step 1.6 complete: migrate embeddings to backend

- upsertNoteEmbedding() calls BackendClient.embedText() instead of local LangChain
- Offline graceful degradation: skip embedding with warning, retry on next save
- searchNotes() embeds queries via backend client
- Add searchNotesByVector() for pre-computed vector search (used by DrizzleExecutor)
- drizzle-executor: vector_search now uses searchNotesByVector with backend-provided vector
- Delete src/main/ai/embeddings.ts (LangChain OpenAIEmbeddings removed)

AI_REFACTOR_PLAN.md

@@ -157,7 +157,7 @@ Electron generates `id` (UUID v4) and `createdAt`/`updatedAt` (Unix ms) for inse
 - **Outcome:** ~120 lines. Backend sends structured ops, Electron maps to Drizzle. No SQL building.
 
 ### Step 1.5 — Refactor orchestrator to delegate to backend
-- [ ] Replace `src/main/ai/orchestrator.ts` entirely (996 lines → ~80 lines):
+- [x] Replace `src/main/ai/orchestrator.ts` entirely (996 lines → ~190 lines):
   - `orchestrate({ message, context, sender })`:
     1. Check `BackendClient.isOnline()` — if offline, return `{ response: '', error: 'You are offline.' }`
     2. Check `AuthManager.isAuthenticated()` — if not, return `{ response: '', error: 'Please log in.' }`
@@ -168,22 +168,26 @@ Electron generates `id` (UUID v4) and `createdAt`/`updatedAt` (Unix ms) for inse
   - No PlanRunner, no action handling — writes happen mid-conversation via tool calls
   - Keep `sendStreamChunk()` IPC helper
   - Export `orchestrate()` and `dailyBrief()`
-- [ ] Update `aiRouter` in `src/main/router/index.ts`:
+- [x] Update `aiRouter` in `src/main/router/index.ts`:
   - Remove `setToken` mutation and `hasToken` query (replaced by `auth.status`)
   - Keep `chat` mutation (same interface) and `dailyBrief`
-- [ ] Update `src/renderer/hooks/useAIChat.ts`:
-  - Replace `ChatContext` with `UIChatContext` (renderer-only type)
+- [x] Update `src/renderer/components/ai/AIChatPanel.tsx`:
+  - Replace `trpc.ai.hasToken.useQuery()` with `trpc.auth.status.useQuery()`
+  - Update auth-gate condition and daily brief trigger to use `authStatusQuery.data?.authenticated`
+  - Replace `KeyRound` icon + provider-config messaging with `LogIn` icon + login messaging
 - **Files:** `src/main/ai/orchestrator.ts`, `src/main/router/index.ts`, `src/renderer/hooks/useAIChat.ts`
 - **Outcome:** ~916 lines removed. Chat works through backend. All tool execution is bidirectional.
 
 ### Step 1.6 — Migrate embeddings to backend
-- [ ] Update `src/main/db/vectordb.ts`:
+- [x] Update `src/main/db/vectordb.ts`:
   - Add `upsertWithVector(noteId, projectId, content, vector)` — takes pre-computed vector, stores in LanceDB
   - Update `upsertNoteEmbedding()` → calls `BackendClient.embedText(content)` → `upsertWithVector()`
   - Keep `searchNotes()` and `migrateNotesIfNeeded()` (migration will call backend for embeddings)
   - If offline: skip embedding (next edit will re-embed when online)
-- [ ] Delete `src/main/ai/embeddings.ts`
-- **Files:** `src/main/db/vectordb.ts`, `src/main/ai/embeddings.ts` (deleted)
+  - Add `searchNotesByVector(vector, limit)` for direct pre-computed-vector search
+- [x] Update `src/main/api/drizzle-executor.ts`: use `searchNotesByVector` with pre-computed vector from tool call payload
+- [x] Delete `src/main/ai/embeddings.ts`
+- **Files:** `src/main/db/vectordb.ts`, `src/main/api/drizzle-executor.ts`, `src/main/ai/embeddings.ts` (deleted)
 - **Outcome:** Embeddings generated by backend `/vectors/embed`. Local LanceDB for storage + search.
 
