{"id":3590,"date":"2026-06-26T13:31:04","date_gmt":"2026-06-26T05:31:04","guid":{"rendered":"https:\/\/ai.wuaishare.cn\/hub\/?p=3590"},"modified":"2026-06-27T02:47:04","modified_gmt":"2026-06-26T18:47:04","slug":"browser-use-skill","status":"publish","type":"post","link":"https:\/\/ai.wuaishare.cn\/hub\/en\/3590","title":{"rendered":"Browser Use"},"content":{"rendered":"\"Shows\n<\/picture>\n
\n \"The\n <\/picture>\n<\/div>\n
\n\"Browser-Use<\/a>\n<\/div>\n

—<\/p>\n

\n\"Demos\"<\/a>
\n\"\"
\n\"Docs\"<\/a>
\n\"\"
\n\"Blog\"<\/a>
\n\"\"
\n\"Merch\"<\/a>
\n\"\"
\n\"Github<\/a>
\n\"\"
\n\"Twitter\"<\/a>
\n\"\"
\n\"Discord\"<\/a>
\n\"\"
\n\"Browser-Use<\/a>\n<\/div>\n

<\/br><\/p>\n

\ud83c\udf24\ufe0f Want to skip the setup? Use our [cloud](https:\/\/cloud.browser-use.com?utm_source=github&utm_medium=readme-skip-setup)<\/b> for faster, scalable, stealth-enabled browser automation!<\/p>\n

# \ud83e\udd16 LLM Quickstart<\/p>\n

1. Direct your favorite coding agent (Cursor, Claude Code, etc) to [Agents.md](https:\/\/docs.browser-use.com\/llms-full.txt)
\n2. Prompt away!<\/p>\n

<\/p>\n

# \ud83d\udc4b Human Quickstart<\/p>\n

Browser Use 0.13 introduces a new beta agent powered by a Rust core and a browser harness built for current frontier models. It gives the model a real browser\/computer action space, persistent tools, and recovery loops inspired by coding agents.<\/p>\n

“`text
\nPython API -> Rust core -> Browser harness -> Web task done
\n“`<\/p>\n

**1. Install Browser Use with the native core runtime (Python>=3.11):**
\n“`bash
\nuv add “browser-use[core]”
\n# or: pip install “browser-use[core]”
\nbrowser
\n“`<\/p>\n

The `[core]` extra installs the native Browser Use runtime for your platform.<\/p>\n

**2. [Optional] Get your API key from [Browser Use Cloud](https:\/\/cloud.browser-use.com\/new-api-key?utm_source=github&utm_medium=readme-quickstart-api-key):**
\n“`
\n# .env
\nBROWSER_USE_API_KEY=your-key
\n# GOOGLE_API_KEY=your-key
\n# ANTHROPIC_API_KEY=your-key
\n“`<\/p>\n

**3. Run your first agent:**<\/p>\n

**Browser Use Terminal:**
\n“`bash
\nuv add “browser-use[core]”
\nbrowser
\n“`<\/p>\n

**Python Script:**
\n“`python
\nfrom browser_use.beta import Agent, BrowserProfile, ChatBrowserUse
\n# from browser_use.beta import ChatOpenAI # ChatOpenAI(model=’gpt-5.5′)
\n# from browser_use.beta import ChatAnthropic # ChatAnthropic(model=’claude-opus-4-8′)
\nimport asyncio<\/p>\n

async def main():
\n agent = Agent(
\n task=”Find the number of stars of the browser-use repo”,
\n llm=ChatBrowserUse(model=’openai\/gpt-5.5′),
\n # llm=ChatBrowserUse(model=’bu-2-0′), # Browser Use’s own optimized model
\n # llm=ChatOpenAI(model=’gpt-5.5′),
\n # llm=ChatAnthropic(model=’claude-opus-4-8′), # Sonnet also works well.
\n browser_profile=BrowserProfile(
\n headless=False,
\n allowed_domains=[“*.github.com”],
\n ),
\n )
\n history = await agent.run()
\n print(history.final_result())<\/p>\n

if __name__ == “__main__”:
\n asyncio.run(main())
\n“`<\/p>\n

Existing Python agent users can keep using `from browser_use import Agent`. The new Rust-powered beta agent is `from browser_use.beta import Agent`.<\/p>\n

Check out the [library docs](https:\/\/docs.browser-use.com\/open-source\/introduction) and the [cloud docs](https:\/\/docs.cloud.browser-use.com?utm_source=github&utm_medium=readme-cloud-docs) for more!<\/p>\n

<\/p>\n

# Open Source vs Cloud<\/p>\n\"BU\n<\/picture>\n

We benchmark Browser Use across 100 real-world browser tasks. Full benchmark is open source: **[browser-use\/benchmark](https:\/\/github.com\/browser-use\/benchmark)**.<\/p>\n

**Use the Open-Source Agent**
\n– You need [custom tools](https:\/\/docs.browser-use.com\/customize\/tools\/basics) or deep code-level integration
\n– We recommend pairing with our [cloud browsers](https:\/\/docs.browser-use.com\/open-source\/customize\/browser\/remote) for leading stealth, proxy rotation, and scaling
\n– Or self-host the open-source agent fully on your own machines<\/p>\n

**Use the [Fully-Hosted Cloud Agent](https:\/\/cloud.browser-use.com?utm_source=github&utm_medium=readme-hosted-agent) (recommended)**
\n– Much more powerful agent for complex tasks (see plot above)
\n– Easiest way to start and scale
\n– Best stealth with proxy rotation and captcha solving
\n– 1000+ integrations (Gmail, Slack, Notion, and more)
\n– Persistent filesystem and memory<\/p>\n

<\/p>\n

# Demos<\/p>\n

### \ud83d\udccb Form-Filling
\n#### Task = “Fill in this job application with my resume and information.”
\n![Job Application Demo](https:\/\/github.com\/user-attachments\/assets\/57865ee6-6004-49d5-b2c2-6dff39ec2ba9)
\n[Example code \u2197](https:\/\/github.com\/browser-use\/browser-use\/blob\/main\/examples\/use-cases\/apply_to_job.py)<\/p>\n

### \ud83c\udf4e Grocery-Shopping
\n#### Task = “Put this list of items into my instacart.”<\/p>\n

https:\/\/github.com\/user-attachments\/assets\/a6813fa7-4a7c-40a6-b4aa-382bf88b1850<\/p>\n

[Example code \u2197](https:\/\/github.com\/browser-use\/browser-use\/blob\/main\/examples\/use-cases\/buy_groceries.py)<\/p>\n

### \ud83d\udcbb Personal-Assistant.
\n#### Task = “Help me find parts for a custom PC.”<\/p>\n

https:\/\/github.com\/user-attachments\/assets\/ac34f75c-057a-43ef-ad06-5b2c9d42bf06<\/p>\n

[Example code \u2197](https:\/\/github.com\/browser-use\/browser-use\/blob\/main\/examples\/use-cases\/pcpartpicker.py)<\/p>\n

