browser-use · MagMueller · Feb 18, 2025 · Feb 13, 2025 · Feb 13, 2025 · Feb 13, 2025
diff --git a/.gitignore b/.gitignore
@@ -183,4 +183,6 @@ gcp-login.json
 .idea
 *.txt
 *.pdf
-*.csv
+*.csv
+*.json
+*.jsonl
diff --git a/browser_use/agent/message_manager/service.py b/browser_use/agent/message_manager/service.py
@@ -2,10 +2,8 @@
 
 import json
 import logging
-from datetime import datetime
 from typing import Dict, List, Optional, Type
 
-from langchain_anthropic import ChatAnthropic
 from langchain_core.language_models import BaseChatModel
 from langchain_core.messages import (
 	AIMessage,
@@ -14,7 +12,6 @@
 	SystemMessage,
 	ToolMessage,
 )
-from langchain_openai import ChatOpenAI
 
 from browser_use.agent.message_manager.views import MessageHistory, MessageMetadata
 from browser_use.agent.prompts import AgentMessagePrompt, SystemPrompt
@@ -35,7 +32,6 @@ def __init__(
 		estimated_characters_per_token: int = 3,
 		image_tokens: int = 800,
 		include_attributes: list[str] = [],
-		max_error_length: int = 400,
 		max_actions_per_step: int = 10,
 		message_context: Optional[str] = None,
 		sensitive_data: Optional[Dict[str, str]] = None,
@@ -49,7 +45,6 @@ def __init__(
 		self.estimated_characters_per_token = estimated_characters_per_token
 		self.IMG_TOKENS = image_tokens
 		self.include_attributes = include_attributes
-		self.max_error_length = max_error_length
 		self.message_context = message_context
 		self.sensitive_data = sensitive_data
 		system_message = self.system_prompt_class(
@@ -95,12 +90,12 @@ def __init__(
 		]
 
 		example_tool_call = AIMessage(
-			content=f'',
+			content='',
 			tool_calls=tool_calls,
 		)
 		self._add_message_with_tokens(example_tool_call)
 		tool_message = ToolMessage(
-			content=f'Browser started',
+			content='Browser started',
 			tool_call_id=str(self.tool_id),
 		)
 		self._add_message_with_tokens(tool_message)
@@ -147,7 +142,8 @@ def add_state_message(
 						msg = HumanMessage(content='Action result: ' + str(r.extracted_content))
 						self._add_message_with_tokens(msg)
 					if r.error:
-						msg = HumanMessage(content='Action error: ' + str(r.error)[-self.max_error_length :])
+						last_line = r.error.split('\n')[-1]
+						msg = HumanMessage(content='Action error: ' + last_line)
 						self._add_message_with_tokens(msg)
 					result = None  # if result in history, we dont want to add it again
 
@@ -156,7 +152,6 @@ def add_state_message(
 			state,
 			result,
 			include_attributes=self.include_attributes,
-			max_error_length=self.max_error_length,
 			step_info=step_info,
 		).get_user_message(use_vision)
 		self._add_message_with_tokens(state_message)

diff --git a/browser_use/agent/prompts.py b/browser_use/agent/prompts.py
@@ -160,12 +160,10 @@ def __init__(
 		state: BrowserState,
 		result: Optional[List[ActionResult]] = None,
 		include_attributes: list[str] = [],
-		max_error_length: int = 400,
 		step_info: Optional[AgentStepInfo] = None,
 	):
 		self.state = state
 		self.result = result
-		self.max_error_length = max_error_length
 		self.include_attributes = include_attributes
 		self.step_info = step_info
 
