Skip to content

Open-source Python library for AI browser agents, plus Browser Use Cloud: stealth browsers and a hosted agent API.

Delete examples/mert.py

Mert Unsala28f4268bf0f
diff --git a/examples/mert.py b/examples/mert.py
deleted file mode 100644
index d4e2ba0e7..000000000
--- a/examples/mert.py
+++ /dev/null
@@ -1,50 +0,0 @@
-import asyncio
-import os
-import sys
-
-from browser_use.llm.google.chat import ChatGoogle
-
-sys.path.append(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
-
-from dotenv import load_dotenv
-
-load_dotenv()
-
-from browser_use import Agent
-
-try:
- from lmnr import Laminar
-
- Laminar.initialize(project_api_key=os.getenv('LMNR_PROJECT_API_KEY'))
-except Exception as e:
- print(f'Error initializing Laminar: {e}')
-
-
-# llm = ChatOpenAI(base_url='https://browseruse--browseruse-v0-serve.modal.run/v1',
-# model='browser-use/Qwen3-VL-8B-Instruct-301025-1epoch',
-# api_key='aitorloveskebabs',
-# temperature=1.0,
-# dont_force_structured_output=True)
-
-llm = ChatGoogle(model='gemini-flash-latest', temperature=1.0)
-# llm = ChatAnthropic(model='claude-sonnet-4-5-20250929', temperature=1.0)
-# llm = ChatOpenAI(model='gpt-5-mini', temperature=1.0)
-
-task = """IMPORTANT RULE: Use only clicking by coordinate and sending keys actions. Do not use any other actions. Go to https://levelshealth.typeform.com/waitlist-uk and input John into the first name field and click on OK button using coordinates."""
-# task = 'go to example.com and use the extract tool to extract the content of the page.'
-
-task = 'How many studio albums were published by Mercedes Sosa between 2000 and 2009 (included)? You can use the latest 2022 version of english wikipedia.'
-
-task = "The Metropolitan Museum of Art has a portrait in its collection with an accession number of 29.100.5. Of the consecrators and co-consecrators of this portrait's subject as a bishop, what is the name of the one who never became pope?"
-
-task = 'go to https://browser-use.github.io/stress-tests/challenges/angularjs-form.html and go back.'
-# Test: llm_screenshot_size should auto-configure to (1400, 850) for Claude Sonnet
-agent = Agent(task=task, llm=llm, highlight_elements=True, flash_mode=True, use_judge=True)
-
-
-async def main():
- await agent.run(max_steps=30)
-
-
-if __name__ == '__main__':
- asyncio.run(main())