### \ud83d\udca1See [more examples here \u2197](https:\/\/docs.browser-use.com\/examples) and give us a star!<\/p>\n

<\/p>\n

# \ud83d\ude80 Template Quickstart<\/p>\n

**Want to get started even faster?** Generate a ready-to-run template:<\/p>\n

“`bash
\nuvx browser-use init –template default
\n“`<\/p>\n

This creates a `browser_use_default.py` file with a working example. Available templates:
\n– `default` – Minimal setup to get started quickly
\n– `advanced` – All configuration options with detailed comments
\n– `tools` – Examples of custom tools and extending the agent<\/p>\n

You can also specify a custom output path:
\n“`bash
\nuvx browser-use init –template default –output my_agent.py
\n“`<\/p>\n

<\/p>\n

# \ud83d\udcbb CLI<\/p>\n

Fast, persistent browser automation from the command line:<\/p>\n

“`bash
\nbrowser-use open https:\/\/example.com # Navigate to URL
\nbrowser-use state # See clickable elements
\nbrowser-use click 5 # Click element by index
\nbrowser-use type “Hello” # Type text
\nbrowser-use screenshot page.png # Take screenshot
\nbrowser-use close # Close browser
\n“`<\/p>\n

The CLI keeps the browser running between commands for fast iteration. See [CLI docs](browser_use\/skill_cli\/README.md) for all commands.<\/p>\n

### Claude Code Skill<\/p>\n

For [Claude Code](https:\/\/claude.ai\/code), install the skill to enable AI-assisted browser automation:<\/p>\n

“`bash
\nmkdir -p ~\/.claude\/skills\/browser-use
\ncurl -o ~\/.claude\/skills\/browser-use\/SKILL.md \\
\n https:\/\/raw.githubusercontent.com\/browser-use\/browser-use\/main\/skills\/browser-use\/SKILL.md
\n“`<\/p>\n

<\/p>\n

## Integrations, hosting, custom tools, MCP, and more on our [Docs \u2197](https:\/\/docs.browser-use.com)<\/p>\n

<\/p>\n

# FAQ<\/p>\n

\nWhat’s the best model to use?<\/b><\/summary>\n

We optimized **ChatBrowserUse()** specifically for browser automation tasks. On avg it completes tasks 3-5x faster than other models with SOTA accuracy.<\/p>\n

For pricing and other LLM providers, see our [supported models documentation](https:\/\/docs.browser-use.com\/supported-models).
\n<\/details>\n

\nCan I use Claude \/ GPT \/ Gemini through ChatBrowserUse?<\/b><\/summary>\n

Yes. `ChatBrowserUse` accepts provider-prefixed model ids, so a single `BROWSER_USE_API_KEY` reaches all of them \u2014 no separate OpenAI\/Anthropic\/Google keys required:<\/p>\n

“`python
\nfrom browser_use import Agent, ChatBrowserUse<\/p>\n

llm = ChatBrowserUse(model=’anthropic\/claude-sonnet-4-6′) # or ‘openai\/gpt-5.5’, ‘google\/gemini-3-pro’
\nagent = Agent(task=’…’, llm=llm)
\n“`<\/p>\n

For the best speed and cost we still recommend the default `bu-*` models.
\n<\/details>\n

\nShould I use the Browser Use system prompt with the open-source preview model?<\/b><\/summary>\n

Yes. If you use `ChatBrowserUse(model=’browser-use\/bu-30b-a3b-preview’)` with a normal `Agent(…)`, Browser Use still sends its default agent system prompt for you.<\/p>\n

You do **not** need to add a separate custom “Browser Use system message” just because you switched to the open-source preview model. Only use `extend_system_message` or `override_system_message` when you intentionally want to customize the default behavior for your task.<\/p>\n

If you want the best default speed\/accuracy, we still recommend the newer hosted `bu-*` models. If you want the open-source preview model, the setup stays the same apart from the `model=` value.
\n<\/details>\n

\nCan I use custom tools with the agent?<\/b><\/summary>\n

Yes! You can add custom tools to extend the agent’s capabilities:<\/p>\n

“`python
\nfrom browser_use import Tools<\/p>\n

tools = Tools()<\/p>\n

@tools.action(description=’Description of what this tool does.’)
\ndef custom_tool(param: str) -> str:
\n return f”Result: {param}”<\/p>\n

agent = Agent(
\n task=”Your task”,
\n llm=llm,
\n browser=browser,
\n tools=tools,
\n)
\n“`<\/p>\n<\/details>\n

\nCan I use this for free?<\/b><\/summary>\n

Yes! Browser-Use is open source and free to use. You only need to choose an LLM provider (like OpenAI, Google, ChatBrowserUse, or run local models with Ollama).
\n<\/details>\n

\nTerms of Service<\/b><\/summary>\n

This open-source library is licensed under the MIT License. For Browser Use services & data policy, see our [Terms of Service](https:\/\/browser-use.com\/legal\/terms-of-service) and [Privacy Policy](https:\/\/browser-use.com\/privacy\/).
\n<\/details>\n

\nHow do I handle authentication?<\/b><\/summary>\n

Check out our authentication examples:
\n– [Using real browser profiles](https:\/\/github.com\/browser-use\/browser-use\/blob\/main\/examples\/browser\/real_browser.py) – Reuse your existing Chrome profile with saved logins
\n– If you want to use temporary accounts with inbox, choose AgentMail
\n– To sync your auth profile with the remote browser, run `curl -fsSL https:\/\/browser-use.com\/profile.sh | BROWSER_USE_API_KEY=XXXX sh` (replace XXXX with your API key)<\/p>\n

These examples show how to maintain sessions and handle authentication seamlessly.
\n<\/details>\n

\nHow do I solve CAPTCHAs?<\/b><\/summary>\n

For CAPTCHA handling, you need better browser fingerprinting and proxies. Use [Browser Use Cloud](https:\/\/cloud.browser-use.com?utm_source=github&utm_medium=readme-faq-captcha) which provides stealth browsers designed to avoid detection and CAPTCHA challenges.
\n<\/details>\n

\nHow do I go into production?<\/b><\/summary>\n

Chrome can consume a lot of memory, and running many agents in parallel can be tricky to manage.<\/p>\n

For production use cases, use our [Browser Use Cloud API](https:\/\/cloud.browser-use.com?utm_source=github&utm_medium=readme-faq-production) which handles:
\n– Scalable browser infrastructure
\n– Memory management
\n– Proxy rotation
\n– Stealth browser fingerprinting
\n– High-performance parallel execution
\n<\/details>\n

<\/p>\n

\n

**Tell your computer what to do, and it gets it done.**<\/p>\n

<\/p>\n

[![Twitter Follow](https:\/\/img.shields.io\/twitter\/follow\/Magnus?style=social)](https:\/\/x.com\/intent\/user?screen_name=mamagnus00)
\n   
\n[![Twitter Follow](https:\/\/img.shields.io\/twitter\/follow\/Gregor?style=social)](https:\/\/x.com\/intent\/user?screen_name=gregpr07)<\/p>\n<\/div>\n

Made with \u2764\ufe0f in Zurich and San Francisco <\/div>\n","protected":false},"excerpt":{"rendered":"