@@ -205,7 +203,7 @@ def get_user_message(self, use_vision: bool = True) -> HumanMessage:
 Current url: {self.state.url}
 Available tabs:
 {self.state.tabs}
-Interactive elements from current page:
+Interactive elements from top layer of the current page inside the viewport:
 {elements_text}
 {step_info_description}
 """
@@ -215,8 +213,8 @@ def get_user_message(self, use_vision: bool = True) -> HumanMessage:
 				if result.extracted_content:
 					state_description += f'\nAction result {i + 1}/{len(self.result)}: {result.extracted_content}'
 				if result.error:
-					# only use last 300 characters of error
-					error = result.error[-self.max_error_length :]
+					# only use last line of error
+					error = result.error.split('\n')[-1]
 					state_description += f'\nAction error {i + 1}/{len(self.result)}: ...{error}'
 
 		if self.state.screenshot and use_vision == True:

diff --git a/browser_use/agent/service.py b/browser_use/agent/service.py
@@ -87,14 +87,13 @@ def __init__(
 			'type',
 			'name',
 			'role',
-			'tabindex',
 			'aria-label',
 			'placeholder',
 			'value',
 			'alt',
 			'aria-expanded',
+			'data-date-format',
 		],
-		max_error_length: int = 400,
 		max_actions_per_step: int = 10,
 		tool_call_in_content: bool = True,
 		initial_actions: Optional[List[Dict[str, Dict[str, Any]]]] = None,
@@ -123,7 +122,6 @@ def __init__(
 		self.save_conversation_path_encoding = save_conversation_path_encoding
 		self._last_result = None
 		self.include_attributes = include_attributes
-		self.max_error_length = max_error_length
 		self.generate_gif = generate_gif
 
 		# Initialize planner
@@ -173,7 +171,6 @@ def __init__(
 			system_prompt_class=self.system_prompt_class,
 			max_input_tokens=self.max_input_tokens,
 			include_attributes=self.include_attributes,
-			max_error_length=self.max_error_length,
 			max_actions_per_step=self.max_actions_per_step,
 			message_context=self.message_context,
 			sensitive_data=self.sensitive_data,
@@ -221,15 +218,15 @@ def _set_version_and_source(self) -> None:
 
 	def _set_model_names(self) -> None:
 		self.chat_model_library = self.llm.__class__.__name__
-		self.model_name = "Unknown"
+		self.model_name = 'Unknown'
 		# Check for 'model_name' attribute first
-		if hasattr(self.llm, "model_name"):
+		if hasattr(self.llm, 'model_name'):
 			model = self.llm.model_name
-			self.model_name = model if model is not None else "Unknown"
+			self.model_name = model if model is not None else 'Unknown'
 		# Fallback to 'model' attribute if needed
-		elif hasattr(self.llm, "model"):
+		elif hasattr(self.llm, 'model'):
 			model = self.llm.model
-			self.model_name = model if model is not None else "Unknown"
+			self.model_name = model if model is not None else 'Unknown'
 
 		if self.planner_llm:
 			if hasattr(self.planner_llm, 'model_name'):
@@ -635,7 +632,6 @@ async def _validate_output(self) -> bool:
 				state=state,
 				result=self._last_result,
 				include_attributes=self.include_attributes,
-				max_error_length=self.max_error_length,
 			)
 			msg = [SystemMessage(content=system_msg), content.get_user_message(self.use_vision)]
 		else:
@@ -646,6 +642,7 @@ class ValidationResult(BaseModel):
 			"""
 			Validation results.
 			"""
+
 			is_valid: bool
 			reason: str
 

diff --git a/browser_use/browser/browser.py b/browser_use/browser/browser.py
@@ -28,7 +28,7 @@ class BrowserConfig:
 		headless: True
 			Whether to run browser in headless mode
 
-		disable_security: False
+		disable_security: True
 			Disable browser security features
 
 		extra_chromium_args: []

diff --git a/browser_use/browser/context.py b/browser_use/browser/context.py
@@ -55,7 +55,7 @@ class BrowserContextConfig:
 	    cookies_file: None
 	        Path to cookies file for persistence
 
-	        disable_security: False
+	        disable_security: True
 	                Disable browser security features
 
 	    minimum_wait_page_load_time: 0.5
@@ -115,7 +115,7 @@ class BrowserContextConfig:
 	maximum_wait_page_load_time: float = 5
 	wait_between_actions: float = 1
 
