refactor: remove the embedding settings, whose only consumer is gone

Embeddings existed here for Learning Hub semantic search — the card said so itself. Learning Hub was removed, and nothing took its place: the clinical corpus is embedded by the indexing service, not by this app. What was left was a settings page that configured a model, tested it, reported its dimensions, and fed nothing. src/utils/embeddings.js had exactly one importer, src/routes/adminConfig .js, which used it for the three routes this deletes. Outside those, the only mentions of embedding in the server were a comment and a settings prefix. Gone: the module, its three admin routes, the dimension probe, the Discover & test kind and its two panels, the admin.js block behind them, the embeddings. prefix from both the writable-settings allowlist and the lockdown list (it can no longer be written at all, so locking it says nothing), and docs/embeddings-setup.md, which documented Learning Hub search end to end. 'embedding' stays in NON_CHAT_MODES — that is the filter keeping embedding models out of the chat-model list, and the gateway still serves them. Docs still describe nine /api/learning endpoints that no longer exist, left from the Learning Hub removal. Not touched here; that is its own subject. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Dv6sqaY6Vq3ChZHMem3cnU
2026-09-12 21:47:09 +02:00 · 2026-09-12 21:47:09 +02:00 · a505244b97
commit a505244b97
parent 9df2230479
17 changed files with 17 additions and 692 deletions
--- a/README.md
+++ b/README.md
@ -169,7 +169,6 @@ Primary references:
 - `docs/ai-providers.md` — provider selection, prompts, injection hardening.
 - `docs/clinical-assistant.md` — MCP-backed assistant behavior and safety rules.
 - `docs/retrieval-tuning.md` — how much corpus each feature retrieves, and what it costs.
- `docs/embeddings-setup.md` — embedding model configuration.
 - `docs/global-prompt-administration.md` — prompt overrides and the conversation budget.
 - `docs/speech.md` — STT, TTS, recording, and audio backups.
 - `docs/my-resources.md` — private teaching material, the slide renderer, and search sources.
--- a/docs/api-reference.md
+++ b/docs/api-reference.md
@ -2070,7 +2070,7 @@ Get all application configuration settings.

 ### PUT /api/admin/config/:key

-Update one application configuration setting. The key must match an allowed prefix: `announcement.`, `feature.`, `email.`, `prompt.`, `registration_enabled`, `registration_invite_only`, `site.`, `smtp.`, `models.`, `tts.`, `stt.`, `embeddings.`, `clinical_assistant.`, or `my_resources.`. Anything else is rejected with 400.
+Update one application configuration setting. The key must match an allowed prefix: `announcement.`, `feature.`, `email.`, `prompt.`, `registration_enabled`, `registration_invite_only`, `site.`, `smtp.`, `models.`, `tts.`, `stt.`, `clinical_assistant.`, or `my_resources.`. Anything else is rejected with 400.

 Some keys are refused here even when allowed: `models.*` must go through the validated model endpoints, `feature.*` values must be `true` or `false`, and any key under lockdown returns 403.

@ -2604,7 +2604,7 @@ available, so a code that never arrives is never a lockout. The rest manage
 | `POST` | `/api/auth/login-code/request` |
 | `POST` | `/api/auth/login-code/verify` |

-Remaining endpoints not listed above are additional admin configuration, model/STT/TTS/embedding discovery and test calls, Learning Hub CMS operations, and the per-feature AI helpers (`/api/dont-miss`, `/api/suggest-codes`, `/api/generate-pe-narrative`, `/api/hospital-course-update`, `/api/hospital-course-clarify`, `/api/well-visit/note`, `/api/milestones-data`, `/api/user/features`, `/api/logs/client-error`, `/api/logs/client-event`, `/api/generated-images/:id`, `/api/image-jobs/:workflow`).
+Remaining endpoints not listed above are additional admin configuration, model/STT/TTS discovery and test calls, and the per-feature AI helpers (`/api/dont-miss`, `/api/suggest-codes`, `/api/generate-pe-narrative`, `/api/hospital-course-update`, `/api/hospital-course-clarify`, `/api/well-visit/note`, `/api/milestones-data`, `/api/user/features`, `/api/logs/client-error`, `/api/logs/client-event`, `/api/generated-images/:id`, `/api/image-jobs/:workflow`).

 ---

--- a/docs/architecture.md
+++ b/docs/architecture.md
@ -47,7 +47,6 @@ src/
    promptSafe.js                # <UNTRUSTED_*> LLM prompt wrapper
    logger.js                    # audit/api/access + Loki shipper
    errors.js                    # generic 500 responder
-    embeddings.js                # LiteLLM embeddings
    sttProvider.js, ttsProvider.js  # speech-to-text and text-to-speech routing
    documentExport.js            # pptx/docx/pdf export
    slideSpec.js, docSpec.js     # markdown -> typed spec for the renderers
@ -187,7 +186,7 @@ Ped-AI is a self-hosted Express application with a browser frontend, PostgreSQL
 | Web app | Ped-AI | Auth, UI, clinical workflows, admin settings, notes, Learning Hub, bedside tools |
 | Database | PostgreSQL | Users, sessions, settings, saved app data, audit/API/access logs |
 | Operational cache | Redis | Prompt suggestions, lightweight state, queue groundwork; not clinical answer caching |
-| Model gateway | LiteLLM | Text, speech, image, embedding model discovery and routing |
+| Model gateway | LiteLLM | Text, speech and image model discovery and routing |
 | Clinical retrieval | MCP service | Nextcloud access, indexing, search, rerank, source metadata |
 | Reverse proxy | Caddy or equivalent | TLS and public routing |

--- a/docs/configuration.md
+++ b/docs/configuration.md
@ -150,7 +150,6 @@ with 2-minute in-memory cache. Writes invalidate the cache immediately.
 | `ai.allow_model_fallback` | Enable silent fallback to secondary model on primary failure. **Default false** — fallback could spill to a non-BAA provider. |
 | `stt.model`, `tts.model`, `tts.voice` | System-wide STT/TTS defaults (users can override per-account). |
 | `prompt.{name}` | Prompt overrides. Any template in `src/utils/prompts.js` can be replaced live. |
-| `embeddings.model`, `embeddings.dimensions` | Override embedding config. |

