2d97ee619d
update system prompts to emphasize recovery
2025-08-22 11:54:50 +01:00
1c27af4eab
Merge branch 'main' into mert/save_coordinates_in_lmnr
2025-08-21 21:30:34 +01:00
59fb17cdbe
record coordinates of click
2025-08-21 21:01:18 +01:00
9d27efaf04
remove agent call method
2025-08-21 11:32:42 -07:00
bef7715fc8
Merge branch 'main' into fast-example
2025-08-21 08:33:54 -07:00
f39934ec3d
Merge branch 'main' into mert/add_unexecuted_actions_into_context
2025-08-21 13:26:32 +01:00
543598853f
Make agent callable
2025-08-20 22:36:18 -07:00
4bad26b899
Default llm
2025-08-20 22:18:07 -07:00
f58f25d7d6
Use copy of cached_selector_map to avoid cache being remove
2025-08-20 21:30:51 -07:00
7d53f108c7
Enhance logging: include 'google_genai' in logging configuration and improve attachment logging format. Log task duration on completion.
2025-08-20 21:07:26 -07:00
fa07427da2
small log change
2025-08-20 20:27:23 -07:00
6c6907bfc2
list unexecuted actions
2025-08-20 23:52:45 +01:00
8f370ec6a8
Increase frequency_penalty and reset timeouts
2025-08-19 17:33:29 -07:00
edfe33c9f6
Bump llm timeout to see real error - revert this later
2025-08-19 15:00:11 -07:00
49e7f5686e
Feedback message if wrong output
2025-08-19 14:59:17 -07:00
12a7fe9ff3
Small color change
2025-08-19 00:25:05 -07:00
7f3ae7469a
Fix new lines
2025-08-19 00:17:58 -07:00
22a86b8a3c
Change colors
2025-08-19 00:11:50 -07:00
13dce77c98
Update info logger to debug and clean up logs
2025-08-19 00:02:42 -07:00
8c1b4bf794
Update info logger to debug
2025-08-18 23:37:31 -07:00
76bd5f21d7
preload url fix punctuation with brackets
2025-08-18 20:45:46 -07:00
2246231cfc
preload url fix punctuation
2025-08-18 20:40:45 -07:00
4f433c1ea0
feat: set default working language to english
2025-08-17 12:44:56 -07:00
76512fbfdd
fix Failed to build DOM tree: cannot reuse already awaited coroutine
2025-08-16 17:58:45 -07:00
8d86832095
fix SwitchTab by using TargetID for all tab_ids
2025-08-15 16:45:27 -07:00
78ee73c949
cleaner logging slightly
2025-08-14 19:06:18 -07:00
f00c271468
Update system prompts with recent feedback ( #2688 )
...
Updated system prompts based on recent feedback and testing
<!-- This is an auto-generated description by cubic. -->
---
## Summary by cubic
Updated all system prompt files and code to clarify how to reference
available file paths and ensure all findings are shared with the user in
the reply.
- **Prompt Improvements**
- Added a reminder to use absolute file paths when referencing files.
- Updated guidance for the `done` action to include all relevant
information in the `text` field.
<!-- End of auto-generated description by cubic. -->
2025-08-15 00:19:10 +01:00
beaede9631
small additions
2025-08-15 00:18:12 +01:00
0c8d863d6c
Merge branch 'main' into mert/fix_error_messages
2025-08-14 22:54:13 +01:00
35e122def1
use go_to_url action to preload urls in tasks instead of force-navigating
2025-08-14 14:32:31 -07:00
86fb443895
remove old unused playwright types
2025-08-14 14:32:31 -07:00
f5284c996f
Merge branch 'main' into mert/fix_error_messages
2025-08-14 20:46:59 +01:00
d243ae7bcb
make sure that read state can handle multiple items nicely
2025-08-14 20:45:02 +01:00
5de5c746fd
Merge branch 'main' into fix/issue-2614-queue-shutdown-clean
2025-08-14 12:37:21 -07:00
6c2e854ee5
dont use preload cache for anything after the first step
2025-08-14 00:11:03 -07:00
e9769ac18e
tweak logging line lengths
2025-08-13 22:55:50 -07:00
079e9e6445
preload urls in tasks by default
2025-08-13 22:55:07 -07:00
c40daa14eb
hot-fix/error-count-only-for-single-steps
2025-08-12 23:42:38 -07:00
2801c84cfd
fix-done-action-not defined
2025-08-12 21:40:52 -07:00
8732b41bdb
highlight actions nicely in logs
2025-08-12 17:50:59 -07:00
56086329fd
Only increase errors if first action fails
2025-08-12 14:41:36 -07:00
5a908851d5
Increase consecutive_failures if action error
2025-08-12 13:35:13 -07:00
a29d4ac8bd
reverted multi-act back to branch_path_hashes
2025-08-12 12:04:23 -07:00
c440e29ea1
fix cached selector map and js popups
2025-08-12 05:00:31 -07:00
7a0889cf20
formatting
2025-08-12 04:14:53 -07:00
d1e569eb31
raise BrowserError exceptions instead of returning success=False
2025-08-12 04:14:53 -07:00
9f9558e49f
Merge branch 'main' into local-remote-split
2025-08-11 13:15:21 -07:00
571bfa78a4
style: Remove trailing whitespace to fix pre-commit formatting
...
🤖 Generated with [Claude Code](https://claude.ai/code )
Co-Authored-By: Claude <noreply@anthropic.com >
2025-08-11 20:15:21 +02:00
8a11d05bf8
Fix #2614 : Prevent QueueShutDown exception when using add_new_task + multiple runs
...
The issue occurred when calling agent.run() after using add_new_task(), because
the second run() attempted to dispatch CreateAgentSessionEvent to an already
shut down event queue from the first run.
This change:
- Adds session_initialized flag to AgentState to track if session events were dispatched
- Only dispatches CreateAgentSessionEvent on the first run() call
- CreateAgentTaskEvent is still dispatched on every run() as tasks can change
- Maintains all existing functionality while preventing the QueueShutDown exception
Fixes the workflow: agent.run() -> agent.add_new_task() -> agent.run() ✅
🤖 Generated with [Claude Code](https://claude.ai/code )
Co-Authored-By: Claude <noreply@anthropic.com >
2025-08-11 20:06:11 +02:00
aa45460617
Fixed multiact cache
2025-08-10 18:25:22 -07:00
mertunsall