65f87b7fca
fix sensitive_data redaction order to prevent substring leaks
2026-04-11 18:16:24 -04:00
39b911398e
fix load_from_dict and final_result crashing on incomplete history entries
2026-03-23 13:20:03 -04:00
cf6a799d63
fix: prevent KeyError/IndexError in AgentHistoryList.load_from_dict and final_result
...
- load_from_dict: use data.get('history', []) and h.get() for safe access
to avoid KeyError when history dict is missing keys (e.g. legacy/partial
history files saved before a schema change)
- final_result: add len() check before accessing result[-1] to prevent
IndexError when the result list is empty, consistent with is_done(),
is_successful(), judgement() and other methods in the same class
2026-03-23 01:46:50 +08:00
653d795f15
changed default compaction to 25 steps
2026-03-17 17:17:20 -07:00
d864bd7a8e
fix(agent): disable json escaping in conversation history ( #3090 )
2026-03-01 14:11:02 -05:00
adf3bf9f43
fixed ruff formatting issues
2026-02-21 09:42:20 -08:00
cb6fce4d1c
remove simple judge
2026-02-20 19:10:02 -08:00
270e418cea
feat(agent): expose max_clickable_elements_length as Agent parameter
...
- Add max_clickable_elements_length parameter to Agent.__init__ with default 40000
- Add max_clickable_elements_length to AgentSettings
- Pass parameter through MessageManager to AgentMessagePrompt
- Bump version to 0.11.10a2
Co-authored-by: Magnus Müller <MagMueller@users.noreply.github.com >
2026-02-17 02:07:09 +00:00
8263a8d91d
change max failure to 5
2026-02-10 19:12:45 -08:00
42cbf9ca15
Changed consecutive failure to 7
2026-02-10 18:31:30 -08:00
33831f2b12
increase default consecutive failure to 5
2026-02-10 17:08:35 -08:00
d1c9a38c00
Merge branch 'main' into message-compaction
2026-02-05 20:29:03 -08:00
6b9910a546
changed the threshold for compaction
2026-02-05 16:41:08 -08:00
79501fe2cd
added message compaction for agent
2026-02-05 12:26:11 -08:00
bb5914f251
added milder nudge for action loop
2026-02-05 11:17:20 -08:00
cec4aea3bc
increased threshold for loop detection
2026-02-04 16:34:42 -08:00
b96b9436fc
fixed agent loop detection
2026-02-03 16:00:08 -08:00
bd4ead5d2e
Action Loop detection
2026-02-03 15:19:59 -08:00
0dba38ed95
Merge branch 'main' into align-agent-success-with-judge
2026-02-02 22:16:11 -08:00
8565c7d267
updated agent success with simple judge
2026-02-02 22:14:45 -08:00
bce72138c4
fixed lint issue
2026-02-02 21:49:48 -08:00
4d5976c427
Merge branch 'main' into improve-agent-planning
2026-02-02 21:45:06 -08:00
74e26e0b93
Merge branch 'main' into align-agent-success-with-judge
2026-02-02 17:14:43 -08:00
f1fbdbc1a8
removed self verify injection to agent
2026-02-02 17:10:01 -08:00
4a522a6f1c
dynamic planing exploration
2026-02-02 16:26:50 -08:00
71e7481034
added one more retry if simple judge thinks it failed the task
2026-02-02 16:13:57 -08:00
defe37dd28
Improve multiact for agent
2026-02-02 12:50:03 -08:00
c33e9494c5
align agent success with simple judge
2026-02-02 12:33:27 -08:00
5ad365e31b
added basic planing to agents
2026-02-02 12:28:29 -08:00
38703d5ad7
perf: lazy import for openai and reportlab to reduce startup latency
...
- agent/views.py: Move RateLimitError import inside format_error() function
to avoid loading openai SDK (~800ms) at module level
- llm/messages.py: Replace openai.BaseModel with pydantic.BaseModel directly
to remove unnecessary openai dependency
- filesystem/file_system.py: Move reportlab imports inside sync_to_disk_sync()
to avoid ~40ms startup cost when PDF generation is not used
- utils.py: Convert OpenAIBadRequestError and GroqBadRequestError to lazy
loaders to avoid loading SDKs at module level
This improves import time for users who don't use OpenAI provider,
especially when using Anthropic, Google, or other providers.
🤖 Generated with [Claude Code](https://claude.com/claude-code )
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com >
2025-12-14 01:48:47 +09:00
9b894150d5
fix structured output for sandboxes
2025-12-12 15:35:34 -08:00
ed7cfa4ab2
fix title problem in schema as well as enum problems
2025-11-25 13:02:39 -08:00
d1b47eba50
Refactor: Reduce max actions per step to 3
...
Co-authored-by: mailmertunsal <mailmertunsal@gmail.com >
2025-11-22 01:29:59 +00:00
5e356108f9
feat: added AI summary for rerun task
2025-11-20 21:40:30 +05:30
e57f7ac14f
added variable detection from agent history
2025-11-20 14:39:32 +05:30
8f8713b001
feat: added step_interval calculation for history
2025-11-20 10:28:28 +05:30
dfbba0e062
add ground truth to judge
2025-11-12 11:16:40 -08:00
f3961e0b39
ask for shorter judge output
2025-11-08 09:55:02 -08:00
d3a88ce351
vision defaults to True
2025-11-06 16:29:09 -08:00
67d2f12665
default to PNG everywhere
2025-11-06 14:51:52 -08:00
e68114662f
fix linter
2025-11-06 13:59:23 -08:00
c4e2dfb39a
Update browser_use/agent/views.py
2025-11-06 13:42:38 -08:00
cb814da33e
Add impossible_task and reached_captcha to JudgementResult
...
Co-authored-by: mailmertunsal <mailmertunsal@gmail.com >
2025-11-06 21:38:18 +00:00
03c9e94dc6
allow reading images such as jpg/png, as well as docx, fixes to the flash system prompt
2025-11-03 16:41:27 -08:00
299cf6a9d0
Merge remote-tracking branch 'upstream/main' into library-judge
2025-10-31 21:03:23 -07:00
15911c8c72
fix bugs in library judge and make it off by default
2025-10-31 20:20:36 -07:00
0f1946cb54
implement simple judge directly into the library on done action
2025-10-31 20:01:07 -07:00
5e428e829b
Merge branch 'main' into feature/remote-code
2025-10-31 12:19:39 -07:00
66c2d7dd97
Output structure msg
2025-10-30 10:54:33 -07:00
78beda65fb
Revert "Set vision to true by default"
...
This reverts commit 62ad269975
2025-10-30 00:44:53 -07:00
Laith Weinberger