 ### Feature flags

@ -197,18 +196,11 @@ OpenAI-compatible gateway — LiteLLM, Bifrost, or other proxies.
   (e.g., `openrouter/gpt-4.1`), while LiteLLM can use deployment aliases
   (e.g., `openrouter-gpt-4.1`). Update model names in:
   - Admin Panel → Models (chat models)
-   - Admin Panel → Settings → `stt.model` (speech-to-text)
-   - Admin Panel → Settings → `tts.model` (text-to-speech)
+   - Admin Panel → Models → Discover & test → Speech / Transcription
+     (**Set** makes a model the default)
   - `LITELLM_TTS_MODEL` env var (if set)

-4. **Embedding model** — Set via Admin Panel → Settings →
-   `embeddings.model`. The embedding vector column is `VECTOR(768)`, so
-   any model producing 768 dimensions works without re-embedding
-   (e.g., `vertex/text-embedding-005`). Switching to a model with
-   different dimensions requires altering the column and re-embedding all
-   content.
-
-5. **Restart the container** — `docker compose up -d --force-recreate` to
+4. **Restart the container** — `docker compose up -d --force-recreate` to
   pick up `.env` changes (a plain `restart` does not re-read `.env`).

 ### Verified gateways
--- a/docs/deployment.md
+++ b/docs/deployment.md
@ -155,7 +155,7 @@ App sets `trust proxy: 1` so rate limiting uses the original client IP.

 | Volume | Contents | Backup priority |
 |---|---|---|
-| `pgdata` | All user data, encounters, memories, audit logs, settings, embeddings | Critical |
+| `pgdata` | All user data, encounters, memories, audit logs, settings | Critical |
 | `scribe-logs` | Filesystem audit log files (JSONL by day) | High for compliance evidence; Postgres also has audit/API/access tables |