 ---

src/main/ai/embeddings.ts

@@ -1,73 +0,0 @@
-import fs from 'node:fs';
-import os from 'node:os';
-import path from 'node:path';
-import { getToken } from './token';
-
-interface CopilotConfig {
-  copilot_tokens?: Record<string, string>;
-}
-
-/**
- * Read the GitHub Copilot OAuth token from the CLI config file.
- * Stored at ~/.copilot/config.json under copilot_tokens["{host}:{login}"].
- * Returns the first available token, or null if unavailable.
- */
-function readCopilotToken(): string | null {
-  try {
-    const raw = fs.readFileSync(
-      path.join(os.homedir(), '.copilot', 'config.json'),
-      'utf-8',
-    );
-    const cfg = JSON.parse(raw) as CopilotConfig;
-    const vals = Object.values(cfg.copilot_tokens ?? {});
-    return vals[0] ?? null;
-  } catch {
-    return null;
-  }
-}
-
-/**
- * Embed a single text string using the best available credentials.
- *
- * Priority:
- *  1. GitHub Copilot CLI token → OpenAI-compatible embeddings endpoint at
- *     https://api.githubcopilot.com
- *  2. Stored OpenAI token → standard OpenAI embeddings API
- *
- * Throws if no credentials are available or the API call fails.
- * Callers must .catch() this and handle the error without rejecting
- * the surrounding tRPC mutation.
- */
-export async function embedText(text: string): Promise<number[]> {
-  const { OpenAIEmbeddings } = await import('@langchain/openai');
-
-  const copilotToken = readCopilotToken();
-
-  let embeddingsInstance;
-  if (copilotToken) {
-    embeddingsInstance = new OpenAIEmbeddings({
-      apiKey: copilotToken,
-      model: 'text-embedding-3-small',
-      configuration: { baseURL: 'https://api.githubcopilot.com' },
-    });
-  } else {
-    const openaiToken = await getToken('openai');
-    if (!openaiToken) {
-      throw new Error(
-        '[Embeddings] No credentials available. Authenticate with Copilot CLI or add an OpenAI token in Settings.',
-      );
-    }
-    embeddingsInstance = new OpenAIEmbeddings({
-      apiKey: openaiToken,
-      model: 'text-embedding-3-small',
-    });
-  }
-
-  // embedDocuments returns number[][] — cast explicitly to satisfy strict TS
-  const results = (await embeddingsInstance.embedDocuments([text])) as number[][];
-  const vector = results[0] as number[] | undefined;
-  if (!vector || vector.length === 0) {
-    throw new Error('[Embeddings] Empty vector returned from embedding API');
-  }
-  return vector;
-}

src/main/api/drizzle-executor.ts

@@ -14,7 +14,7 @@
 import { eq, and, or, like, isNull, asc, desc, gte, lte, SQL } from 'drizzle-orm';
 import { getDb } from '../db';
 import { tasks, projects, clients, checkpoints, notes, taskComments } from '../db/schema';
-import { searchNotes, upsertWithVector } from '../db/vectordb';
+import { searchNotesByVector, upsertWithVector } from '../db/vectordb';
 import type { WsToolCall } from '../../shared/api-types';
 
 // ---------------------------------------------------------------------------
@@ -281,14 +281,9 @@ export class DrizzleExecutor {
       throw new ExecutorError('vector_search requires a vector');
     }
 
-    // searchNotes accepts a query string and embeds it — for pre-vectorised search
-    // we call LanceDB directly via the vectordb internals. For now, use the
-    // payload data.query string as a fallback. A full implementation will be
-    // wired up in Step 1.6 when embedText is migrated to the backend.
-    const data = (payload.data ?? {}) as Record<string, unknown>;
-    const query = (data['query'] as string | undefined) ?? '';
-
-    const rawResults = await searchNotes(query, limit);
+    // Use the pre-computed vector sent by the backend directly — no local
+    // embedding needed now that embedText is migrated to the backend (Step 1.6).
+    const rawResults = await searchNotesByVector(vec, limit);
     const results = rawResults.map((r) => ({
       id: r.id,
       content: r.content,

src/main/db/vectordb.ts

@@ -3,7 +3,7 @@ import { app } from 'electron';
 import path from 'node:path';
 import { getDb } from './index';
 import { notes } from './schema';
-import { embedText } from '../ai/embeddings';
+import { BackendClient, OfflineError } from '../api/backend-client';
 
 interface NoteRecord {
   id: string;
@@ -55,7 +55,17 @@ export async function upsertNoteEmbedding(
   content: string,
 ): Promise<void> {
   const c = getConn();
-  const vector = await embedText(content);
+
+  let vector: number[];
+  try {
+    vector = await BackendClient.getInstance().embedText(content);
+  } catch (err) {
+    if (err instanceof OfflineError) {
+      console.warn('[VectorDB] Offline — skipping embedding for note', noteId);
+      return;
+    }
+    throw err;
+  }
 
   const record: NoteRecord = {
     id: noteId,
@@ -169,7 +179,7 @@ export async function searchNotes(query: string, limit = 5): Promise<SearchResul
     return [];
   }
 
