = new ClientRagPipeline((int)$client['id'], $gatewayUrl, 30);
+                $sharedChunks = $rag->searchAll($query, $limit, null, [
+                    'search_private' => true,
+                    'search_shared' => true,
+                    'package_ids' => [(int)$package['id']],
+                    'chunk_limit' => $limit,
+                    'search_method' => 'keyword',
+                    'min_private' => 0,
+                    'include_beta_website' => true,
+                ]);
+
+                $privateChunks = [];
+                if ($personalClientId > 0) {
+                    try {
+                        $ragPrivate = new ClientRagPipeline($personalClientId, $gatewayUrl, 30);
+                        $privateChunks = $ragPrivate->searchAll($query, $limit, null, [
+                            'search_private' => true,
+                            'search_shared' => false,
+                            'chunk_limit' => $limit,
+                            'search_method' => 'keyword',
+                            'min_private' => 0,
+                        ]);
+                    } catch (Throwable $e) {
+                        error_log('[search] personal corpus query failed for client ' . $personalClientId . ': ' . $e->getMessage());
+                    }
+                }
+
+                // Merge by score descending, cap at $limit
+                $merged = array_merge($sharedChunks, $privateChunks);
+                usort($merged, fn($a, $b) => ($b['score'] ?? 0) <=> ($a['score'] ?? 0));
+                $chunks = array_slice($merged, 0, $limit);
+            }
 
             // Apply temporal reranking after retrieval (optional)
             if ($temporalMode === 'legal_conservative' && !empty($chunks)) {

includes/layout_footer.php

@@ -26,5 +26,26 @@
 <?php if (!empty($extraScripts) && is_array($extraScripts)): foreach ($extraScripts as $extraScript): ?>
 <script src="<?= htmlspecialchars((string)$extraScript) ?>" defer></script>
 <?php endforeach; endif; ?>
+<script src="assets/js/corpus-save.js" defer></script>
+
+<!-- Save-to-corpus dialog (shared across all tool pages) -->
+<dialog id="save-corpus-dialog" class="save-corpus-dialog">
+  <form method="dialog" id="save-corpus-form">
+    <h3>Save to corpus</h3>
+    <p class="save-corpus-hint">This will be indexed and searchable in your private corpus.</p>
+    <label>
+      <span>Title <span aria-hidden="true">*</span></span>
+      <input id="save-corpus-title" type="text" required placeholder="Give this entry a title…" autocomplete="off">
+    </label>
+    <label>
+      <span>Tags <span class="save-corpus-optional">(comma-separated)</span></span>
+      <input id="save-corpus-tags" type="text" placeholder="e.g. barnevern, 2024, kjennelse">
+    </label>
+    <menu>
+      <button type="submit" class="btn-primary">Save to corpus</button>
+      <button type="button" id="save-corpus-cancel">Cancel</button>
+    </menu>
+  </form>
+</dialog>
 </body>
 </html>

includes/tool_form.php

@@ -19,6 +19,13 @@
                         <input type="number" id="numSpeakersInput" name="num_speakers" min="2" max="10" placeholder="auto" class="num-speakers-input" aria-label="Expected speaker count">
                     </div>
 
+                    <div class="control-row is-hidden" id="corpusScopeControl">
+                        <span class="control-label">Search</span>
+                        <label><input type="radio" name="corpusScope" value="both" checked> Legal Library + My Docs</label>
+                        <label><input type="radio" name="corpusScope" value="shared"> Legal Library only</label>
+                        <label><input type="radio" name="corpusScope" value="private"> My Docs only</label>
+                    </div>
+
                     <div class="control-row is-hidden" id="redactionControl">
                         <span class="control-label">Mode</span>
                         <label><input type="radio" name="redactionMode" value="standard" checked> Standard</label>