 ### Postgres backup / restore
--- a/docs/developer-guide.md
+++ b/docs/developer-guide.md
@ -37,7 +37,6 @@ src/
    fileType.js                  magic-byte upload verifier
    errors.js                    generic 500 responder
    logger.js                    audit + api + access + Loki shipper
-    embeddings.js                LiteLLM embeddings
    notify.js                    ntfy push
    transcribe.js, tts.js        LiteLLM STT / TTS routes
  routes/                        Express routers for auth, AI workflows, education, logs, and user data
--- a/docs/embeddings-setup.md
+++ b/docs/embeddings-setup.md
@ -1,241 +0,0 @@
-# Embeddings And Semantic Search Setup
-
-This guide explains how to set up and use the new vector-based semantic search for the Learning Hub.
-
-## What This Enables
-
- **Semantic search** - Find content by meaning, not just keywords
- **3 search modes**:
-  - **Keyword** (`/api/learning/search`) - Traditional text matching
-  - **Semantic** (`/api/learning/search/semantic`) - AI-powered vector similarity
-  - **Hybrid** (`/api/learning/search/hybrid`) - Combines both for best results
- **Auto-embedding** - Content is automatically vectorized when created/updated
- **Gateway-routed** - Uses LiteLLM embeddings so provider policy stays in one place
-
-## Prerequisites
-
-### 1. Install pgvector Extension
-
-The database needs the `pgvector` extension for vector operations:
-
-```bash
-# For PostgreSQL 16 on Ubuntu/Debian
-sudo apt-get install postgresql-16-pgvector
-
-# For PostgreSQL 15
-sudo apt-get install postgresql-15-pgvector
-
-# For Docker (add to Dockerfile or docker-compose)
-# The postgres:16-alpine base image doesn't include pgvector by default
-# You'll need to use a custom image or install at runtime
-```
-
-**For Docker deployments**, use this postgres image instead:
-```yaml
-postgres:
-  image: pgvector/pgvector:pg16
-  # ... rest of your config
-```
-
-### 2. Configure LiteLLM Embeddings
-
-Add to your `.env` file:
-
-```bash
-LITELLM_API_BASE=http://localhost:4000
-LITELLM_API_KEY=your-key
-EMBEDDING_MODEL=openai-text-embedding-3-large
-EMBEDDING_DIMENSIONS=3072
-```
-
-## Available Embedding Models
-
-The Admin embedding search reads LiteLLM `/model/info` and only shows models with `model_info.mode = "embedding"`. Do not add app-side built-in Vertex/OpenAI embedding lists; configure those choices in LiteLLM.
-
-The local LiteLLM instance currently exposes examples such as `openai-text-embedding-3-large`, `openai-text-embedding-3-small`, and Mistral embedding models. Dimensions are read from LiteLLM metadata when available.
-
-## Setup Steps
-
-### 1. Database Migration
-
-The database will automatically:
- Enable the `pgvector` extension
- Add `embedding vector(768)` column to `learning_content`
- Create IVFFLAT index for fast similarity search (after 10+ embeddings)
-
-Just restart your server after installing pgvector.
-
-### 2. Generate Embeddings for Existing Content
-
-Two options:
-
-**Option A: Admin API (recommended)**
-```bash
-curl -X POST http://localhost:3000/api/admin/learning/embeddings/generate \
-  -H "Authorization: Bearer YOUR_JWT_TOKEN" \
-  -H "Content-Type: application/json" \
-  -d '{"regenerateAll": false}'
-```
-
-**Option B: Via Admin Panel**
- Go to Admin → Learning Hub → Settings
- Click "Generate Embeddings" button
- Check status at `/api/admin/learning/embeddings/status`
-
-### 3. Verify Setup
-
-Check embedding status:
-```bash
-curl http://localhost:3000/api/admin/learning/embeddings/status \
-  -H "Authorization: Bearer YOUR_JWT_TOKEN"
-```
-
-Response:
-```json
-{
-  "success": true,
-  "enabled": true,
-  "total": 50,
-  "withEmbeddings": 50,
-  "missing": 0,
-  "model": "openai-text-embedding-3-large",
-  "dimensions": 3072
-}
-```
-
-## Using Semantic Search
-
-### Keyword Search (existing)
-```bash
-GET /api/learning/search?q=pneumonia
-```
-Returns exact text matches in title/subject/body.
-
-### Semantic Search (new)
-```bash
-GET /api/learning/search/semantic?q=childhood breathing problems&limit=10&threshold=0.5
-```
-Returns content similar by **meaning** (e.g., finds "pediatric asthma" articles).
-
-**Parameters:**
- `q` (required) - Search query
- `limit` (optional, default 10, max 50) - Max results
- `threshold` (optional, default 0.5) - Similarity threshold (0-1, higher = more similar)
- `contentType` (optional) - Filter by type: article, quiz, pearl, presentation
-
-### Hybrid Search (recommended)
-```bash
-GET /api/learning/search/hybrid?q=fever management
-```
-Combines keyword + semantic for best results. Automatically deduplicates and ranks by relevance.
-
-## How It Works
-
-1. **Content Creation/Update**:
-   - Text is extracted from `title`, `subject`, and `body` (HTML stripped)
-   - Sent to the configured LiteLLM embedding model
-   - Returns an embedding vector
-   - Stored in `learning_content.embedding` column
-
-2. **Semantic Search**:
-   - Query text → embedding vector
-   - PostgreSQL pgvector computes cosine similarity
-   - Returns top N most similar documents
-   - Similarity score 0-1 (1 = identical, 0 = unrelated)
-
-3. **Hybrid Search**:
-   - Runs both keyword + semantic searches in parallel
-   - Merges results (semantic first for quality)
-   - Deduplicates by content ID
-   - Sorts by relevance score
-
-## Cost Estimate
-
-Embedding cost depends on the upstream configured in LiteLLM.
-
-## Troubleshooting
-
-### "pgvector extension not available"
- Install: `apt-get install postgresql-16-pgvector`
- For Docker: Use `pgvector/pgvector:pg16` image
-
-### "Embeddings not configured"
- Verify `.env` has `LITELLM_API_BASE`
- Test: `curl http://localhost:3000/api/admin/learning/embeddings/status`
-
-### "Embedding generation failed"
- Check logs for API errors
- Verify LiteLLM `/model/info` shows the selected model with `mode: embedding`
- Check content isn't empty (skips empty bodies)
-
-### "No results from semantic search"
- Check if embeddings exist: `/api/admin/learning/embeddings/status`
- Lower threshold: `?threshold=0.3` (default 0.5)
- Verify pgvector index exists: `\di` in psql
-
-## Performance
-
- **Embedding generation**: latency depends on the LiteLLM upstream
- **Search latency**:
-  - Keyword: 10-50ms
-  - Semantic: 20-100ms (with IVFFLAT index)
-  - Hybrid: 30-150ms
- **Index build time**: ~1-5 seconds per 1,000 articles
-
-## Security And Compliance
-
- **Compliance**: controlled by the upstream provider configured in LiteLLM
- **Data retention**: Embeddings stored in your database only
- **No PHI**: Only article content (not patient data) is embedded
- **Encryption**: TLS in transit, at-rest encryption via PostgreSQL
-
-## Example Queries
-
-**Before (keyword):**
-```
-Query: "fever in babies"
-Results: Only articles with exact words "fever" or "babies"
-```
-
-**After (semantic):**
-```
-Query: "fever in babies"
-Results:
- Infant hyperthermia management (similarity: 0.89)
- Pediatric fever evaluation (similarity: 0.87)
- Febrile seizures in toddlers (similarity: 0.82)
- Neonatal temperature regulation (similarity: 0.78)
-```
-
-**Hybrid (best):**
-```
-Query: "asthma"
-Results:
- Childhood asthma management (keyword + semantic: 1.0)
- Pediatric breathing difficulties (semantic: 0.91)
- Reactive airway disease (semantic: 0.86)
- Bronchiolitis vs asthma (keyword: 1.0)
-```
-
-## API Reference
-
-### Admin Endpoints
-
- `POST /api/admin/learning/embeddings/generate` - Backfill embeddings
- `GET /api/admin/learning/embeddings/status` - Check status
- `GET /api/admin/learning/stats` - Includes embedding count
-
-### User Endpoints
-
- `GET /api/learning/search` - Keyword search
- `GET /api/learning/search/semantic` - Semantic search
- `GET /api/learning/search/hybrid` - Hybrid search (recommended)
-
-All endpoints require authentication (JWT token).
-
---
-
-**Questions?** Check logs for detailed error messages, or review the code in:
- `/src/utils/embeddings.js` - Core embedding logic
- `/src/routes/learningHub.js` - Search endpoints
- `/src/routes/learningAdmin.js` - Admin management
--- a/docs/logic/auth-admin-learning.md
+++ b/docs/logic/auth-admin-learning.md
@ -38,14 +38,6 @@ Learning Hub has two surfaces:
 - learner/user-facing routes under `/api/learning`
 - moderator/admin CMS routes under `/api/admin/learning`

-Content types include articles, pearls, quizzes, and presentations. AI content
-generation can use topic text, uploaded files, or connected Nextcloud/WebDAV
-sources. Semantic search uses pgvector embeddings on `learning_content` when an
-embedding provider is configured.
-
-See
-[`../embeddings-setup.md`](../embeddings-setup.md).
-
 ## Security Rules