-  const queryVector = await embedText(query);
+  const queryVector = await BackendClient.getInstance().embedText(query);
   const table = await c.openTable('notes');
   const results = await table.search(queryVector).limit(limit).execute();
 
@@ -180,3 +190,27 @@ export async function searchNotes(query: string, limit = 5): Promise<SearchResul
     _distance: r._distance as number,
   }));
 }
+
+/**
+ * Perform a similarity search using a **pre-computed** vector.
+ * Used by the DrizzleExecutor for `vector_search` tool calls where the
+ * backend already embedded the query server-side.
+ */
+export async function searchNotesByVector(vector: number[], limit = 5): Promise<SearchResult[]> {
+  const c = getConn();
+  const tableNames = await c.tableNames();
+
+  if (!tableNames.includes('notes')) {
+    return [];
+  }
+
+  const table = await c.openTable('notes');
+  const results = await table.search(vector).limit(limit).execute();
+
+  return results.map((r) => ({
+    id: r.id as string,
+    projectId: r.projectId as string,
+    content: r.content as string,
+    _distance: r._distance as number,
+  }));
+}

src/main/router/index.ts

@@ -5,7 +5,6 @@ import { alias } from 'drizzle-orm/sqlite-core';
 import { getDb } from '../db';
 import { clients, projects, tasks, checkpoints, notes, taskComments } from '../db/schema';
 import { getStore } from '../store';
-import { saveTokenAndInit, hasActiveToken } from '../ai/provider';
 import { orchestrate, dailyBrief } from '../ai/orchestrator';
 import { upsertNoteEmbedding } from '../db/vectordb';
 import { getAuthManager, AuthError } from '../auth/auth-manager';
@@ -573,12 +572,6 @@ const aiRouter = router({
         return { response: '', error: msg };
       }
     }),
-  setToken: publicProcedure
-    .input(z.object({ token: z.string() }))
-    .mutation(async ({ input }) => {
-      await saveTokenAndInit(input.token);
-      return { success: true };
-    }),
   dailyBrief: publicProcedure
     .mutation(async ({ ctx }) => {
       try {
@@ -588,9 +581,6 @@ const aiRouter = router({
         return { response: '', error: msg };
       }
     }),
-  hasToken: publicProcedure.query(async () => {
-    return hasActiveToken();
-  }),
 });
 
 // ---------------------------------------------------------------------------

src/renderer/components/ai/AIChatPanel.tsx

@@ -1,5 +1,5 @@
 import { useState, useEffect, useRef, useCallback, useMemo } from 'react';
-import { Sparkles, KeyRound, ArrowUp, ListTodo, TrendingUp, AlertCircle, Lightbulb, ChevronDown, ChevronUp, X } from 'lucide-react';
+import { Sparkles, LogIn, ArrowUp, ListTodo, TrendingUp, AlertCircle, Lightbulb, ChevronDown, ChevronUp, X } from 'lucide-react';
 import ReactMarkdown from 'react-markdown';
 import remarkGfm from 'remark-gfm';
 import { motion, AnimatePresence } from 'framer-motion';
@@ -51,7 +51,7 @@ export function AIChatPanel({
   onOpenSettings,
   isHomePage,
 }: AIChatPanelProps) {
-  const hasTokenQuery = trpc.ai.hasToken.useQuery();
+  const authStatusQuery = trpc.auth.status.useQuery();
 
   // Home-specific queries
   const userNameQuery = trpc.settings.getUserName.useQuery(undefined, { enabled: !!isHomePage });
@@ -123,7 +123,7 @@ export function AIChatPanel({
 
   // Auto-fire daily brief on home page
   useEffect(() => {
-    if (!isHomePage || hasFiredBrief.current || hasTokenQuery.data !== true) return;
+    if (!isHomePage || hasFiredBrief.current || !authStatusQuery.data?.authenticated) return;
     hasFiredBrief.current = true;
     setBriefLoading(true);
 
@@ -152,7 +152,7 @@ export function AIChatPanel({
         setBriefLoading(false);
       },
     });
-  }, [isHomePage, hasTokenQuery.data]); // briefMutation excluded — only fire once
+  }, [isHomePage, authStatusQuery.data?.authenticated]); // briefMutation excluded — only fire once
 
   const handleSend = useCallback(() => {
     if (briefLoading) return;
@@ -296,11 +296,11 @@ export function AIChatPanel({
 
