Stagehand agent can now call tools and integrate with third-party MCP servers! 
This increases efficiency for certain tasks (web search, authentication, google workspace) and expands the scope of what our agent can do (db insertions, memory/reasoning, custom tools). 

![](https://cdn.sanity.io/images/v773ixxh/staging/7f78a6597fddcf39eb88500745194b5d7091fd75-2048x1638.png?w=450)


Stagehand agent can now call tools

**Stagehand TS 2.4.3**

Shadow DOM support! Set experimental:true in the Stagehand constructor to try it out
* Fix for same-process iframes
* Enabling scrolling within iframes
* Handle namespaced elements in xpaths
* Bump Zod to be compatible with v4
* Patch for new GPT-5 api format

**Stagehand Python 0.5.1**
* Shadow DOM support for env="BROWSERBASE" (also requires experimental=True)
* Multi-tab support! S/o Arun
* LLM api_base customization! S/o @Sanveer Singh Osahan
* Fix for stagehand.metrics on Browserbase
* Fix for agent max_steps parameter
* Fix for Litellm dependency errors with OpenAI on newer versions
* Patch automatic download behavior! S/o @Shubhankar
* Patch for new GPT-5 api format

Stagehand: New release feat. shadow DOM support

We’ve been hard at work expanding what you can do with Browserbase. 

Here’s what’s new:
* Cloudflare Signed Agents - Powered by Browserbase Identity, select customers will be allowed to bypass Cloudflare in an official capacity.
* Expanded coverage for the hardest bot defenses
* Our own in-house solvers for increased reliability

Curious about our Advanced Stealth features? [Request a trial -> here](/https://www.browserbase.com/contact-web-bot-auth)

Stealth just got a huge upgrade

Lots of people use Director for web scraping, but it's not useful if you can't download the data your agent spent so much time gathering.

* Scroll up in the chat to find the "CSV download card"
* For now, this works for all CSVs < 1,000 rows


![](https://cdn.sanity.io/images/v773ixxh/staging/c851cc810703affce17395aaa9f33fd16486789f-1544x982.png?w=450)

You can now download CSVs in Director.ai

Now in the sessions list: filter by status!

![](https://cdn.sanity.io/images/v773ixxh/staging/afd15deee37f33efe4f67be0e2d630964b92feb9-860x604.png?w=450)

Filter your sessions by status

Now when a session has ended and you can no longer interact with a task, the chat is disabled! Huge UX improvement, thanks for the feedback!
![](https://cdn.sanity.io/images/v773ixxh/staging/3e1317164a152a8a99e00deda83f0a5a255201a3-3124x1718.png?w=450)

Director.ai - Chat disabled when session ends

We're so excited to announce that we're SOC 2 Type II certified. We also have SSO for Scale plan customers.

SOC 2 Type II, officially.

Just shipped a few changes to uploading extensions in API:
* We now accept zipped extensions up to 100 MB
* We now return a 400 if an extension exceeds 100 MB instead of a 500

Uploading extensions in API

Never lose your session again!

You can now search for a specific session ID in the dashboard.

Enter the sessionID and find any session from the last 30 days.

You can now search for a specific session ID in the dashboard

Introducing the Browserbase MCP🛠️

We've updated our tools to be powered by Stagehand, our AI browser automation SDK that is more powerful than direct Playwright calls.

You can now use our remote hosted server and run browser sessions in parallel, unlocking more use cases than ever.

Try it out: https://github.com/browserbase/mcp-server-browserbase

Introducing the Browserbase MCP

Web automation for everyone!

Director makes it so you don't have to start writing code to use Browserbase. You can start with the task that you already know—the boring work that you want to automate. And then, using AI by chatting with the application, it outputs a Stagehand script that controls the browser.

Where's that browser running? Browserbase—the best infrastructure platform for running browsers at scale in the cloud.

Try it now at [Director.ai](https://director.ai)

 
![](https://cdn.sanity.io/images/v773ixxh/staging/a4ed9aed410ad243d76ebbe79a52b107c2363176-1714x1175.png)


Introducing Director

We're excited to finally announce Stagehand for Python.