 - Never expose raw secrets in admin health/config responses.
--- a/public/components/admin.html
+++ b/public/components/admin.html
@ -225,13 +225,12 @@
            <!-- One search for every kind of model. The kind decides which
                 gateway list is asked and what the row buttons do: chat and
                 image models are added to a roster; speech, transcription and
-                 embedding models are single defaults, so their button is Set. -->
+                 transcription models are single defaults, so their button is Set. -->
            <div class="admin-kind-switch" role="group" aria-label="Kind of model">
              <button type="button" class="admin-discover-kind" id="admin-discover-kind-chat" data-kind="chat" aria-pressed="true"><i class="fas fa-comments"></i> Chat</button>
              <button type="button" class="admin-discover-kind" id="admin-discover-kind-image" data-kind="image" aria-pressed="false"><i class="fas fa-image"></i> Image</button>
              <button type="button" class="admin-discover-kind" id="admin-discover-kind-tts" data-kind="tts" aria-pressed="false"><i class="fas fa-volume-up"></i> Speech (TTS)</button>
              <button type="button" class="admin-discover-kind" id="admin-discover-kind-stt" data-kind="stt" aria-pressed="false"><i class="fas fa-microphone"></i> Transcription (STT)</button>
-              <button type="button" class="admin-discover-kind" id="admin-discover-kind-embedding" data-kind="embedding" aria-pressed="false"><i class="fas fa-network-wired"></i> Embedding</button>
            </div>

            <!-- Provider status for the chosen kind. -->
@ -253,11 +252,6 @@
              <span id="admin-stt-info" class="admin-note">Loading...</span>
              <span class="admin-note"><strong>Set</strong> makes a model the default for dictation.</span>
            </div>
-            <div class="admin-kind-panel admin-kind-status" data-kind="embedding" hidden>
-              <span id="admin-embed-provider-badge" class="admin-badge">Loading...</span>
-              <span id="admin-embed-info" class="admin-note">Loading...</span>
-              <span class="admin-note">Learning Hub semantic search. Corpus (MCP) embeddings are configured on the indexing service, not here. <strong>Set</strong> makes a model the default.</span>
-            </div>

            <div class="admin-search-row">
              <input type="search" id="admin-discover-search" placeholder="Filter by name (e.g. gemini, gpt, llama)" aria-label="Filter models by name" style="font-size:13px;padding:6px 10px;border:1px solid var(--g300);border-radius:6px;flex:1;min-width:200px;">
@ -309,13 +303,6 @@
                  <div id="admin-stt-meta" style="font-size:11px;color:var(--g400);margin-top:4px;"></div>
                </div>
              </div>
-              <div class="admin-kind-panel" data-kind="embedding" hidden>
-                <div style="display:flex;gap:8px;flex-wrap:wrap;align-items:center;">
-                  <input type="text" id="admin-embed-test-text" value="Pediatric patient with fever and cough" aria-label="Sample text to embed" style="font-size:13px;padding:6px 10px;border:1px solid var(--g300);border-radius:6px;flex:1;min-width:200px;" placeholder="Sample text to embed...">
-                  <button id="btn-test-embedding" class="btn-sm btn-primary" type="button"><i class="fas fa-code-branch"></i> Generate</button>
-                </div>
-                <div id="admin-embed-result" role="status" class="admin-note" style="margin-top:8px;"></div>
-              </div>
            </div>
          </div>
        </details>
@ -348,7 +335,7 @@
              <p class="admin-note">Loading...</p>
            </div>

-            <p class="admin-note" style="border-top:1px solid var(--g100);padding-top:10px;">Speech, transcription and embedding models are single defaults rather than a roster: choose them with <strong>Set</strong> under Discover &amp; test.</p>
+            <p class="admin-note" style="border-top:1px solid var(--g100);padding-top:10px;">Speech and transcription models are single defaults rather than a roster: choose them with <strong>Set</strong> under Discover &amp; test.</p>
          </div>
        </details>

--- a/public/js/admin.js
+++ b/public/js/admin.js
@ -1210,7 +1210,7 @@ initImageSettings();
 // ============================================================
 // ADMIN DISCOVERY — one search box for every kind of model
 // ============================================================
-// Chat, image, speech, transcription and embedding models each have their own
+// Chat, image, speech and transcription models each have their own
 // gateway list and their own row buttons, and they used to have a card each,
 // scattered down the page. The kind switch keeps the five discovery calls as
 // they are and only decides which one the Search button asks. The switch
@ -1221,8 +1221,7 @@ initImageSettings();
    chat: 'Filter by name (e.g. gemini, gpt, llama)',
    image: 'Filter by name (e.g. dall-e, imagen, flux)',
    tts: 'Filter voices or models (e.g. Journey, Neural, alloy)',
-    stt: 'Filter by name (e.g. gemini, whisper)',
-    embedding: 'Filter by name (e.g. embedding, vertex)'
+    stt: 'Filter by name (e.g. gemini, whisper)'
  };

