Feat(ai-gemini): Gemini Realtime Adapter

.changeset/huge-lizards-admire.md

@@ -0,0 +1,5 @@
+---
+'@tanstack/ai-gemini': minor
+---
+
+Added Gemini Realtime Adapter

packages/typescript/ai-gemini/package.json

@@ -40,14 +40,16 @@
     "adapter"
   ],
   "dependencies": {
-    "@google/genai": "^1.43.0"
+    "@google/genai": "^1.46.0"
   },
   "peerDependencies": {
-    "@tanstack/ai": "workspace:^"
+    "@tanstack/ai": "workspace:^",
+    "@tanstack/ai-client": "workspace:^"
   },
   "devDependencies": {
     "@tanstack/ai": "workspace:*",
+    "@tanstack/ai-client": "workspace:*",
     "@vitest/coverage-v8": "4.0.14",
-    "vite": "^7.2.7"
+    "vite": "^7.3.1"
   }
 }

packages/typescript/ai-gemini/src/realtime/adapter.ts

@@ -0,0 +1,52 @@
+import type {
+  AnyClientTool,
+  RealtimeEvent,
+  RealtimeEventHandler,
+  RealtimeToken,
+} from '@tanstack/ai'
+import type { RealtimeAdapter, RealtimeConnection } from '@tanstack/ai-client'
+import type { GeminiRealtimeOptions } from './types'
+
+/**
+ * Creates a Gemini realtime adapter for client-side use.
+ *
+ * @param options - Optional configuration
+ * @returns A RealtimeAdapter for use with RealtimeClient
+ *
+ * @example
+ * ```typescript
+ * import { RealtimeClient } from '@tanstack/ai-client'
+ * import { geminiRealtime } from '@tanstack/ai-gemini'
+ *
+ * const client = new RealtimeClient({
+ *   getToken: () => fetch('/api/realtime-token').then(r => r.json()),
+ *   adapter: geminiRealtime(),
+ * })
+ * ```
+ */
+export function geminiRealtime(
+  options: GeminiRealtimeOptions = {},
+): RealtimeAdapter {
+  return {
+    provider: 'gemini',
+
+    connect(
+      token: RealtimeToken,
+      _clientTools?: ReadonlyArray<AnyClientTool>,
+    ): Promise<RealtimeConnection> {
+      return createWebSocketConnection(token)
+    },
+  }
+}
+
+/**
+ * Creates a WebSocket connection to Gemini's realtime API
+ */
+function createWebSocketConnection(
+  token: RealtimeToken,
+): Promise<RealtimeConnection> {
+  const model = token.config.model ?? 'gemini-live-2.5-flash-native-audio'
+  const eventHandlers = new Map<RealtimeEvent, Set<RealtimeEventHandler<any>>>()
+
+  return new Promise((resolve, reject) => {})
+}

packages/typescript/ai-gemini/src/realtime/index.ts

@@ -0,0 +1,12 @@
+// Token adapter for server-side use
+export { geminiRealtimeToken } from './token'
+
+// Client adapter for browser use
+export { geminiRealtime } from './adapter'
+
+// Types
+export type {
+  GeminiRealtimeModel,
+  GeminiRealtimeTokenOptions,
+  GeminiRealtimeOptions,
+} from './types'

packages/typescript/ai-gemini/src/realtime/token.ts

@@ -0,0 +1,78 @@
+import { GoogleGenAI, Modality } from '@google/genai'
+import { getGeminiApiKeyFromEnv } from '../utils'
+import type { RealtimeToken, RealtimeTokenAdapter } from '@tanstack/ai'
+import type { GeminiRealtimeModel, GeminiRealtimeTokenOptions } from './types'
+
+/**
+ * Creates a Google Gemini realtime token adapter.
+ *
+ * This adapter generates ephemeral tokens for client-side WebSocket connections.
+ *
+ * @param options - Configuration options for the realtime session
+ * @returns A RealtimeTokenAdapter for use with realtimeToken()
+ *
+ * @example
+ * ```typescript
+ * import { realtimeToken } from '@tanstack/ai'
+ * import { geminiRealtimeToken } from '@tanstack/ai-gemini'
+ *
+ * const token = await realtimeToken({
+ *   adapter: geminiRealtimeToken({
+ *     model: 'gemini-live-2.5-flash-native-audio',
+ *   }),
+ * })
+ * ```
+ */
+export function geminiRealtimeToken(
+  options: GeminiRealtimeTokenOptions = {},
+): RealtimeTokenAdapter {
+  const apiKey = getGeminiApiKeyFromEnv()
+
+  const client = new GoogleGenAI({
+    apiKey,
+  })
+
+  // Defaults to 30 minutes
+  const expireTime = options.expiresAt ?? Date.now() + 30 * 60 * 1000
+
+  return {
+    provider: 'gemini',
+    async generateToken(): Promise<RealtimeToken> {
+      const model: GeminiRealtimeModel =
+        options.model ?? 'gemini-live-2.5-flash-native-audio'
+
+      const token = await client.authTokens.create({
+        config: {
+          uses: 1, // The default
+          expireTime: new Date(expireTime).toISOString(),
+          liveConnectConstraints: {
+            model,
+            config: {
+              sessionResumption: {},
+              maxOutputTokens: options.maxOutputTokens,
+              responseModalities: [Modality.AUDIO],
+            },
+          },
+          httpOptions: {
+            apiVersion: 'v1alpha',
+          },
+        },
+      })
+
+      if (!token.name) {
+        throw new Error('Gemini realtime token creation failed')
+      }
+
+      return {
+        provider: 'gemini',
+        token: token.name,
+        expiresAt: expireTime,
+        config: {
+          model,
+          maxOutputTokens: options.maxOutputTokens,
+          outputModalities: ['audio'],
+        },
+      }
+    },
+  }
+}

packages/typescript/ai-gemini/src/realtime/types.ts

@@ -0,0 +1,27 @@
+/**
+ * Gemini realtime model options
+ */
+export type GeminiRealtimeModel = 'gemini-live-2.5-flash-native-audio'
+
+/**
+ * Options for the Gemini realtime client adapter
+ */
+export interface GeminiRealtimeOptions {
+  /** Connection mode (default: 'websocket' in browser) */
+  connectionMode?: 'websocket'
+}
+
+/**
+ * Options for the Gemini realtime token adapter
+ */
+export interface GeminiRealtimeTokenOptions {
+  /** Model to use (default: 'gemini-live-2.5-flash-native-audio') */
+  model?: GeminiRealtimeModel
+  expiresAt?: number
+  maxOutputTokens?: number
+}
+
+/**
+ * Gemini Realtime session response from the API
+ */
+export interface GeminiRealtimeSessionResponse {}