Open-source browser automation for AI agents, with a Rust-powered core, browser harness and Python API.<\/p>\n","protected":false},"author":1,"featured_media":3582,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"ai_share_source_url":"https:\/\/github.com\/browser-use\/browser-use","ai_share_source_title":"Browser Use","ai_share_status":"\u5df2\u53d1\u5e03","ai_share_official_url":"https:\/\/browser-use.com","ai_share_original_content_language":"en","ai_share_i18n_original_source_language":"en","ai_share_author_label":"browser-use","ai_share_version_label":"0.13.2","ai_share_install_command":"uv add \"browser-use[core]\"","ai_share_primary_category":"\u5de5\u4f5c\u6d41\u81ea\u52a8\u5316","ai_share_tool_type":"skill","ai_share_resource_type":"workflow_skill_pack","ai_share_platform_support":"Python,CLI","ai_share_collection":"LobeHub Skills","ai_share_last_verified_at":"2026-05-07","ai_share_last_commit_at":"2026-06-20T18:21:29Z","ai_share_license":"MIT","ai_share_language_support":"en","ai_share_target_user":"\u5f00\u53d1\u8005,\u81ea\u52a8\u5316\u7528\u6237","ai_share_github_topics":"ai-agents, ai-tools, browser-automation, browser-use, llm, playwright, python","ai_share_repo_default_branch":"main","ai_share_pricing_model":"open_source","ai_share_deployment_mode":"self_hosted","ai_share_delivery_mode":"cli","ai_share_region_availability":"global","ai_share_source_platform":"GitHub","ai_share_release_url":"https:\/\/github.com\/browser-use\/browser-use\/releases\/tag\/0.13.2","ai_share_github_releases_json":"[{\"tagName\":\"0.13.2\",\"name\":\"0.13.2\",\"url\":\"https:\/\/github.com\/browser-use\/browser-use\/releases\/tag\/0.13.2\",\"body\":\"## What's Changed\\n* [codex] add BU3 browser use models by @MagMueller in https:\/\/github.com\/browser-use\/browser-use\/pull\/5002\\n* [ENG-4762] publish: gate publish_to_pypi behind release env by @LarsenCundric in https:\/\/github.com\/browser-use\/browser-use\/pull\/5013\\n* feat(llm): accept provider-prefixed models in ChatBrowserUse by @sauravpanda in https:\/\/github.com\/browser-use\/browser-use\/pull\/5007\\n* chore(llm): remove bu-3 \/ bu-3-max, default beta example to openai\/gpt-5.5 by @sauravpanda in https:\/\/github.com\/browser-use\/browser-use\/pull\/5020\\n* Bump version from 0.13.1 to 0.13.2 by @MagMueller in https:\/\/github.com\/browser-use\/browser-use\/pull\/5026\\n* Bump browser-use-core to 0.13.2 by @laithrw in https:\/\/github.com\/browser-use\/browser-use\/pull\/5027\\n\\n\\n**Full Changelog**: https:\/\/github.com\/browser-use\/browser-use\/compare\/0.13.1...0.13.2\",\"bodyExcerpt\":\"## What's Changed * [codex] add BU3 browser use models by @MagMueller in https:\/\/github.com\/browser-use\/browser-use\/pull\/5002 * [ENG-4762] publish: gate publish_to_pypi behind release env by @LarsenCundric in https:\/\/github.com\/browser-use\/browser-use\/pull\/5013 * feat(llm): accept provider-prefixed models in ChatBrowserUse by @sauravpanda in https:\/\/github.com\/browser-use\/browser-use\/pull\/5007 * chore(llm): remove bu-3 \/ bu-3-max, default beta example to openai\/gpt-5.5 by @sauravpanda in https:\/\/github.com\/browser-\",\"bodyTruncated\":false,\"publishedAt\":\"2026-06-12T22:45:04Z\",\"prerelease\":false,\"draft\":false,\"assetCount\":0,\"assets\":[]},{\"tagName\":\"0.13.1\",\"name\":\"0.13.1\",\"url\":\"https:\/\/github.com\/browser-use\/browser-use\/releases\/tag\/0.13.1\",\"body\":\"## What's Changed\\n* support claude fable 5 by @laithrw in https:\/\/github.com\/browser-use\/browser-use\/pull\/5008\\n* anthropic: tool choice auto when thinking (cannot force it) by @laithrw in https:\/\/github.com\/browser-use\/browser-use\/pull\/5009\\n* Fix packaged agent tools env by @gregpr07 in https:\/\/github.com\/browser-use\/browser-use\/pull\/5010\\n\\n\\n**Full Changelog**: https:\/\/github.com\/browser-use\/browser-use\/compare\/0.13.0...0.13.1\",\"bodyExcerpt\":\"## What's Changed * support claude fable 5 by @laithrw in https:\/\/github.com\/browser-use\/browser-use\/pull\/5008 * anthropic: tool choice auto when thinking (cannot force it) by @laithrw in https:\/\/github.com\/browser-use\/browser-use\/pull\/5009 * Fix packaged agent tools env by @gregpr07 in https:\/\/github.com\/browser-use\/browser-use\/pull\/5010 **Full Changelog**: https:\/\/github.com\/browser-use\/browser-use\/compare\/0.13.0...0.13.1\",\"bodyTruncated\":false,\"publishedAt\":\"2026-06-10T01:00:59Z\",\"prerelease\":false,\"draft\":false,\"assetCount\":0,\"assets\":[]},{\"tagName\":\"0.13.0\",\"name\":\"0.13.0 - Rebuilt in Rust [beta]\",\"url\":\"https:\/\/github.com\/browser-use\/browser-use\/releases\/tag\/0.13.0\",\"body\":\"\\\"browser-use_a_hooded_traveler_in_saffron_robes_with_a_long_staf_99dc26d6-efe0-4f9b-b2ea-7329f26cdb2a\\n\\n\\n\\nBrowser Use 0.13.0 introduces a new Rust-backed beta agent.\\n\\nIt gives modern models a more direct browser control loop, guided by robust helpers instead of brittle browser abstractions.\\n\\n```bash\\nuv add \\\"browser-use[core]\\\"\\n```\\n\\n```python\\nfrom browser_use.beta import Agent\\n```\\n\\nThe existing Python agent remains unchanged.