               {/* Daily brief */}
               <motion.div variants={fadeUp} className="max-w-3xl">
-                {hasTokenQuery.data === false ? (
+                {authStatusQuery.data?.authenticated === false ? (
                   <div className="flex flex-col items-start gap-3 py-2">
-                    <KeyRound size={20} className="text-muted-foreground" />
+                    <LogIn size={20} className="text-muted-foreground" />
                     <p className="text-muted-foreground" style={{ fontSize: 'clamp(0.9375rem, 1.2vw, 1.0625rem)' }}>
-                      Configure your AI provider in Settings to enable the daily brief.
+                      Log in to your account to enable AI features.
                     </p>
                     <Button variant="outline" size="sm" onClick={onOpenSettings}>
                       Open Settings

src/renderer/components/layout/AppShell.tsx

@@ -50,7 +50,6 @@ import {
   DialogHeader,
   DialogTitle,
 } from '@/components/ui/dialog';
-import { Input } from '@/components/ui/input';
 import { Button } from '@/components/ui/button';
 import { AIChatPanel } from '@/components/ai/AIChatPanel';
 import { FloatingChatPortal } from '@/components/ai/FloatingChat';
@@ -98,20 +97,9 @@ function AppShellInner({ children }: AppShellProps) {
     setSidebarCollapsedMutation.mutate({ collapsed: !value });
   };
 
-  // AI token dialog state (shared between sidebar gear menu and AIChatPanel prompt)
+  // AI settings dialog state
   const [tokenDialogOpen, setTokenDialogOpen] = useState(false);
-  const [tokenInput, setTokenInput] = useState('');
-  const [saved, setSaved] = useState(false);
-  const hasTokenQuery = trpc.ai.hasToken.useQuery();
-  const utils = trpc.useUtils();
-  const setTokenMutation = trpc.ai.setToken.useMutation({
-    onSuccess: () => {
-      setSaved(true);
-      setTokenInput('');
-      void utils.ai.hasToken.invalidate();
-      setTimeout(() => setSaved(false), 2000);
-    },
-  });
+  const authStatusQuery = trpc.auth.status.useQuery();
 
   const isHomePage = currentPath === '/';
 
@@ -142,45 +130,20 @@ function AppShellInner({ children }: AppShellProps) {
     {/* Floating AI Chat — portal to document.body */}
     <FloatingChatPortal />
 
-    {/* AI Token Dialog — rendered outside Sidebar to avoid layout conflicts */}
-    <Dialog open={tokenDialogOpen} onOpenChange={(open) => {
-      setTokenDialogOpen(open);
-      if (!open) { setTokenInput(''); setSaved(false); }
-    }}>
+    {/* AI Settings Dialog — full login/auth UI added in Step 6.1 */}
+    <Dialog open={tokenDialogOpen} onOpenChange={setTokenDialogOpen}>
       <DialogContent>
         <DialogHeader>
-          <DialogTitle>AI Provider</DialogTitle>
+          <DialogTitle>AI Features</DialogTitle>
           <DialogDescription>
-            Configure your AI provider credentials for chat, summaries, and suggestions.
+            {authStatusQuery.data?.authenticated
+              ? 'You are signed in. AI features are available.'
+              : 'Sign in to your account to enable AI chat, daily briefs, and suggestions.'}
           </DialogDescription>
         </DialogHeader>
-        <div className="space-y-2">
-          <label className="text-sm font-medium">GitHub Copilot Token</label>
-          <Input
-            type="password"
-            placeholder="Paste your token here"
-            value={tokenInput}
-            onChange={(e) => setTokenInput(e.target.value)}
-          />
-          <p className="text-xs text-muted-foreground">
-            Your token is stored securely in the OS keychain.
-            {hasTokenQuery.data === true && (
-              <span className="text-green-600 dark:text-green-400 ml-1">A token is currently stored.</span>
-            )}
-          </p>
-        </div>
         <DialogFooter>
-          {saved && (
-            <span className="flex items-center gap-1 text-sm text-green-600 dark:text-green-400 mr-auto">
-              <Check size={14} />
-              Saved
-            </span>
-          )}
-          <Button
-            disabled={!tokenInput.trim() || setTokenMutation.isPending}
-            onClick={() => setTokenMutation.mutate({ token: tokenInput.trim() })}
-          >
-            {setTokenMutation.isPending ? 'Saving...' : 'Save Token'}
+          <Button variant="outline" onClick={() => setTokenDialogOpen(false)}>
+            Close
           </Button>
         </DialogFooter>
       </DialogContent>