To get started:

1.  Run pip install stagehand
    
2.  Review the Github: [https://github.com/browserbase/stagehand-python](https://github.com/browserbase/stagehand-python)
    
3.  Automate the web
    

Have fun!

Stagehand for Python

The Browserbase playground is used to develop and test scripts before putting them into production. Previously, our playground only supported Playwright, but we've now added our open source browser automation framework, Stagehand. 

You'll see new script templates, a "convert to Stagehand" button, and you can add in new commands in the console after running. 


![](https://cdn.sanity.io/images/v773ixxh/staging/557a4ef7a3575765b45efac6c3ffa9e0bf30115b-2048x1136.png)


Stagehand is now in the playground

Our new Hobby plan is only $20 a month, renamed → "Developer Plan". For current Hobby customers, this will be reflected in your next invoice.

We've also increased concurrency across every plan!

See all the changes on [https://browserbase.com/pricing](https://browserbase.com/pricing)


![](https://cdn.sanity.io/images/v773ixxh/staging/1e969792c6f214559a3499ce8023549424cdbbe8-1024x475.png)



Massive price decrease!

Leveraging the Stagehand API, Browserbase can now offer a deeper look into Stagehand script executions. This enhanced observability will help users better understand and optimize their Stagehand usage and also provide debugging insights that significantly reduce troubleshooting time and improve the overall developer experience.

Key features include method inspection views, element highlighting for actions, and detailed execution metrics.


![](https://cdn.sanity.io/images/v773ixxh/staging/b6057ca54e40dcc25d8f944ab27c678d176e860f-2048x1116.png)


Stagehand now in the Browserbase dashboard

Browserbase has moved all remaining customers from single user accounts to multi-user and multi-project supported Organizations.

This means your team can have their own logins, see sessions across shared projects, and manage multiple projects under one org. Or if you're building alone, you can have multiple projects within your account.

No action required on your part. [Read more here.](https://docs.browserbase.com/guides/manage-account)

New account structure - Organizations

We now ship the `captchaImageSelector` and `captchaInputSelector` browser settings in the API docs and in the SDKs!

  

```
const sesssion = await bb.sessions.create(body: {
  projectID: BROWSERBASE\_PROJECT\_ID,
  browserSettings: {
    captchaImageSelector: "",
    captchaInputSelector: "",
    }
});
```

We also now have two new versions of our SDKs with a lot of internal improvements!

*   Node: v2.6.0: [https://www.npmjs.com/package/@browserbasehq/sdk/v/2.6.0](https://www.npmjs.com/package/@browserbasehq/sdk/v/2.6.0)
    
*   Python: v1.4.0: [https://pypi.org/project/browserbase/1.4.0/](https://pypi.org/project/browserbase/1.4.0/)
    

**Read more**: [https://docs.browserbase.com/features/stealth-mode#custom-captcha-solving](https://docs.browserbase.com/features/stealth-mode#custom-captcha-solving)

New CAPTCHA details in API docs & SDKs

**Keep-alive is more reliable now!**

When a session is created with `keepAlive: true`, a protection is set on the browser. This prevents it from being stopped on disconnect. The session is only stopped if:

*   The user stops the session via the API or dashboard
    
*   The session expires
    
*   Connect stops the session due to an error
    

Previously, the keepAlive protection expired based on a coarse timer. It now uses more precise timing, making protection behavior more accurate and reliable.

**Read more**: [https://docs.browserbase.com/guides/long-running-sessions#keep-alive-sessions](https://docs.browserbase.com/guides/long-running-sessions#keep-alive-sessions)

Increased reliability for keepAlive

Session cookies are now fully persisted resulting in fewer session timeouts!

**What changed?**

Basic & Advanced Stealth were both updated such that contexts will now persist session cookies (cookies with `expires: -1`). The cache that is saved from session to session has been slimmed down, decreasing the rate at which contexts grow in size with `persist: true`.

**What results should customers see?**

Primarily, this increases the lifespan of the context 🎉 Sessions created with contexts are significantly less likely to time out when you try to connect.

This also means login flows are persisted across sessions more reliably.