\\n\\n## What's Changed\\n* add x402 skill by @laithrw in https:\/\/github.com\/browser-use\/browser-use\/pull\/4782\\n* fix by @laithrw in https:\/\/github.com\/browser-use\/browser-use\/pull\/4950\\n* Magnus\/rust sdk server clean pr by @MagMueller in https:\/\/github.com\/browser-use\/browser-use\/pull\/4975\\n* Validate Rust terminal SDK server binary by @laithrw in https:\/\/github.com\/browser-use\/browser-use\/pull\/4996\\n* Add browser-use-core binary discovery by @gregpr07 in https:\/\/github.com\/browser-use\/browser-use\/pull\/4994\\n* Fix prerelease version upgrade check by @gregpr07 in https:\/\/github.com\/browser-use\/browser-use\/pull\/4998\\n* Prepare browser-use rc4 by @gregpr07 in https:\/\/github.com\/browser-use\/browser-use\/pull\/5003\\n* Close beta SDK browser resources by @gregpr07 in https:\/\/github.com\/browser-use\/browser-use\/pull\/5004\\n* Prepare browser-use 0.13.0 by @gregpr07 in https:\/\/github.com\/browser-use\/browser-use\/pull\/5005\\n\\n\\n**Full Changelog**: https:\/\/github.com\/browser-use\/browser-use\/compare\/0.12.9...0.13.0\",\"bodyExcerpt\":\"Browser Use 0.13.0 introduces a new Rust-backed beta agent. It gives modern models a more direct browser control loop, guided by robust helpers instead of brittle browser abstractions. ```bash uv add \\\"browser-use[core]\\\" ``` ```python from browser_use.beta import Agent ``` The existing Python agent remains unchanged. ## What's Changed * add x402 skill by @laithrw in https:\/\/github.com\/browser-use\/browser-use\/pull\/4782 * fix by @laithrw in https:\/\/github.com\/browser-use\/browser-use\/pull\/4950 * Magnus\/rust sdk server\",\"bodyTruncated\":false,\"publishedAt\":\"2026-06-08T22:07:47Z\",\"prerelease\":false,\"draft\":false,\"assetCount\":0,\"assets\":[]},{\"tagName\":\"0.12.9\",\"name\":\"0.12.9\",\"url\":\"https:\/\/github.com\/browser-use\/browser-use\/releases\/tag\/0.12.9\",\"body\":\"## What's Changed\\n* Pass session id to judge LLM calls by @MagMueller in https:\/\/github.com\/browser-use\/browser-use\/pull\/4918\\n* fix(agent): skip screenshots on new tab pages by @MagMueller in https:\/\/github.com\/browser-use\/browser-use\/pull\/4920\\n\\n\\n**Full Changelog**: https:\/\/github.com\/browser-use\/browser-use\/compare\/0.12.8...0.12.9\",\"bodyExcerpt\":\"## What's Changed * Pass session id to judge LLM calls by @MagMueller in https:\/\/github.com\/browser-use\/browser-use\/pull\/4918 * fix(agent): skip screenshots on new tab pages by @MagMueller in https:\/\/github.com\/browser-use\/browser-use\/pull\/4920 **Full Changelog**: https:\/\/github.com\/browser-use\/browser-use\/compare\/0.12.8...0.12.9\",\"bodyTruncated\":false,\"publishedAt\":\"2026-05-26T03:45:22Z\",\"prerelease\":false,\"draft\":false,\"assetCount\":0,\"assets\":[]},{\"tagName\":\"0.12.8\",\"name\":\"0.12.8\",\"url\":\"https:\/\/github.com\/browser-use\/browser-use\/releases\/tag\/0.12.8\",\"body\":\"## What's Changed\\n* fix(daemon): restrict unix socket file to owner-only access by @sauravpanda in https:\/\/github.com\/browser-use\/browser-use\/pull\/4870\\n* docs: clarify integration example placement by @felix-windsor in https:\/\/github.com\/browser-use\/browser-use\/pull\/4856\\n* fix(tools): refuse evaluate() on restricted browser profiles by @sauravpanda in https:\/\/github.com\/browser-use\/browser-use\/pull\/4871\\n* chore(llm): default ChatBrowserUse to bu-2-0 by @sauravpanda in https:\/\/github.com\/browser-use\/browser-use\/pull\/4876\\n* feat: add client header to GoogleChat by @markmcd in https:\/\/github.com\/browser-use\/browser-use\/pull\/4884\\n* chore(llm): recommend gemini-3-flash-preview in examples and tests by @sauravpanda in https:\/\/github.com\/browser-use\/browser-use\/pull\/4885\\n* Add OpenRouter pricing fallback for token costs by @sauravpanda in https:\/\/github.com\/browser-use\/browser-use\/pull\/4886\\n* agent(history): freeze HistoryItem + lock byte-prefix property by @sauravpanda in https:\/\/github.com\/browser-use\/browser-use\/pull\/4890\\n* agent(prompts): move per-step metadata out of into a tail block by @sauravpanda in https:\/\/github.com\/browser-use\/browser-use\/pull\/4891\\n* feat: forward cached_content into generate_content by @sauravpanda in https:\/\/github.com\/browser-use\/browser-use\/pull\/4889\\n* Move user request before agent history by @MagMueller in https:\/\/github.com\/browser-use\/browser-use\/pull\/4897\\n* Bump version to 0.12.8 by @MagMueller in https:\/\/github.com\/browser-use\/browser-use\/pull\/4899\\n* Revert ChatGoogle cached_content forwarding by @MagMueller in https:\/\/github.com\/browser-use\/browser-use\/pull\/4900\\n* Revert evaluate restriction on restricted profiles by @MagMueller in https:\/\/github.com\/browser-use\/browser-use\/pull\/4901\\n* Trim HistoryItem freeze follow-up by @MagMueller in https:\/\/github.com\/browser-use\/browser-use\/pull\/4902\\n\\n## New Contributors\\n* @felix-windsor made their first contribution in https:\/\/github.com\/browser-use\/browser-use\/pull\/4856\\n\\n**Full Changelog**: https:\/\/github.com\/browser-use\/browser-use\/compare\/0.12.7...0.12.8\",\"bodyExcerpt\":\"## What's Changed * fix(daemon): restrict unix socket file to owner-only access by @sauravpanda in https:\/\/github.com\/browser-use\/browser-use\/pull\/4870 * docs: clarify integration example placement by @felix-windsor in https:\/\/github.com\/browser-use\/browser-use\/pull\/4856 * fix(tools): refuse evaluate() on restricted browser profiles by @sauravpanda in https:\/\/github.com\/browser-use\/browser-use\/pull\/4871 * chore(llm): default ChatBrowserUse to bu-2-0 by @sauravpanda in https:\/\/github.com\/browser-use\/browser-use\/pull\",\"bodyTruncated\":false,\"publishedAt\":\"2026-05-23T19:29:53Z\",\"prerelease\":false,\"draft\":false,\"assetCount\":0,\"assets\":[]},{\"tagName\":\"0.12.7\",\"name\":\"0.12.7\",\"url\":\"https:\/\/github.com\/browser-use\/browser-use\/releases\/tag\/0.12.7\",\"body\":\"## What's Changed\\n* another big cli update by @ShawnPana in https:\/\/github.com\/browser-use\/browser-use\/pull\/4514\\n* fix: security and correctness issues found in #4514 review by @sauravpanda in https:\/\/github.com\/browser-use\/browser-use\/pull\/4590\\n* fix: upgrade aiohttp to 3.13.4 to patch memory exhaustion vulnerability by @sauravpanda in https:\/\/github.com\/browser-use\/browser-use\/pull\/4596\\n* fix: upgrade requests to 2.33.0 to patch temp-file path-traversal vulnerability by @sauravpanda in https:\/\/github.com\/browser-use\/browser-use\/pull\/4597\\n* fix: add per-session auth token to daemon socket by @sauravpanda in