  function activeDiscoverKind() {
@ -2124,159 +2123,3 @@ initImageSettings();
      });
  }
 }
-
-// ============================================================
-// ADMIN EMBEDDING MODELS MANAGEMENT
-// ============================================================
-{
-  document.addEventListener('tabChanged', function(e) {
-    if (e.detail && e.detail.tab === 'admin') loadEmbeddingConfig();
-  });
-  // Catch-up for a tab that is already active and loaded at module init.
-  if (adminTabActive()) loadEmbeddingConfig();
-  document.addEventListener('click', function(e) {
-    if (e.target.closest('#btn-test-embedding'))     testEmbedding();
-    if (e.target.closest('.admin-embed-set-btn')) {
-      var btn = e.target.closest('.admin-embed-set-btn');
-      setEmbeddingDefault(btn.dataset.id, btn.dataset.dims, btn);
-    }
-  });
-  document.addEventListener('admin-discover', function(e) {
-    if (e.detail && e.detail.kind === 'embedding') discoverEmbeddings();
-  });
-
-  const esc = adminEscapeHtml;
-
-  function loadEmbeddingConfig() {
-    fetch('/api/admin/config/embeddings', { headers: getAuthHeaders() })
-      .then(function(r) { return r.json(); })
-      .then(function(data) {
-        if (!data.success) return;
-        var badge = document.getElementById('admin-embed-provider-badge');
-        if (badge) {
-          badge.textContent = (data.provider || 'none').toUpperCase();
-          badge.style.background = data.configured ? 'var(--g100)' : 'var(--red-light)';
-          badge.style.color = data.configured ? 'var(--g600)' : 'var(--red)';
-        }
-        var info = document.getElementById('admin-embed-info');
-        if (info) {
-          var parts = [];
-          if (data.dbModel) parts.push('DB model: ' + data.dbModel);
-          else if (data.envModel) parts.push('Env model: ' + data.envModel);
-          parts.push('Dims: ' + (data.currentDimensions || '?'));
-          if (!data.configured) parts.push('⚠️ Not configured');
-          info.textContent = parts.join(' · ');
-        }
-        var modelsEl = document.getElementById('admin-embed-models');
-        if (modelsEl && data.models) {
-          modelsEl.innerHTML = data.models.map(function(m) {
-            var isCurrent = m.id === data.currentModel;
-            return '<div style="display:flex;align-items:center;gap:8px;padding:5px 8px;border-radius:6px;background:var(--g50);font-size:13px;">' +
-              '<button class="btn-sm admin-embed-set-btn" data-id="' + esc(m.id) + '" data-dims="' + m.dims + '" style="padding:2px 8px;font-size:11px;background:var(--g100);color:var(--g700);border:none;border-radius:4px;cursor:pointer;">Set</button>' +
-              '<span style="flex:1;"><strong>' + esc(m.name) + '</strong> <span style="color:var(--g500);font-size:11px;">(' + m.dims + 'd)</span></span>' +
-              '<span style="font-size:11px;padding:1px 6px;border-radius:4px;background:var(--g100);color:var(--g600);">' + esc(m.tag || '') + '</span>' +
-              (isCurrent ? '<span style="font-size:11px;padding:1px 6px;border-radius:4px;background:var(--green-light,#d1fae5);color:var(--green);">ACTIVE</span>' : '') +
-              '</div>';
-          }).join('');
-        }
-      })
-      .catch(function() {});
-  }
-
-  function discoverEmbeddings() {
-    var search = (document.getElementById('admin-discover-search') || {}).value || '';
-    var container = document.getElementById('admin-discover-results');
-    var hint = document.getElementById('admin-discover-hint');
-    if (!container) return;
-    container.innerHTML = '<p style="font-size:13px;color:var(--g400);"><i class="fas fa-spinner fa-spin"></i> Querying provider...</p>';
-    if (hint) hint.hidden = true;
-
-    fetch('/api/admin/config/embeddings/discover?q=' + encodeURIComponent(search), { headers: getAuthHeaders() })
-      .then(function(r) { return r.json(); })
-      .then(function(data) {
-        if (!data.success) {
-          container.innerHTML = '<p style="font-size:13px;color:var(--red);">Error: ' + esc(data.error || 'Unknown') + '</p>';
-          return;
-        }
-        var items = data.models || [];
-        if (items.length === 0) {
-          container.innerHTML = '<p style="font-size:13px;color:var(--g400);">No models found' + (search ? ' matching "' + esc(search) + '"' : '') + '</p>';
-          return;
-        }
-        container.innerHTML = '<p style="font-size:12px;color:var(--g500);margin:0 0 6px;">Found ' + data.count + ' models (provider: ' + esc(data.provider) + ')</p>' +
-          items.map(function(m) {
-            return '<div style="display:flex;align-items:center;gap:8px;padding:5px 8px;border-radius:6px;background:var(--g50);font-size:13px;">' +
-              '<button class="btn-sm btn-primary admin-embed-set-btn" data-id="' + esc(m.id) + '" data-dims="' + (m.dims || '') + '" style="padding:2px 8px;font-size:11px;">Set</button>' +
-              '<span style="flex:1;">' + esc(m.name || m.id) + (m.dims && m.dims !== '?' ? ' <span style="color:var(--g500);font-size:11px;">(' + m.dims + 'd)</span>' : '') + '</span>' +
-              '<span style="font-size:10px;color:var(--g400);">' + esc(m.source || '') + '</span>' +
-              '</div>';
-          }).join('');
-      })
-      .catch(function(err) {
-        container.innerHTML = '<p style="font-size:13px;color:var(--red);">Request failed: ' + esc(err.message) + '</p>';
-      });
-  }
-
-  function setEmbeddingDefault(modelId, dims, btn) {
-    var origText = btn ? btn.textContent : '';
-    adminSetButtonText(btn, '...', true);
-    var promises = [
-      fetch('/api/admin/config/' + encodeURIComponent('embeddings.model'), {
-        method: 'PUT', headers: getAuthHeaders(), body: JSON.stringify({ value: modelId })
-      }).then(function(r) { return r.json(); })
-    ];
-    if (dims && dims !== '?') {
-      promises.push(
-        fetch('/api/admin/config/' + encodeURIComponent('embeddings.dimensions'), {
-          method: 'PUT', headers: getAuthHeaders(), body: JSON.stringify({ value: String(dims) })
-        }).then(function(r) { return r.json(); })
-      );
-    }
-    Promise.all(promises)
-      .then(function(results) {
-        var ok = results.every(function(r) { return r.success; });
-        adminSetButtonText(btn, 'Set', false);
-        adminFlashButtonBackground(btn, ok ? 'var(--green)' : '');
-        if (ok) { showToast('Embedding model set to: ' + modelId + (dims ? ' (' + dims + 'd)' : ''), 'success'); loadEmbeddingConfig(); }
-        else showToast(results[0].error || 'Failed', 'error');
-      })
-      .catch(function() {
-        adminSetButtonText(btn, origText, false);
-        showToast('Request failed', 'error');
-      });
-  }
-
-  function testEmbedding() {
-    var text = (document.getElementById('admin-embed-test-text') || {}).value || 'test';
-    var resultEl = document.getElementById('admin-embed-result');
-    var btn = document.getElementById('btn-test-embedding');
-    adminSetButtonHtml(btn, '<i class="fas fa-spinner fa-spin"></i>', true);
-    if (resultEl) resultEl.textContent = 'Generating...';
-
-    fetch('/api/admin/config/embeddings/test', {
-      method: 'POST',
-      headers: getAuthHeaders(),
-      body: JSON.stringify({ text: text })
-    })
-    .then(function(r) { return r.json(); })
-    .then(function(data) {
-      adminSetButtonHtml(btn, '<i class="fas fa-code-branch"></i> Generate', false);
-      if (!data.success) {
-        if (resultEl) resultEl.innerHTML = '<span style="color:var(--red);">Error: ' + esc(data.error || 'Failed') + '</span>';
-        return;
-      }
-      if (resultEl) {
-        resultEl.innerHTML =
-          '<strong>Dimensions:</strong> ' + data.dimensions + ' &nbsp;|&nbsp; ' +
-          '<strong>Model:</strong> ' + esc(data.model) + ' &nbsp;|&nbsp; ' +
-          '<strong>' + data.duration + 'ms</strong>' +
-          '<div style="margin-top:4px;font-family:monospace;font-size:11px;color:var(--g400);">Sample: [' + (data.sample || []).join(', ') + ', ...]</div>';
-      }
-    })
-    .catch(function(err) {
-      adminSetButtonHtml(btn, '<i class="fas fa-code-branch"></i> Generate', false);
-      if (resultEl) resultEl.textContent = 'Request failed: ' + err.message;
-    });
-  }
-
-}
--- a/src/routes/adminConfig.js
+++ b/src/routes/adminConfig.js
@ -15,7 +15,6 @@ var { gatewayUrl, serverError } = require('../utils/errors');
 var { getTTSEnvProvider, getLiteLLMTTSDiscoveryItems, getLiteLLMTTSRequestOptions, getLiteLLMTTSVoicesForModel, isLiteLLMTTSVoiceCompatible, getTTSProvider } = require('../utils/ttsProvider');
 var { getLiteLLMHeaders, getLiteLLMAdminHeaders } = require('../utils/litellm');
 var { getSTTDependencies, getLiteLLMSTTModels, getSTTModelLists, getSTTProvider } = require('../utils/sttProvider');
-var { getLiteLLMEmbeddingModels } = require('../utils/embeddings');