**Resources**:

*   **Learn more**: [https://docs.browserbase.com/features/contexts](https://docs.browserbase.com/features/contexts)
    
*   **Code Examples:** [Contexts code example](https://github.com/browserbase/playbook/tree/main/node/playwright/_tools/context) & [using Stagehand to login to Github](https://github.com/browserbase/playbook/blob/main/node/stagehand/authenticate/githubLogin.ts)

New and improved Contexts API

Browserbase now supports UI-based workflows involving native `select` elements! Previously, clicking on a `select` HTML element in the live view wouldn't show the native `options` popup, which meant that the only way users could edit these elements was via keyboard. This is now fixed! Users in the live view will now see a menu as expected. Click away :)

Native select support is out!

We're thrilled to announce the release of Stagehand 2.0, bringing significant improvements to make browser automation more powerful, faster, and easier to use than ever before.

### 🚀 New Features

*   **Introducing** `stagehand.agent`: A powerful new way to integrate SOTA Computer use models or Browserbase's [Open Operator](https://operator.browserbase.com/) into Stagehand with one line of code! Perfect for multi-step workflows and complex interactions. [Learn more](https://docs.stagehand.dev/concepts/agent)
    
*   **Lightning-fast** `act` **and** `extract`: Major performance improvements to make your automations run significantly faster.
    
*   **Enhanced Logging**: Better visibility into what's happening during automation with improved logging and debugging capabilities.
    
*   **Comprehensive Documentation**: A completely revamped documentation site with better examples, guides, and best practices.
    
*   **Improved Error Handling**: More descriptive errors and better error recovery to help you debug issues faster.
    

### 🛠️ Developer Experience

*   **Better TypeScript Support**: Enhanced type definitions and better IDE integration
    
*   **Better Error Messages**: Clearer, more actionable error messages to help you debug faster
    
*   **Improved Caching**: More reliable action caching for better performance
    

We're excited to see what you build with Stagehand 2.0! For questions or support, join our [Slack community](https://stagehand.dev/slack).

For more details, check out our [documentation](https://docs.stagehand.dev/).

Stagehand V2

Now updated for Chrome version 132 - Better defaults (like laptop sizing), new features (like realistic screen dimension sizes), and increased stealth (new device data).

[Talk to our team](https://browserbase.com/contact) for access, and trial, or more info!

Advanced Stealth Update

Develop agents that can browse the web with Browserbase and Inngest.

There are two ways to use Browserbase with AgentKit:

1.  Create your own Browserbase tools - If you want to build simple actions on webpages with manual browser control.
    
2.  Use Browserbase's [Stagehand](https://stagehand.dev) library as tools - If you want autonomous browser and resilient scraping.
    

[Get started today](https://agentkit.inngest.com/integrations/browserbase)

New Inngest + Browserbase guide

The v1.14.0 Stagehand release includes performance improvements to the `act()` method using `observe()` under the hood, adds targeted extraction via XPath strings in the `selector` parameter, and introduces support for new AI models like GPT-4.5-preview, Claude 3-7-sonnet-latest, Cerebras LLMs, and Groq LLMs.

[See the full release](https://github.com/browserbase/stagehand/releases/tag/v1.14.0)

Stagehand v1.14 comes with new models & a faster "act"

We've added 8 new tutorials to our playground to make it easier to get started and trial features.

*   "**Enabled AdBlocker**" to test out visiting websites with Ads
    
*   "**Two Tabs**" guide for managing more than one tab
    
*   "**GitHub Sign In**" tutorial to show how to sign in within Browser live view
    
*   "**Training Mode**" playground for generating code
    
*   "**AI** **Codegen**" playground to test out our AI code gen feature
    
*   "**Playwright** **Console**" playground to test our playwright code gen
    
*   "**Advanced** **Tips**" section with feature examples

New playground tutorials

We’re excited to announce our new free plan for side projects & exploration. Users have always been able to test out Browserbase with 10 free browser sessions, but we’re increasing that usage to an hour of session time with 7-day data retention.

