03c80a7493
Reduce max actions per step in Agent class from 10 to 1 for improved control over action execution.
2025-06-19 16:34:34 +02:00
f463a9383c
init and config fixes
2025-06-19 07:20:43 -07:00
9bc619877c
small change to system prompt
2025-06-19 15:20:37 +02:00
129590672d
small update to descriptions
2025-06-19 14:09:06 +02:00
778536a3ce
move all sync code into sync dir
2025-06-19 01:28:22 -07:00
ccd11e26f5
remove unneeded event_type fields
2025-06-19 01:19:30 -07:00
0b6ebea431
Merge branch 'main' into mert/new_everything
2025-06-19 09:49:43 +02:00
e2e11966d3
Merge branch 'main' into typing-improvements
2025-06-18 21:20:47 -04:00
927b286d71
Handle sensitive data in agent state and handle user request being visible in agent history in case of update
2025-06-18 15:53:35 +02:00
463f147e6a
Refactor add_new_task method in service.py to update agent history description with new user requests, removing the previous message creation logic.
2025-06-18 14:30:04 +02:00
bbc8d26f63
- Updated ActionResult to include an attachments field for displaying files in the done message.
...
- Enhanced logging in the Agent class to inform users about available attachments after task completion.
- Revised system prompt guidelines to clarify the use of text and files_to_display in the done action.
2025-06-18 14:24:27 +02:00
600febfc63
Add guideline for format specification in done action of system prompt
2025-06-18 13:19:04 +02:00
474476f0ae
switch to external bubus lib
2025-06-18 01:41:16 -07:00
f35ea5be4d
fixes
2025-06-17 22:38:59 -07:00
d0ec528802
fixed events
2025-06-17 22:38:59 -07:00
b8621e1089
Improve logging and action result handling: update messages for new elements detection and adjust success flag in action results so that agent doesn't stop.
2025-06-17 18:58:17 +02:00
a0e60a1503
Update prompts and service files: increase max clickable elements length and enhance file writing instructions to specify allowed extensions.
2025-06-17 16:01:21 +02:00
f7a8e99f13
Enhance user guidance in data extraction: add reminder to save information during read state and refine instructions for structured data extraction to clarify usage and file saving practices.
2025-06-17 15:10:07 +02:00
95fcd75832
Refine system prompt instructions: clarify transient data display conditions, update terminology for data extraction, and emphasize the importance of saving information during the read state.
2025-06-17 14:51:42 +02:00
2a7985df5f
Update prompts.py to enhance user guidance in scrolling and data extraction, improve formatting of agent and browser state descriptions, and streamline the structure of the output for better readability.
2025-06-17 14:49:22 +02:00
34d2894d56
Refactor MessageManager initialization: remove unnecessary agent history description, adjust message formatting, and clean up commented-out code for clarity.
2025-06-17 14:49:03 +02:00
0c34c399db
Merge branch 'main' into mert/new_everything
2025-06-17 11:19:06 +02:00
8ea4b4c71a
centralize playwright/patchwright imports, and update typing/checks to reduce lint errors
2025-06-16 22:57:20 -04:00
d728148200
tweak logger emoji
2025-06-14 01:28:47 -07:00
66fbb8b282
Merge branch 'main' into eventbus
2025-06-14 01:19:15 -07:00
b3ed9cbbba
Update browser_use/agent/cloud_events.py
...
Co-authored-by: cubic-dev-ai[bot] <191113872+cubic-dev-ai[bot]@users.noreply.github.com>
2025-06-14 03:42:42 -04:00
3c65acf225
improve conversation saving
2025-06-13 17:02:29 -07:00
582c039c15
make save_context and other file operations async, setup traces and other dirs on startup
2025-06-12 18:38:12 -07:00
e8f2467ab2
Refactor action result handling in message_manager and controller services
...
Updated the logic for processing action results in service.py files. Changed the conditional check for memory in message_manager to use 'elif' for clarity. In controller, modified the return statement to handle errors more effectively by returning an ActionResult with the error message instead of memory, improving error logging and debugging capabilities.