https:\/\/github.com\/browser-use\/browser-use\/pull\/4598\\n* improve model docs by @laithrw in https:\/\/github.com\/browser-use\/browser-use\/pull\/4616\\n* fix: prevent KeyError in load_from_dict and IndexError in final_result by @r266-tech in https:\/\/github.com\/browser-use\/browser-use\/pull\/4464\\n* fix: use setattr for LLM ainvoke patching to avoid pydantic crash by @laithrw in https:\/\/github.com\/browser-use\/browser-use\/pull\/4641\\n* fix: guard against missing stdin in MCP stdio server startup by @laithrw in https:\/\/github.com\/browser-use\/browser-use\/pull\/4642\\n* chore: update browser-use-sdk from 2.0.15 to 3.4.2 by @sauravpanda in https:\/\/github.com\/browser-use\/browser-use\/pull\/4644\\n* Improve OSS-to-cloud conversion: UTM tracking, better error messages, and cloud nudges by @Alezander9 in https:\/\/github.com\/browser-use\/browser-use\/pull\/4646\\n* Add per-link utm_medium slugs to README cloud links by @Alezander9 in https:\/\/github.com\/browser-use\/browser-use\/pull\/4653\\n* fix(#4631): clear dom cache after scroll to prevent stale extract data by @laithrw in https:\/\/github.com\/browser-use\/browser-use\/pull\/4658\\n* fix asyncio.get_event_loop for python 3.14 compat by @laithrw in https:\/\/github.com\/browser-use\/browser-use\/pull\/4659\\n* fix sensitive_data redaction order to prevent substring leaks by @laithrw in https:\/\/github.com\/browser-use\/browser-use\/pull\/4660\\n* fix pagination classifier to prioritize semantic labels over glyph syms by @laithrw in https:\/\/github.com\/browser-use\/browser-use\/pull\/4661\\n* fix: handle BrokenPipeError gracefully when MCP client disconnects by @laithrw in https:\/\/github.com\/browser-use\/browser-use\/pull\/4663\\n* prefer Playwright chromium over sys Chrome by default by @laithrw in https:\/\/github.com\/browser-use\/browser-use\/pull\/4664\\n* close alias for BrowserSession stop by @laithrw in https:\/\/github.com\/browser-use\/browser-use\/pull\/4665\\n* ci: pin stale workflow action by @grtninja in https:\/\/github.com\/browser-use\/browser-use\/pull\/4656\\n* improve connect failure UX: fix chrome:\/\/inspect link and add fallbac\u2026 by @ShawnPana in https:\/\/github.com\/browser-use\/browser-use\/pull\/4678\\n* fix: add utf-8 encoding to Local State file read in list_chrome_profiles by @voidborne-d in https:\/\/github.com\/browser-use\/browser-use\/pull\/4690\\n* fix input clear fallbacks and clarify clear-then-type behavior by @laithrw in https:\/\/github.com\/browser-use\/browser-use\/pull\/4692\\n* Fix: Catch TypeError during lmnr import to prevent CLI startup crash (#4046) by @AtharvaJaiswal005 in https:\/\/github.com\/browser-use\/browser-use\/pull\/4104\\n* Add `record start\/stop` CLI command for session video capture by @sauravpanda in https:\/\/github.com\/browser-use\/browser-use\/pull\/4710\\n* fix(tools): enforce per-action timeout to prevent hung event handlers by @sauravpanda in https:\/\/github.com\/browser-use\/browser-use\/pull\/4711\\n* Bump pillow to 12.2.0 (CVE-2026-40192) by @sauravpanda in https:\/\/github.com\/browser-use\/browser-use\/pull\/4713\\n* Bump pypdf to 6.10.2 (CVE-2026-40260) by @sauravpanda in https:\/\/github.com\/browser-use\/browser-use\/pull\/4714\\n* Bump aiohttp to 3.13.4 (CVE-2026-34515) by @sauravpanda in https:\/\/github.com\/browser-use\/browser-use\/pull\/4715\\n* docs: fix malformed HTML attributes in README picture\/img tags by @abhicris in https:\/\/github.com\/browser-use\/browser-use\/pull\/4712\\n* fix(element): redact sensitive values in fill() debug\",\"bodyExcerpt\":\"## What's Changed * another big cli update by @ShawnPana in https:\/\/github.com\/browser-use\/browser-use\/pull\/4514 * fix: security and correctness issues found in #4514 review by @sauravpanda in https:\/\/github.com\/browser-use\/browser-use\/pull\/4590 * fix: upgrade aiohttp to 3.13.4 to patch memory exhaustion vulnerability by @sauravpanda in https:\/\/github.com\/browser-use\/browser-use\/pull\/4596 * fix: upgrade requests to 2.33.0 to patch temp-file path-traversal vulnerability by @sauravpanda in https:\/\/github.com\/browser-\",\"bodyTruncated\":true,\"publishedAt\":\"2026-05-19T01:11:20Z\",\"prerelease\":false,\"draft\":false,\"assetCount\":0,\"assets\":[]},{\"tagName\":\"0.12.6\",\"name\":\"0.12.6\",\"url\":\"https:\/\/github.com\/browser-use\/browser-use\/releases\/tag\/0.12.6\",\"body\":\"## What's Changed\\n\\\"image\\\"\\n\\n* fix: set default temperature to 1.0 for gemini-3 models by @markmcd in https:\/\/github.com\/browser-use\/browser-use\/pull\/4489\\n* fix bedrock structured output w SchemaOptimizer to flatten schema by @laithrw in https:\/\/github.com\/browser-use\/browser-use\/pull\/4524\\n* fix: use login shell for PATH configuration in install script by @techiro in https:\/\/github.com\/browser-use\/browser-use\/pull\/4457\\n* fix(utils): add option to disable SignalHandler for host app signal control by @BillionClaw in https:\/\/github.com\/browser-use\/browser-use\/pull\/4387\\n* rm pass by @laithrw in https:\/\/github.com\/browser-use\/browser-use\/pull\/4525\\n* fix daemon orphaning when external browser dies on --connect\/--cdp-url by @laithrw in https:\/\/github.com\/browser-use\/browser-use\/pull\/4526\\n* Feature\/cdp navigation timeout by @laithrw in https:\/\/github.com\/browser-use\/browser-use\/pull\/4527\\n* feat(tokens): allow overriding model pricing URL by @laithrw in https:\/\/github.com\/browser-use\/browser-use\/pull\/4528\\n* fix anthropic action field double-serialization by @laithrw in https:\/\/github.com\/browser-use\/browser-use\/pull\/4529\\n* fix(agent): prevent stale history and stuck step counter on timeout by @warren618 in https:\/\/github.com\/browser-use\/browser-use\/pull\/4481\\n* fix tunnel process kill and daemon spawn on windows by @laithrw in https:\/\/github.com\/browser-use\/browser-use\/pull\/4530\\n* fix: stop cloud sessions when reconnecting via cdp_url by @iafisch in https:\/\/github.com\/browser-use\/browser-use\/pull\/4096\\n* fix(mcp): remove oneOf from browser_click schema that breaks Claude API clients by @gabros20 in https:\/\/github.com\/browser-use\/browser-use\/pull\/4212\\n* feat(cli): add install-lite.sh script for the CLI by @ShawnPana in https:\/\/github.com\/browser-use\/browser-use\/pull\/4534\\n* fix: close httpx client pool on session stop to prevent OOM on Lambda by @LarsenCundric in https:\/\/github.com\/browser-use\/browser-use\/pull\/4577\\n* perf: fix O(n\u00b2) bottlenecks in DOM capture for heavy pages by @MagMueller in https:\/\/github.com\/browser-use\/browser-use\/pull\/4587\\n\\n## New Contributors\\n* @markmcd made their first contribution in https:\/\/github.com\/browser-use\/browser-use\/pull\/4489\\n* @techiro made their first contribution in https:\/\/github.com\/browser-use\/browser-use\/pull\/4457\\n* @BillionClaw made their first contribution in https:\/\/github.com\/browser-use\/browser-use\/pull\/4387\\n* @warren618 made their first contribution in https:\/\/github.com\/browser-use\/browser-use\/pull\/4481\\n* @iafisch made their first contribution in https:\/\/github.com\/browser-use\/browser-use\/pull\/4096\\n* @gabros20 made their first contribution in https:\/\/github.com\/browser-use\/browser-use\/pull\/4212\\n\\n**Full Changelog**: https:\/\/github.com\/browser-use\/browser-use\/compare\/0.12.5...0.12.6\",\"bodyExcerpt\":\"## What's Changed * fix: set default temperature to 1.0 for gemini-3 models by @markmcd in https:\/\/github.com\/browser-use\/browser-use\/pull\/4489 * fix bedrock structured output w SchemaOptimizer to flatten schema by @laithrw in https:\/\/github.com\/browser-use\/browser-use\/pull\/4524 * fix: use login shell for PATH configuration in install script by @techiro in https:\/\/github.com\/browser-use\/browser-use\/pull\/4457 * fix(utils): add option to disable SignalHandler for host app signal control by @BillionClaw in https:\/\/git\",\"bodyTruncated\":false,\"publishedAt\":\"2026-04-02T07:54:28Z\",\"prerelease\":false,\"draft\":false,\"assetCount\":0,\"assets\":[]},{\"tagName\":\"0.12.5\",\"name\":\"0.12.5\",\"url\":\"https:\/\/github.com\/browser-use\/browser-use\/releases\/tag\/0.12.5\",\"body\":\"## Security fix\\n\\nRemove `litellm` from core dependencies in response to the litellm supply chain attack (versions 1.82.7 & 1.82.8 backdoored on March 24, 2026).\\n\\n- `pip install browser-use` no longer installs litellm\\n- `ChatLiteLLM` wrapper is preserved \u2014 install litellm separately if needed: `pip install litellm`\\n\\nSee: https:\/\/github.com\/browser-use\/browser-use\/pull\/4515\",\"bodyExcerpt\":\"## Security fix Remove `litellm` from core dependencies in response to the litellm supply chain attack (versions 1.82.7 & 1.82.8 backdoored on March 24, 2026). - `pip install browser-use` no longer installs litellm - `ChatLiteLLM` wrapper is preserved \u2014 install litellm separately if needed: `pip install litellm` See: https:\/\/github.com\/browser-use\/browser-use\/pull\/4515\",\"bodyTruncated\":false,\"publishedAt\":\"2026-03-25T05:17:51Z\",\"prerelease\":false,\"draft\":false,\"assetCount\":0,\"assets\":[]}]","ai_share_view_count":7024,"ai_share_bookmark_count":455,"ai_share_download_count":312,"ai_share_first_party_view_count":54,"ai_share_github_stars":100667,"ai_share_github_watchers":437,"ai_share_github_forks":11195,"ai_share_github_open_issues":283,"ai_share_api_available":true,"ai_share_open_source":true,"ai_share_self_hostable":true,"ai_share_local_first":true,"ai_share_repo_url":"https:\/\/github.com\/browser-use\/browser-use","ai_share_documentation_url":"https:\/\/docs.browser-use.com\/open-source\/customize\/browser\/remote","creator_id":"github-browser-use","creator_slug":"browser-use","creator_name":"browser-use","creator_type":"org","creator_status":"external","creator_intro":"browser-use is the GitHub organization behind this repository.","creator_intro_zh_cn":"browser-use \u662f Browser Use \u9879\u76ee\u7684 GitHub \u7ef4\u62a4\u7ec4\u7ec7\uff0c\u8d1f\u8d23\u5f00\u6e90\u4ed3\u5e93\u3001\u7248\u672c\u53d1\u5e03\u548c\u9879\u76ee\u534f\u4f5c\u3002","creator_github_url":"https:\/\/github.com\/browser-use","creator_avatar_url":"https:\/\/avatars.githubusercontent.com\/u\/192012301?v=4","ai_share_github_readme_path":"README.md","ai_share_github_readme_sha":"37705cf2b20dbc750ce186ead4aa722368439d24","ai_share_github_readme_hash":"757a130cc5daeed6ce85cd9d4b5b4194ecf5f57e525dee6e80bfe8055b92cbdb","ai_share_github_readme_language":"en","ai_share_github_readme_format":"markdown","ai_share_github_description":"Open-source browser automation for AI agents, with a Rust-powered core, browser harness and Python API.","ai_share_github_description_zh_cn":"\u5f00\u6e90 Python \u6d4f\u89c8\u5668\u81ea\u52a8\u5316\u5de5\u5177\uff0c\u8ba9 AI Agent \u64cd\u4f5c\u771f\u5b9e\u7f51\u9875\u5b8c\u6210\u5bfc\u822a\u3001\u70b9\u51fb\u3001\u8f93\u5165\u3001\u622a\u56fe\u548c\u7ed3\u679c\u9a8c\u8bc1\u3002","ai_share_one_sentence_definition":"Open-source browser automation for AI agents, with a Rust-powered core, browser harness and Python API.","ai_share_one_sentence_definition_zh_cn":"\u8ba9 AI Agent \u64cd\u4f5c\u771f\u5b9e\u6d4f\u89c8\u5668\u5b8c\u6210\u7f51\u9875\u5bfc\u822a\u3001\u70b9\u51fb\u3001\u8f93\u5165\u4e0e\u9a8c\u8bc1\u7684\u5f00\u6e90 Python \u5de5\u5177\u3002","ai_share_one_sentence_definition_en":"Open-source browser automation for AI agents, with a Rust-powered core, browser harness and Python API.","ai_share_github_readme_url":"https:\/\/github.com\/browser-use\/browser-use\/blob\/main\/README.md","ai_share_github_readme_raw_url":"https:\/\/raw.githubusercontent.com\/browser-use\/browser-use\/main\/README.md","ai_share_github_readme_markdown":"\n \n \n \"Shows\n<\/picture>\n\n