Read more ([blog](https://www.browserbase.com/blog/free-plan)) & try it out today ([sign up](https://browserbase.com/sign-up))

Browserbase now has a free plan!

Stagehand version 1.13 has been released with the option to customize your local browser.

Configuration highlights:

*   Chrome executable path (for pointing it at your local chrome)
    
*   Custom user data directory
    
*   Viewport settings
    

The release also made updates to Act() which will automatically attempt to self-heal from a stale observe result and shipped a better developer experience around iframes.

[See the full release here](https://github.com/browserbase/stagehand/pull/469).

Customize your local browser with cookies, viewport, and more

Now supported in [Stagehand](https://Stagehand.dev), [Langchain](https://www.langchain.com/) is a Python and Node.js framework to build applications on top of large-language models (OpenAI, Llama, Gemini).

`npm run langchain` to run our new example ([demo](https://x.com/Stagehanddev/status/1890499688479985886)).

You can also add Browserbase to your Langchain application ([docs](https://docs.browserbase.com/integrations/langchain/introduction)).

Stagehand now supports Langchain

The Browserbase Playground was built to help you quickly run sessions, generate code with AI, and debug Playwright scripts. The new version of the Playground includes a more intuitive user experience, clarified onboarding language, and an easier user interface to explore the full capabilities of browserbase.

Try it out today at [browserbase.com/playground](https://browserbase.com/playground) and reach out to hello@browserbase.com if you have any feedback!

New and improved Playground experience

Lot of new changes for `observe` this week. If you set the `returnAction` flag to `true`, it'll return the suggested Playwright method to run.

Also, Stagehand leverages the accessibility (a11y) tree to improve speed, accuracy, and costs fewer tokens. Using the a11y tree, however, means Stagehand may be able to act on items that aren’t directly visible (i.e. hidden behind a dropdown). To use these new features, set `onlyVisible: false`

```
const results = await page.observe({
     instruction: "click the sign in button",
     returnAction: true, // Return a Playwright action
     onlyVisible: false // Leverage the a11y tree (faster/better/cheaper)
})

// Act on an observation
// NO ADDITIONAL LLM CALL HERE! Just does the thing :)
await page.act(results[0])
```



This returns a list like this, which now contains `method` and `arguments`

```
[
  {
    description: 'The button saying sign in...'
    method: 'click', // or another playwright method
    arguments: [],
    selector: 'xpath=/html/body/div[1]/main/section/div[1]'
  }
]
```

Now, you can cache the selector, method, and arguments if you want to repeat the same Playwright without more LLM calls.

We also made some huge improvements to create-browser-app, making vercel AI SDK + ollama first-class citizens and greatly improving cursorrules.

Stagehand version 1.11 improves speed and accuracy

Normally, sessions begin the shut down process on disconnect. But you can change this behavior so that sessions stay alive till they either timeout or you stop them manually.

Previously, this was reserved for Scale plan customers, but we've opened it up for all plans! This makes it possible for any customer to clone our newest open source project, Open Operator.

Read more about keeping sessions alive in our [documentation](https://docs.browserbase.com/guides/keep-alive#keep-sessions-alive), or view the [open operator project](https://operator.browserbase.com/) & [repository](https://github.com/browserbase/open-operator)

Keep alive works on all plans

Our [List Sessions](https://docs.browserbase.com/reference/api/list-sessions) endpoint has always supported filtering sessions by status, but didn’t give the level of configuration sometimes required. With user metadata, you can now attach arbitrary JSON data to sessions, which then gives you additional flexibility when querying sessions.

Learn more in our documentation: [https://docs.browserbase.com/features/sessions#user-metadata](https://docs.browserbase.com/features/sessions#user-metadata)

You can now attach user metadata to a session

Usage data for current and last billing cycle has now been added to the Browserbase Overview page. You can now see your usage segmented by last 7 days, last 24 hours, or your billing cycle. Data includes total sessions, average session duration, total proxy bandwidth, and average session proxy bandwidth.

More billing and usage data is now on your Overview page

When using the Browserbase playground the code will now mute when sessions are running. There is also a new “clear code” button to reset back to fresh state as well as a share button for copying code or sharing the URL.