 router.use(authMiddleware);

@ -86,20 +85,6 @@ async function liteLLMVisionSupport(modelId) {
  }
 }

-async function probeLiteLLMEmbeddingDimensions(modelId) {
-  try {
-    var axios = require('axios');
-    var resp = await axios.post(gatewayUrl('/embeddings'), {
-      model: modelId,
-      input: 'dimension probe'
-    }, { headers: getLiteLLMHeaders('application/json'), timeout: 30000 });
-    var embedding = resp.data && resp.data.data && resp.data.data[0] && resp.data.data[0].embedding;
-    return Array.isArray(embedding) ? embedding.length : '?';
-  } catch (e) {
-    logger.warn('LiteLLM embedding dimension probe failed for ' + modelId + ': ' + e.message);
-    return '?';
-  }
-}

 // ── GET announcement (any authenticated user) ──────────────────────────────
 router.get('/config/announcement', async function(req, res) {
@ -925,87 +910,6 @@ router.post('/config/stt/test', async function(req, res) {
  }
 });

-// ── GET embedding model config ────────────────────────────────────────────
-router.get('/config/embeddings', async function(req, res) {
-  try {
-    var { isEmbeddingsAvailable, DEFAULT_MODEL, DEFAULT_DIMS } = require('../utils/embeddings');
-    var dbModel = await db.getSetting('embeddings.model') || '';
-    var dbDims  = await db.getSetting('embeddings.dimensions') || '';
-    var envModel = process.env.EMBEDDING_MODEL || DEFAULT_MODEL;
-    var envDims  = parseInt(process.env.EMBEDDING_DIMENSIONS) || DEFAULT_DIMS;
-
-    var provider = 'none';
-    if (process.env.LITELLM_API_BASE) provider = 'litellm';
-
-    res.json({
-      success: true,
-      provider: provider,
-      configured: isEmbeddingsAvailable(),
-      currentModel: dbModel || envModel,
-      currentDimensions: dbDims ? parseInt(dbDims) : envDims,
-      dbModel: dbModel,
-      dbDimensions: dbDims,
-      envModel: envModel,
-      envDimensions: envDims,
-      models: []
-    });
-  } catch (e) { res.status(500).json({ error: 'Request failed' }); }
-});
-
-// ── GET discover embedding models from provider ───────────────────────────
-router.get('/config/embeddings/discover', async function(req, res) {
-  try {
-    var search = (req.query.q || '').toLowerCase().trim();
-    var axios = require('axios');
-    var discovered = [];
-
-    var provider = 'none';
-    if (process.env.LITELLM_API_BASE) provider = 'litellm';
-
-    if (provider === 'litellm') {
-      try {
-        var eResp = await axios.get(liteLLMBaseUrl() + '/model/info', { headers: getLiteLLMAdminHeaders(), timeout: 10000 });
-        var embeddingModels = getLiteLLMEmbeddingModels(eResp.data && eResp.data.data);
-        for (var i = 0; i < embeddingModels.length; i++) {
-          var m = embeddingModels[i];
-          var dims = m.dims === '?' ? await probeLiteLLMEmbeddingDimensions(m.id) : m.dims;
-          discovered.push({ id: m.id, name: m.name, dims: dims, source: 'gateway-api', mode: 'embedding', capability: 'embedding' });
-        }
-      } catch(e) { logger.warn('LiteLLM embedding metadata discovery failed: ' + e.message); }
-    }
-
-    if (search) {
-      discovered = discovered.filter(function(d) {
-        return matchesDiscoverySearch(d, search, 'embedding vector');
-      });
-    }
-    res.json({ success: true, provider: provider, models: discovered, count: discovered.length });
-  } catch (e) { res.status(500).json({ error: 'Request failed' }); }
-});
-
-// ── POST test embedding ───────────────────────────────────────────────────
-router.post('/config/embeddings/test', async function(req, res) {
-  try {
-    var text = (req.body.text || 'Pediatric patient with fever').substring(0, 500);
-    var { generateEmbedding, DEFAULT_MODEL } = require('../utils/embeddings');
-    var db = require('../db/database');
-    var dbModel = await db.getSetting('embeddings.model') || '';
-    var start = Date.now();
-    var vector = await generateEmbedding(text);
-    var dims = Array.isArray(vector) ? vector.length : 0;
-    var sample = Array.isArray(vector) ? vector.slice(0, 8).map(function(v) { return v.toFixed(4); }) : [];
-    res.json({
-      success: true,
-      dimensions: dims,
-      sample: sample,
-      model: dbModel || process.env.EMBEDDING_MODEL || DEFAULT_MODEL,
-      duration: Date.now() - start
-    });
-  } catch (e) {
-    res.json({ success: false, error: e.message });
-  }
-});
-
 // ============================================================
 // WILDCARD CONFIG — Must come AFTER all specific model routes above
 // :key(*) matches slashes, so it would intercept /config/models/toggle
@ -1023,7 +927,7 @@ router.put('/config/:key(*)', async function(req, res) {
    }