\n \n \n \n \"The\n <\/picture>\n<\/div>\n\n
\n\"Browser-Use<\/a>\n<\/div>\n\n---\n\n
\n\"Demos\"<\/a>\n\"\"\n\"Docs\"<\/a>\n\"\"\n\"Blog\"<\/a>\n\"\"\n\"Merch\"<\/a>\n\"\"\n\"Github<\/a>\n\"\"\n\"Twitter\"<\/a>\n\"\"\n\"Discord\"<\/a>\n\"\"\n\"Browser-Use<\/a>\n<\/div>\n\n<\/br>\n\n\ud83c\udf24\ufe0f Want to skip the setup? Use our [cloud](https:\/\/cloud.browser-use.com?utm_source=github&utm_medium=readme-skip-setup)<\/b> for faster, scalable, stealth-enabled browser automation!\n\n# \ud83e\udd16 LLM Quickstart\n\n1. Direct your favorite coding agent (Cursor, Claude Code, etc) to [Agents.md](https:\/\/docs.browser-use.com\/llms-full.txt)\n2. Prompt away!\n\n\n\n# \ud83d\udc4b Human Quickstart\n\nBrowser Use 0.13 introduces a new beta agent powered by a Rust core and a browser harness built for current frontier models. It gives the model a real browser\/computer action space, persistent tools, and recovery loops inspired by coding agents.\n\n```text\nPython API -> Rust core -> Browser harness -> Web task done\n```\n\n**1. Install Browser Use with the native core runtime (Python>=3.11):**\n```bash\nuv add \"browser-use[core]\"\n# or: pip install \"browser-use[core]\"\nbrowser\n```\n\nThe `[core]` extra installs the native Browser Use runtime for your platform.\n\n**2. [Optional] Get your API key from [Browser Use Cloud](https:\/\/cloud.browser-use.com\/new-api-key?utm_source=github&utm_medium=readme-quickstart-api-key):**\n```\n# .env\nBROWSER_USE_API_KEY=your-key\n# GOOGLE_API_KEY=your-key\n# ANTHROPIC_API_KEY=your-key\n```\n\n**3. Run your first agent:**\n\n**Browser Use Terminal:** \n```bash\nuv add \"browser-use[core]\"\nbrowser\n```\n\n**Python Script:**\n```python\nfrom browser_use.beta import Agent, BrowserProfile, ChatBrowserUse\n# from browser_use.beta import ChatOpenAI # ChatOpenAI(model='gpt-5.5')\n# from browser_use.beta import ChatAnthropic # ChatAnthropic(model='claude-opus-4-8')\nimport asyncio\n\nasync def main():\n agent = Agent(\n task=\"Find the number of stars of the browser-use repo\",\n llm=ChatBrowserUse(model='openai\/gpt-5.5'),\n # llm=ChatBrowserUse(model='bu-2-0'), # Browser Use's own optimized model\n # llm=ChatOpenAI(model='gpt-5.5'),\n # llm=ChatAnthropic(model='claude-opus-4-8'), # Sonnet also works well.\n browser_profile=BrowserProfile(\n headless=False,\n allowed_domains=[\"*.github.com\"],\n ),\n )\n history = await agent.run()\n print(history.final_result())\n\nif __name__ == \"__main__\":\n asyncio.run(main())\n```\n\nExisting Python agent users can keep using `from browser_use import Agent`. The new Rust-powered beta agent is `from browser_use.beta import Agent`.\n\nCheck out the [library docs](https:\/\/docs.browser-use.com\/open-source\/introduction) and the [cloud docs](https:\/\/docs.cloud.browser-use.com?utm_source=github&utm_medium=readme-cloud-docs) for more!\n\n\n\n# Open Source vs Cloud\n\n\n \n \n \"BU\n<\/picture>\n\nWe benchmark Browser Use across 100 real-world browser tasks. Full benchmark is open source: **[browser-use\/benchmark](https:\/\/github.com\/browser-use\/benchmark)**.\n\n**Use the Open-Source Agent**\n- You need [custom tools](https:\/\/docs.browser-use.com\/customize\/tools\/basics) or deep code-level integration\n- We recommend pairing with our [cloud browsers](https:\/\/docs.browser-use.com\/open-source\/customize\/browser\/remote) for leading stealth, proxy rotation, and scaling\n- Or self-host the open-source agent fully on your own machines\n\n**Use the [Fully-Hosted Cloud Agent](https:\/\/cloud.browser-use.com?utm_source=github&utm_medium=readme-hosted-agent) (recommended)**\n- Much more powerful agent for complex tasks (see plot above)\n- Easiest way to start and scale\n- Best stealth with proxy rotation and captcha solving\n- 1000+ integrations (Gmail, Slack, Notion, and more)\n- Persistent filesystem and memory\n\n\n\n# Demos\n\n\n### \ud83d\udccb Form-Filling\n#### Task = \"Fill in this job application with my resume and information.\"\n![Job Application Demo](https:\/\/github.com\/user-attachments\/assets\/57865ee6-6004-49d5-b2c2-6dff39ec2ba9)\n[Example code \u2197](https:\/\/github.com\/browser-use\/browser-use\/blob\/main\/examples\/use-cases\/apply_to_job.py)\n\n\n### \ud83c\udf4e Grocery-Shopping\n#### Task = \"Put this list of items into my instacart.\"\n\nhttps:\/\/github.com\/user-attachments\/assets\/a6813fa7-4a7c-40a6-b4aa-382bf88b1850\n\n[Example code \u2197](https:\/\/github.com\/browser-use\/browser-use\/blob\/main\/examples\/use-cases\/buy_groceries.py)\n\n\n### \ud83d\udcbb Personal-Assistant.\n#### Task = \"Help me find parts for a custom PC.\"\n\nhttps:\/\/github.com\/user-attachments\/assets\/ac34f75c-057a-43ef-ad06-5b2c9d42bf06\n\n[Example code \u2197](https:\/\/github.com\/browser-use\/browser-use\/blob\/main\/examples\/use-cases\/pcpartpicker.py)\n\n\n### \ud83d\udca1See [more examples here \u2197](https:\/\/docs.browser-use.com\/examples) and give us a star!\n\n\n\n# \ud83d\ude80 Template Quickstart\n\n**Want to get started even faster?** Generate a ready-to-run template:\n\n```bash\nuvx browser-use init --template default\n```\n\nThis creates a `browser_use_default.py` file with a working example. Available templates:\n- `default` - Minimal setup to get started quickly\n- `advanced` - All configuration options with detailed comments\n- `tools` - Examples of custom tools and extending the agent\n\nYou can also specify a custom output path:\n```bash\nuvx browser-use init --template default --output my_agent.py\n```\n\n\n\n# \ud83d\udcbb CLI\n\nFast, persistent browser automation from the command line:\n\n```bash\nbrowser-use open https:\/\/example.com # Navigate to URL\nbrowser-use state # See clickable elements\nbrowser-use click 5 # Click element by index\nbrowser-use type \"Hello\" # Type text\nbrowser-use screenshot page.png # Take screenshot\nbrowser-use close # Close browser\n```\n\nThe CLI keeps the browser running between commands for fast iteration. See [CLI docs](browser_use\/skill_cli\/README.md) for all commands.\n\n### Claude Code Skill\n\nFor [Claude Code](https:\/\/claude.ai\/code), install the skill to enable AI-assisted browser automation:\n\n```bash\nmkdir -p ~\/.claude\/skills\/browser-use\ncurl -o ~\/.claude\/skills\/browser-use\/SKILL.md \\\n https:\/\/raw.githubusercontent.com\/browser-use\/browser-use\/main\/skills\/browser-use\/SKILL.md\n```\n\n\n\n## Integrations, hosting, custom tools, MCP, and more on our [Docs \u2197](https:\/\/docs.browser-use.com)\n\n\n\n# FAQ\n\n