2025-06-11 11:36:06 +02:00
656a129fba
include original LLM status_code in raised exception
2025-06-10 05:33:08 -07:00
9a08e3b031
add better logging with reference ids for each object
2025-06-10 02:10:57 -07:00
fd4d7ea07a
use llm.ainvoke non-blocking async call
2025-06-10 02:03:02 -07:00
a8eb2bac2f
make markdownify non-blocking async
2025-06-10 02:01:16 -07:00
023c702f4d
pass logger object down the chain to sub-objects for re-use
2025-06-10 00:38:38 -07:00
930a7aa4de
use self.logger for timing log lines
2025-06-10 00:24:50 -07:00
18a97acf76
Enhance agent state representation and update system prompt documentation
...
Refactored the agent state output in prompts.py to include XML-like tags for user requests, file system details, and state descriptions, improving structure and readability. Updated system prompt documentation to clarify memory tracking instructions, emphasizing the importance of counting pages visited and items found for better progress tracking.
2025-06-10 08:51:04 +02:00
1875c34416
Refactor argument naming and enhance success message in service.py
...
Updated the key from 'arguments' to 'args' for consistency in the tool call structure. Revised the success message to specify that data was written to 'todo.md' and 'github.md', improving clarity and user feedback.
2025-06-10 08:04:02 +02:00
1c7b41eacd
Update argument naming and message content in service.py
...
Changed the key 'args' to 'arguments' for clarity in the tool call structure. Additionally, updated the success message to reflect that data was written to both 'todo.md' and 'github.md', improving the accuracy of feedback provided to users.
2025-06-10 07:59:35 +02:00
c0df2feb11
Enhance system prompt and message formatting in service.py
...
Updated the system prompt documentation to clarify user request handling and added specific instructions for open-ended tasks. Modified message formatting in service.py to include XML-like tags for context, sensitive data, and user requests, improving structure and readability of messages. This change aims to enhance the clarity of communication between the agent and users.
2025-06-10 07:51:12 +02:00
1c7eb2b849
add some newlines in logging output to visually separate steps
2025-06-09 22:05:37 -07:00
f0df1f195f
tweak mem0 warning
2025-06-09 22:04:54 -07:00
64e2476f22
gracefully degrade when mem0 db is corrupt
2025-06-09 21:00:04 -07:00
39bc9327f7
silence faiss-cpu swig deprecation warnings
2025-06-09 20:34:45 -07:00
5a70f1f1ce
dont reset BrowserSession foreground page on Agent.__init__
2025-06-09 18:51:30 -07:00
0cd31c8b2a
throw a hard error requiring BrowserSession to be initialized before passing it in if keep_alive=True
2025-06-09 18:35:44 -07:00
8481e2cc04
Add control character handling in JSON extraction utility
...
Enhance the `extract_json_from_model_output` function to fix control characters in JSON strings before parsing. Introduced a new helper function, `_fix_control_characters_in_json`, to ensure valid JSON formatting by escaping control characters within string values. This improves the robustness of JSON parsing and error handling in the utility.
2025-06-09 23:11:46 +02:00
d998e95ac0
Improve error handling in LLM API calls by appending exception messages to LLMException. This change enhances the clarity of error logs in both utils.py and service.py, facilitating better debugging and traceability.
2025-06-09 22:47:03 +02:00
db74f1ade3
Refactor response logging in service.py to improve emoji representation based on evaluation state. Updated checks to be case-insensitive for 'success' and 'failure', enhancing robustness in response handling.
2025-06-09 22:19:25 +02:00
7c038ae944
Update system prompt documentation to improve clarity and consistency in reasoning patterns. Standardized references to user requests and enhanced instructions for evaluating previous actions and preparing tool calls. Emphasized the importance of tracking progress and verifying completeness before finalizing actions.
2025-06-09 22:19:20 +02:00
7298617f02
Update system prompt documentation to enhance clarity and structure. Revised definitions for User Request, Agent State, and Browser State, and emphasized the importance of user satisfaction in task completion. Improved readability of action and reasoning rules to facilitate better understanding of user interactions.
2025-06-09 22:02:48 +02:00
mertunsall