    // Security: only allow known key prefixes
-    var allowed = ['announcement.', 'feature.', 'email.', 'prompt.', 'registration_enabled', 'registration_invite_only', 'site.', 'smtp.', 'models.', 'tts.', 'stt.', 'embeddings.', 'clinical_assistant.', 'my_resources.'];
+    var allowed = ['announcement.', 'feature.', 'email.', 'prompt.', 'registration_enabled', 'registration_invite_only', 'site.', 'smtp.', 'models.', 'tts.', 'stt.', 'clinical_assistant.', 'my_resources.'];
    var isAllowed = allowed.some(function(p) { return key === p || key.startsWith(p); });
    if (!isAllowed) {
      return res.status(400).json({ error: 'Unknown config key' });
--- a/src/utils/adminLockdown.js
+++ b/src/utils/adminLockdown.js
@ -21,7 +21,6 @@ var LOCKED_PREFIXES = Object.freeze([
  'models.',              // model policy: default, custom, enabled set
  'tts.',
  'stt.',
-  'embeddings.',
  'smtp.',                // where mail goes and who it authenticates as
  'email.'                // the templates that mail sends
 ]);
--- a/src/utils/embeddings.js
+++ b/src/utils/embeddings.js
@ -1,123 +0,0 @@
-// ============================================================
-// EMBEDDINGS UTILITY — Generate & search through LiteLLM embeddings
-// ============================================================
-
-var axios = require('axios');
-var { gatewayUrl } = require('./errors');
-var { getLiteLLMHeaders } = require('./litellm');
-
-var DEFAULT_MODEL = 'openai-text-embedding-3-large';
-var DEFAULT_DIMS = 3072;
-
-/**
- * Generate embedding for text using configured provider
- * @param {string} text - Text to embed (max ~2000 tokens)
- * @param {object} opts - Options: { model, dimensions }
- * @returns {Promise<number[]>} - Embedding vector
- */
-async function generateEmbedding(text, opts) {
-  opts = opts || {};
-  var dbModel, dbDims;
-  try {
-    var db = require('../db/database');
-    dbModel = await db.getSetting('embeddings.model') || '';
-    dbDims  = await db.getSetting('embeddings.dimensions') || '';
-  } catch(e) { /* DB not available during startup */ }
-  var model = opts.model || dbModel || process.env.EMBEDDING_MODEL || DEFAULT_MODEL;
-  var dimensions = opts.dimensions || (dbDims ? parseInt(dbDims) : 0) || parseInt(process.env.EMBEDDING_DIMENSIONS) || DEFAULT_DIMS;
-
-  // Truncate text to ~2000 tokens (~8000 chars) to avoid API errors
-  // NOTE: A large PDF is truncated to the first ~8000 chars for embedding.
-  // The full PDF content is still extracted and stored in the database body field.
-  // This is expected behavior - embeddings are semantic representations, not full-text storage.
-  var truncated = text.substring(0, 8000);
-  if (!truncated.trim()) {
-    throw new Error('Empty text provided for embedding');
-  }
-
-  if (process.env.LITELLM_API_BASE) {
-    return await generateEmbeddingLiteLLM(truncated, model, dimensions);
-  }
-
-  throw new Error('No embedding provider configured. Set LITELLM_API_BASE');
-}
-
-/**
- * Generate embedding via LiteLLM proxy
- */
-async function generateEmbeddingLiteLLM(text, model, dimensions) {
-  try {
-    var payload = {
-      model: model,
-      input: text
-    };
-
-    if (dimensions) {
-      payload.dimensions = dimensions;
-    }
-
-    var response = await axios.post(gatewayUrl('/embeddings'), payload, {
-      headers: getLiteLLMHeaders('application/json'),
-      timeout: 30000
-    });
-
-    if (!response.data || !response.data.data || !response.data.data[0]) {
-      throw new Error('Invalid response from LiteLLM embeddings API');
-    }
-
-    return response.data.data[0].embedding;
-  } catch (err) {
-    console.error('[Embeddings] LiteLLM error:', err.response?.data || err.message);
-    throw new Error('LiteLLM embedding failed: ' + (err.response?.data?.error || err.message));
-  }
-}
-
-/**
- * Search for similar content using cosine similarity
- * @param {string} queryText - Search query
- * @param {object} opts - Options: { limit, threshold, contentType }
- * @returns {Promise<Array>} - Matching content with similarity scores
- */
-function getLiteLLMModelId(model) {
-  if (!model) return '';
-  if (typeof model === 'string') return model;
-  return model.id || model.model_name || '';
-}
-
-function isLiteLLMEmbeddingModel(model) {
-  var mode = model && model.model_info && model.model_info.mode ? String(model.model_info.mode) : '';
-  return mode === 'embedding';
-}
-
-function getLiteLLMEmbeddingDimensions(model) {
-  var info = model && model.model_info ? model.model_info : {};
-  var dims = info.output_vector_size || info.dimensions || info.embedding_dimensions || model.output_vector_size || model.dimensions || model.embedding_dimensions;
-  var parsed = parseInt(dims, 10);
-  return Number.isFinite(parsed) ? parsed : '?';
-}
-
-function getLiteLLMEmbeddingModels(models) {
-  return (models || [])
-    .filter(isLiteLLMEmbeddingModel)
-    .map(function(model) {
-      var id = getLiteLLMModelId(model);
-      return { id: id, name: id, dims: getLiteLLMEmbeddingDimensions(model) };
-    })
-    .filter(function(model) { return !!model.id; });
-}
-
-/**
- * Check if embeddings are available (provider configured)
- */
-function isEmbeddingsAvailable() {
-  return !!process.env.LITELLM_API_BASE;
-}
-
-module.exports = {
-  generateEmbedding,
-  getLiteLLMEmbeddingModels,
-  isEmbeddingsAvailable,
-  isLiteLLMEmbeddingModel,
-  DEFAULT_MODEL,
-  DEFAULT_DIMS
-};
--- a/test/admin-clinical-assistant-wiring.test.js
+++ b/test/admin-clinical-assistant-wiring.test.js
@ -178,18 +178,18 @@ test('image model discovery is a kind in the shared Discover & test card, and en
  // One search box, one Search button, one result list and one hint for every kind.
  ['admin-discover-search', 'btn-discover', 'admin-discover-results', 'admin-discover-hint']
    .forEach(id => assert.ok(html.includes('id="' + id + '"'), 'admin.html has #' + id));
-  for (const kind of ['chat', 'image', 'tts', 'stt', 'embedding']) {
+  for (const kind of ['chat', 'image', 'tts', 'stt']) {
    assert.ok(html.includes('id="admin-discover-kind-' + kind + '"'), 'a kind switch for ' + kind);
  }
  assert.equal((html.match(/id="admin-discover-search"/g) || []).length, 1, 'exactly one search box');
  // No leftover per-kind search boxes from the five cards this replaced.
-  for (const id of ['admin-model-search', 'admin-image-search', 'admin-tts-search', 'admin-stt-search', 'admin-embed-search']) {
+  for (const id of ['admin-model-search', 'admin-image-search', 'admin-tts-search', 'admin-stt-search']) {
    assert.doesNotMatch(html, new RegExp('id="' + id + '"'), 'no separate #' + id);
  }