-	disable_security: bool = False
+	disable_security: bool = True
 
 	browser_window_size: BrowserContextWindowSize = field(default_factory=lambda: {'width': 1280, 'height': 1100})
 	no_viewport: Optional[bool] = None
@@ -939,33 +939,30 @@ async def _input_text_element_node(self, element_node: DOMElementNode, text: str
 		"""
 		try:
 			# Highlight before typing
-			if element_node.highlight_index is not None:
-				await self._update_state(focus_element=element_node.highlight_index)
+			# if element_node.highlight_index is not None:
+			# 	await self._update_state(focus_element=element_node.highlight_index)
 
-			page = await self.get_current_page()
 			element_handle = await self.get_locate_element(element_node)
 
 			if element_handle is None:
 				raise BrowserError(f'Element: {repr(element_node)} not found')
 
 			# Ensure element is ready for input
-			await element_handle.wait_for_element_state('stable', timeout=2000)
-			await element_handle.scroll_into_view_if_needed(timeout=2100)
+			try:
+				await element_handle.wait_for_element_state('stable', timeout=1000)
+				await element_handle.scroll_into_view_if_needed(timeout=1000)
+			except Exception:
+				pass
 
 			# Get element properties to determine input method
 			is_contenteditable = await element_handle.get_property('isContentEditable')
 
 			# Different handling for contenteditable vs input fields
-			try:
-				if await is_contenteditable.json_value():
-					await element_handle.evaluate('el => el.textContent = ""')
-					await element_handle.type(text, delay=5)
-				else:
-					await element_handle.fill(text)
-			except Exception:
-				logger.debug('Could not type text into element. Trying to click and type.')
-				await element_handle.click()
+			if await is_contenteditable.json_value():
+				await element_handle.evaluate('el => el.textContent = ""')
 				await element_handle.type(text, delay=5)
+			else:
+				await element_handle.fill(text)
 
 		except Exception as e:
 			logger.debug(f'Failed to input text into element: {repr(element_node)}. Error: {str(e)}')
@@ -979,8 +976,8 @@ async def _click_element_node(self, element_node: DOMElementNode) -> Optional[st
 
 		try:
 			# Highlight before clicking
-			if element_node.highlight_index is not None:
-				await self._update_state(focus_element=element_node.highlight_index)
+			# if element_node.highlight_index is not None:
+			# 	await self._update_state(focus_element=element_node.highlight_index)
 
 			element_handle = await self.get_locate_element(element_node)
 

diff --git a/browser_use/controller/service.py b/browser_use/controller/service.py
@@ -221,13 +221,25 @@ async def scroll_up(params: ScrollAction, browser: BrowserContext):
 
 		# send keys
 		@self.registry.action(
-			'Send strings of special keys like Backspace, Insert, PageDown, Delete, Enter, Shortcuts such as `Control+o`, `Control+Shift+T` are supported as well. This gets used in keyboard.press. Be aware of different operating systems and their shortcuts',
+			'Send strings of special keys like Escape,Backspace, Insert, PageDown, Delete, Enter, Shortcuts such as `Control+o`, `Control+Shift+T` are supported as well. This gets used in keyboard.press. ',
 			param_model=SendKeysAction,
 		)
 		async def send_keys(params: SendKeysAction, browser: BrowserContext):
 			page = await browser.get_current_page()
 
-			await page.keyboard.press(params.keys)
+			try:
+				await page.keyboard.press(params.keys)
+			except Exception as e:
+				if 'Unknown key' in str(e):
+					# loop over the keys and try to send each one
+					for key in params.keys:
+						try:
+							await page.keyboard.press(key)
+						except Exception as e:
+							logger.debug(f'Error sending key {key}: {str(e)}')
+							raise e
+				else:
+					raise e
 			msg = f'⌨️  Sent keys: {params.keys}'
 			logger.info(msg)
 			return ActionResult(extracted_content=msg, include_in_memory=True)
-Original file line number
+Diff line change
@@ Expand Up / @@ -183,4 +183,6 @@ gcp-login.json @@
     .idea
     *.txt
     *.pdf
-    *.csv
+    *.csv
+    *.json
+    *.jsonl