\nWhat's the best model to use?<\/b><\/summary>\n\nWe optimized **ChatBrowserUse()** specifically for browser automation tasks. On avg it completes tasks 3-5x faster than other models with SOTA accuracy.\n\nFor pricing and other LLM providers, see our [supported models documentation](https:\/\/docs.browser-use.com\/supported-models).\n<\/details>\n\n
\nCan I use Claude \/ GPT \/ Gemini through ChatBrowserUse?<\/b><\/summary>\n\nYes. `ChatBrowserUse` accepts provider-prefixed model ids, so a single `BROWSER_USE_API_KEY` reaches all of them \u2014 no separate OpenAI\/Anthropic\/Google keys required:\n\n```python\nfrom browser_use import Agent, ChatBrowserUse\n\nllm = ChatBrowserUse(model='anthropic\/claude-sonnet-4-6') # or 'openai\/gpt-5.5', 'google\/gemini-3-pro'\nagent = Agent(task='...', llm=llm)\n```\n\nFor the best speed and cost we still recommend the default `bu-*` models.\n<\/details>\n\n
\nShould I use the Browser Use system prompt with the open-source preview model?<\/b><\/summary>\n\nYes. If you use `ChatBrowserUse(model='browser-use\/bu-30b-a3b-preview')` with a normal `Agent(...)`, Browser Use still sends its default agent system prompt for you.\n\nYou do **not** need to add a separate custom \"Browser Use system message\" just because you switched to the open-source preview model. Only use `extend_system_message` or `override_system_message` when you intentionally want to customize the default behavior for your task.\n\nIf you want the best default speed\/accuracy, we still recommend the newer hosted `bu-*` models. If you want the open-source preview model, the setup stays the same apart from the `model=` value.\n<\/details>\n\n
\nCan I use custom tools with the agent?<\/b><\/summary>\n\nYes! You can add custom tools to extend the agent's capabilities:\n\n```python\nfrom browser_use import Tools\n\ntools = Tools()\n\n@tools.action(description='Description of what this tool does.')\ndef custom_tool(param: str) -> str:\n return f\"Result: {param}\"\n\nagent = Agent(\n task=\"Your task\",\n llm=llm,\n browser=browser,\n tools=tools,\n)\n```\n\n<\/details>\n\n
\nCan I use this for free?<\/b><\/summary>\n\nYes! Browser-Use is open source and free to use. You only need to choose an LLM provider (like OpenAI, Google, ChatBrowserUse, or run local models with Ollama).\n<\/details>\n\n
\nTerms of Service<\/b><\/summary>\n\nThis open-source library is licensed under the MIT License. For Browser Use services & data policy, see our [Terms of Service](https:\/\/browser-use.com\/legal\/terms-of-service) and [Privacy Policy](https:\/\/browser-use.com\/privacy\/).\n<\/details>\n\n
\nHow do I handle authentication?<\/b><\/summary>\n\nCheck out our authentication examples:\n- [Using real browser profiles](https:\/\/github.com\/browser-use\/browser-use\/blob\/main\/examples\/browser\/real_browser.py) - Reuse your existing Chrome profile with saved logins\n- If you want to use temporary accounts with inbox, choose AgentMail\n- To sync your auth profile with the remote browser, run `curl -fsSL https:\/\/browser-use.com\/profile.sh | BROWSER_USE_API_KEY=XXXX sh` (replace XXXX with your API key)\n\nThese examples show how to maintain sessions and handle authentication seamlessly.\n<\/details>\n\n
\nHow do I solve CAPTCHAs?<\/b><\/summary>\n\nFor CAPTCHA handling, you need better browser fingerprinting and proxies. Use [Browser Use Cloud](https:\/\/cloud.browser-use.com?utm_source=github&utm_medium=readme-faq-captcha) which provides stealth browsers designed to avoid detection and CAPTCHA challenges.\n<\/details>\n\n
\nHow do I go into production?<\/b><\/summary>\n\nChrome can consume a lot of memory, and running many agents in parallel can be tricky to manage.\n\nFor production use cases, use our [Browser Use Cloud API](https:\/\/cloud.browser-use.com?utm_source=github&utm_medium=readme-faq-production) which handles:\n- Scalable browser infrastructure\n- Memory management\n- Proxy rotation\n- Stealth browser fingerprinting\n- High-performance parallel execution\n<\/details>\n\n\n\n
\n\n**Tell your computer what to do, and it gets it done.**\n\n\n\n[![Twitter Follow](https:\/\/img.shields.io\/twitter\/follow\/Magnus?style=social)](https:\/\/x.com\/intent\/user?screen_name=mamagnus00)\n   \n[![Twitter Follow](https:\/\/img.shields.io\/twitter\/follow\/Gregor?style=social)](https:\/\/x.com\/intent\/user?screen_name=gregpr07)\n\n<\/div>\n\n
Made with \u2764\ufe0f in Zurich and San Francisco <\/div>"},"categories":[2462],"tags":[1751,1754,1757,1760,1577,1763,1766],"ai_series":[],"ai_topic":[2670],"ai_creator":[710],"ai_collection":[884],"ai_resource_type":[105],"ai_tool_tag":[],"class_list":["post-3590","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-workflow_skill_pack","tag-ai-agents","tag-ai-tools","tag-browser-automation","tag-browser-use","tag-llm","tag-playwright","tag-python","ai_topic-workflow-automation","ai_creator-browser-use","ai_collection-lobehub-skills","ai_resource_type-workflow_skill_pack"],"source_url":"","preserve_modified":false,"lang":"en","translations":{"en":3590,"zh-cn":52},"pll_sync_post":{},"_links":{"self":[{"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/posts\/3590","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/comments?post=3590"}],"version-history":[{"count":3,"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/posts\/3590\/revisions"}],"predecessor-version":[{"id":3610,"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/posts\/3590\/revisions\/3610"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/media\/3582"}],"wp:attachment":[{"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/media?parent=3590"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/categories?post=3590"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/tags?post=3590"},{"taxonomy":"ai_series","embeddable":true,"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/ai_series?post=3590"},{"taxonomy":"ai_topic","embeddable":true,"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/ai_topic?post=3590"},{"taxonomy":"ai_creator","embeddable":true,"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/ai_creator?post=3590"},{"taxonomy":"ai_collection","embeddable":true,"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/ai_collection?post=3590"},{"taxonomy":"ai_resource_type","embeddable":true,"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/ai_resource_type?post=3590"},{"taxonomy":"ai_tool_tag","embeddable":true,"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/ai_tool_tag?post=3590"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}