fixes

Author: NikhilShahi

libs/langgraph-cua/package.json

@@ -33,7 +33,7 @@
   "license": "MIT",
   "dependencies": {
     "@hyperbrowser/sdk": "^0.40.0",
-    "playwright-core": "^1.51.1",
+    "puppeteer-core": "^24.5.0",
     "scrapybara": "^2.4.4",
     "zod": "^3.23.8"
   },

libs/langgraph-cua/src/index.ts

@@ -7,6 +7,7 @@ import {
   StateGraph,
 } from "@langchain/langgraph";
 import { SystemMessage } from "@langchain/core/messages";
+import { CreateSessionParams } from "@hyperbrowser/sdk/types";
 import { callModel } from "./nodes/call-model.js";
 import { createVMInstance } from "./nodes/create-vm-instance.js";
 import { takeComputerAction } from "./nodes/take-computer-action.js";
@@ -89,7 +90,7 @@ interface CreateCuaParams<
    * For more information on the available parameters, see the [Hyperbrowser API documentation](https://docs.hyperbrowser.ai/sessions/overview/session-parameters).
    * @default undefined
    */
-  sessionParams?: Record<string, unknown>;
+  sessionParams?: CreateSessionParams;
 
   /**
    * The number of hours to keep the virtual machine running before it times out.

libs/langgraph-cua/src/nodes/create-vm-instance.ts

@@ -1,17 +1,28 @@
 import { LangGraphRunnableConfig } from "@langchain/langgraph";
-import { chromium } from "playwright-core";
+import type { Browser } from "puppeteer-core";
+import { connect } from "puppeteer-core";
 import { UbuntuInstance, BrowserInstance, WindowsInstance } from "scrapybara";
 import { SessionDetail } from "@hyperbrowser/sdk/types";
 import { CUAState, CUAUpdate, getConfigurationWithDefaults } from "../types.js";
 import { getHyperbrowserClient, getScrapybaraClient } from "../utils.js";
 
+export const getActivePage = async (browser: Browser) => {
+  const pages = await browser.pages();
+  for (const page of pages) {
+    const isHidden = await page.evaluate("document.hidden");
+    if (isHidden === false) {
+      return page;
+    }
+  }
+  return pages[0];
+};
+
 async function createHyperbrowserInstance(
   state: CUAState,
   config: LangGraphRunnableConfig
 ): Promise<CUAUpdate> {
   const { hyperbrowserApiKey, sessionParams } =
     getConfigurationWithDefaults(config);
-  let { browserState } = state;
 
   if (!hyperbrowserApiKey) {
     throw new Error(
@@ -22,18 +33,16 @@ async function createHyperbrowserInstance(
   const client = getHyperbrowserClient(hyperbrowserApiKey);
   const session: SessionDetail = await client.sessions.create(sessionParams);
 
-  if (!browserState && session.wsEndpoint) {
-    const browser = await chromium.connectOverCDP(
-      `${session.wsEndpoint}&keepAlive=true`
-    );
-    const currPage = browser.contexts()[0].pages()[0];
-    if (currPage.url() === "about:blank") {
-      await currPage.goto("https://www.google.com");
+  if (session.wsEndpoint) {
+    const browser = await connect({
+      browserWSEndpoint: `${session.wsEndpoint}&keepAlive=true`,
+      defaultViewport: null,
+    });
+    const page = await getActivePage(browser);
+
+    if (page.url() === "about:blank") {
+      await page.goto("https://www.google.com");
     }
-    browserState = {
-      browser,
-      currentPage: currPage,
-    };
   }
 
   if (!state.streamUrl) {
@@ -43,13 +52,11 @@ async function createHyperbrowserInstance(
     return {
       instanceId: session.id,
       streamUrl,
-      browserState,
     };
   }
 
   return {
     instanceId: session.id,
-    browserState,
   };
 }