New Playground UX Improvements

While Basic Stealth automatically detects and solves Captchas, Advanced Stealth uses more sophisticated techniques to improve performance against common anti-bot systems.

Primarily, Advanced Stealth reduces the changes of being flagged as a bot by using a custom version of the Chrome browser, built and maintained by the Browserbase Stealth Team.

[Read more](https://docs.browserbase.com/features/stealth-mode)

Advanced Stealth is now available for Scale Plan customers

All `stagehand` AI has been moved directly onto the `page` object so now you can convert your Playwright code directly into Stagehand with just **one line changed.**

Just import `@browserbasehq/stagehand` and init the page like below.

 Stagehand Version 1.8.0

Our dashboard overview page just got a makeover! Login to see our quickstart tips & videos on how to get started. 

New overview page

With our multi-region deployment, users can specify their preferred region when connecting. By choosing a server closer to their location, they can significantly reduce the distance each CDP command needs to travel. Read more here: https://www.browserbase.com/blog/becoming-multi-region

New: Multi-region connectivity 

Transform prompts into powerful web automation. Natural language → browser automation is the new standard, and we just published a guide showing how to automate agents for any website.

AI codegen guide is out now

Multi-region browser support: You can now run sessions locally in your region, offering major performance gains. To learn more, check out our [multi-region guide](https://docs.browserbase.com/guides/multi-region).

You can now run sessions locally in your region

We’ve recreated our [Python](https://github.com/browserbase/sdk-python) and [Node](https://github.com/browserbase/sdk-node/) SDKs from scratch, offering greater capability, higher performance and generally improved quality.

New Browserbase SDKs

We’ve released an open-source AI web browsing framework that offers simple APIs (act, extract, and observe) to enable natural language-driven web automation. Built on top of Playwright, Stagehand focuses on atomic, reliable actions and supports multiple AI models. To learn more, check out the [Stagehand repo](https://github.com/browserbase/stagehand).

Announcing Stagehand: An open-source AI web browsing framework

**HIPAA Compliance.** The Browserbase platform is now HIPAA compliant.

Browserbase is now HIPAA compliant

We’re excited to announce that Browserbase is now certified SOC 2 Type 1 compliant.

Browserbase is now SOC 2 Type 1 compliant

[Performance Tuning](https://docs.browserbase.com/guides/performance) your Browserbase app. This guide offers recommendations and approaches for optimizing your app to get the best performance out of Browserbase.

New guide on performance tuning 

Introducing our new Browserbase operational status board:

[https://status.browserbase.com/](https://status.browserbase.com/)

Use it to get a real-time picture of platform health and to track any maintenance or incidents impacting our services. Subscribe to it to receive incident notifications.

New operational status board

You can now disable session recording by setting `browserSettings.recordSession` to `false` in the create session API.

New: Optional session recording

[Building your own AI Codegen](https://www.notion.so/browserbase/How-to-build-your-own-AI-Codegen-0f1d1ddb08f740d4b8ef6d242bf0aa08?pvs=4) - A walkthrough on how we’ve built the popular codegen feature in our playground [(demo)](https://x.com/alexdphan/status/1821618745191899304).

New walkthrough on how we’ve built the popular codegen feature in our playground

You can now record your actions in the playground with playground training mode, making your dev experience much easier.

You can now record your actions in the playground

When using Selenium, it’s no longer necessary to specify the debugger address in WebDriver requests.

No longer necessary to specify the debugger address in WebDriver requests

The `api.browserbase.com` endpoint has been deprecated. Please use `connect.browserbase.com` instead.

Api.browserbase.com endpoint has been deprecated

Sessions can now be configured to block ads and to enable or disable captcha solving.



Sessions can now be configured to block ads

Sessions can now be configured to use custom proxies which can be combined with Browserbase proxies. Use domain filters to conditionally enable them. You can now also specify the geographic location for Browserbase proxies.

Sessions can now be configured to use custom proxies

Session Live URLs now returns all pages in the browser. Playground uses this to now support multiple tabs.

Browserbase

Stagehand

Director

MCP

Browserbase Changelog