  // The kind switch dispatches; each discovery loader answers for its own kind.
  assert.match(js, /CustomEvent\('admin-discover'/);
-  for (const kind of ['chat', 'image', 'tts', 'stt', 'embedding']) {
+  for (const kind of ['chat', 'image', 'tts', 'stt']) {
    assert.match(js, new RegExp("e\\.detail\\.kind === '" + kind + "'\\) discover\\w+\\(\\);"), kind + ' listens');
  }
  assert.match(js, /'\/api\/admin\/config\/image-models\/discover\?q=' \+ encodeURIComponent\(search\)/,
--- a/test/backend-hardening.test.js
+++ b/test/backend-hardening.test.js
@ -237,7 +237,7 @@ test('admin lockdown refuses configuration writes at the server', () => {

  // On, configuration is locked and day-to-day operation is not.
  for (const key of ['prompt.hpi', 'clinical_assistant.chat_model', 'models.default',
-    'tts.voice', 'stt.model', 'embeddings.model', 'smtp.host', 'email.verify.subject']) {
+    'tts.voice', 'stt.model', 'smtp.host', 'email.verify.subject']) {
    assert.equal(lockdown.isLocked(key, on), true, key + ' is locked');
  }
  for (const key of ['announcement.text', 'registration_enabled',
--- a/test/embeddings-provider.test.js
+++ b/test/embeddings-provider.test.js
@ -1,25 +0,0 @@
-const { test } = require('node:test');
-const assert = require('node:assert/strict');
-
-test('LiteLLM embedding discovery uses embedding metadata only', () => {
-  const embeddings = require('../src/utils/embeddings');
-  assert.equal(embeddings.isLiteLLMEmbeddingModel({ id: 'openai-text-embedding-3-large' }), false);
-  assert.equal(embeddings.isLiteLLMEmbeddingModel({ id: 'mistral-codestral-embed' }), false);
-  assert.equal(embeddings.isLiteLLMEmbeddingModel({ model_name: 'openai-text-embedding-3-large', model_info: { mode: 'embedding' } }), true);
-  assert.equal(embeddings.isLiteLLMEmbeddingModel({ model_name: 'local-parakeet-v3', model_info: { mode: 'audio_transcription' } }), false);
-});
-
-test('LiteLLM embedding extraction includes metadata dimensions', () => {
-  const embeddings = require('../src/utils/embeddings');
-  assert.deepEqual(embeddings.getLiteLLMEmbeddingModels([
-    { model_name: 'openai-text-embedding-3-large', model_info: { mode: 'embedding', output_vector_size: 3072 } },
-    { model_name: 'openai-text-embedding-3-small', model_info: { mode: 'embedding', dimensions: '1536' } },
-    { model_name: 'mistral-embed', model_info: { mode: 'embedding' } },
-    { model_name: 'looks-like-embed' },
-    { model_name: 'local-kokoro-tts', model_info: { mode: 'audio_speech' } }
-  ]), [
-    { id: 'openai-text-embedding-3-large', name: 'openai-text-embedding-3-large', dims: 3072 },
-    { id: 'openai-text-embedding-3-small', name: 'openai-text-embedding-3-small', dims: 1536 },
-    { id: 'mistral-embed', name: 'mistral-embed', dims: '?' }
-  ]);
-});
--- a/test/prompt-administration.test.js
+++ b/test/prompt-administration.test.js
@ -110,7 +110,7 @@ async function application(t, svc, env = {}) {
    '../db/database': svc.db, '../middleware/auth': auth, '../utils/prompts': svc.prompts,
    '../utils/promptCatalog': svc.catalog, '../utils/promptRevisions': svc.revisions,
    '../utils/logger': { audit(actor, action, detail, req, meta) { logs.push({ actor, action, detail, meta }); } }, '../utils/errors': {},
-    '../utils/ttsProvider': {}, '../utils/litellm': {}, '../utils/sttProvider': {}, '../utils/embeddings': {}
+    '../utils/ttsProvider': {}, '../utils/litellm': {}, '../utils/sttProvider': {}
  }, env);
  const app = express();
  app.use(express.json()); app.use('/api/admin', router);