0b5e2b5d22
disable compaction by default
2026-02-09 21:25:39 -05:00
f3ed2acb11
Merge branch 'main' into fix/sensitive-data-placeholder-instructions
2026-02-09 14:40:45 -05:00
7f3e26bef0
Merge branch 'main' into fix/sensitive-data-placeholder-instructions
2026-02-09 14:30:09 -05:00
f469617aa7
Merge branch 'main' into fix/issue-3970-browser-behavior
2026-02-09 14:28:52 -05:00
9131c88f98
Merge branch 'main' into 3615
2026-02-08 17:34:07 -05:00
e1e170a5fd
fix agent not stopping: consecutive_failures counter was resetting on multi-action error steps
2026-02-08 17:25:39 -05:00
53c1b0dfb9
Merge origin/issue-2748 into HEAD
2026-02-08 16:37:44 -05:00
1737c5bea4
lint
2026-02-08 16:36:45 -05:00
1724d314de
Merge branch 'main' into issue-2748
2026-02-08 16:30:56 -05:00
4d05218d7d
fix: redirect logging to stderr so MCP stdio doesn't break ( #2748 )
2026-02-08 16:29:05 -05:00
1629799dd3
fix(openai): structured output call shouldn't depend on schema injection
...
Fix indentation so structured-output request executes even when add_schema_to_system_prompt is false; add regression tests for choices=null in both structured and non-structured paths.
2026-02-08 16:03:21 -05:00
d15f0a32c7
Merge branch 'main' into fix/3897-llm-response-parsing
2026-02-08 15:43:05 -05:00
62d19f0c9a
fix(openai): handle proxy responses with missing/null choices; undo test change
...
Avoid indexing response.choices[0] when proxies return choices=null/empty; raise a clear ModelProviderError (with base_url hint) and use the validated first choice for content/finish_reason
2026-02-08 15:37:59 -05:00
880b636422
Merge branch 'main' into 3895
2026-02-08 11:15:44 -05:00
9e95241bbb
stop agent if browser closed
2026-02-08 11:14:49 -05:00
ab83ee7a8a
fix: improve sensitive_data placeholder instructions for LLM clarity
...
Fixes #3970
The LLM was typing literal placeholder names (e.g., 'user_name') instead
of wrapping them in <secret> tags (e.g., '<secret>user_name</secret>').
Changes to _get_sensitive_data_description():
- Added clear 'SENSITIVE DATA' header to draw attention
- Changed placeholder list format from Python list to bullet points
- Added explicit 'MUST' instruction emphasizing the required format
- Added concrete example using the first actual placeholder name
- Added explanation that the system will replace tags with actual values
This makes it much clearer to the LLM that it needs to wrap placeholder
names in <secret> tags for the automatic replacement mechanism to work.
2026-02-06 18:10:14 +05:30
d86be39eeb
Merge branch 'main' into fix/structured-output-schema-collision
2026-02-05 23:44:03 -05:00
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
04da1fd30c
strip success from JSON schema so invisible to the LLM, avoiding confusion
2026-02-05 16:49:36 -05:00
79501fe2cd
added message compaction for agent
2026-02-05 12:26:11 -08:00
3c57cdc756
Merge branch 'main' into update-action-loop-prompt
2026-02-05 12:08:33 -08:00
995cb0176b
updated default max step to 500
2026-02-05 11:20:47 -08:00
bb5914f251
added milder nudge for action loop
2026-02-05 11:17:20 -08:00
99be7b1b48
Merge branch 'main' into fix/no-tab-switch
2026-02-04 21:24:34 -05:00
9d81f64b2e
Merge branch 'main' into kalil0321/har_watchdog
2026-02-04 21:17:10 -05:00
7455b3c1af
deduplicate code; allow CDP events to propagate; error handling
2026-02-04 21:14:14 -05:00
d66181cdc6
notify agent if new tab opened
2026-02-04 21:02:13 -05:00
332694bf1a
Added date info to simple judge
2026-02-04 17:04:39 -08:00
cec4aea3bc
increased threshold for loop detection
2026-02-04 16:34:42 -08:00
d24dea0aa4
Merge branch 'main' into improve-input-field-handling
2026-02-03 18:22:19 -08:00
5d7ff041d3
improved concatenation heuristic
2026-02-03 18:02:58 -08:00
bbf3cc0164
Merge branch 'main' into action-detection-loop
2026-02-03 17:18:06 -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
f9c36f11a4
lint
2026-02-03 17:59:24 -05:00
bcb5a2ad63
Merge branch 'main' into feat/smart-pdf-reading
2026-02-03 17:44:37 -05:00
799e714c2b
mitigate injection; don't drop overbudget pages
...
- restricted read_long_content file access to only paths in available_file_paths or browser_session.download_files to mitigate injection
- made PDF page selection truncate pages that exceed char budget instead of dropping them entirely
- fixed truncation hint in file_system.py to reference read_long_context instead of search_page
2026-02-03 17:42:02 -05:00
a511ba2f1e
Merge branch 'main' into 3619
2026-02-03 17:32:46 -05:00
2a5768f3da
fixed pyright issues
2026-02-03 13:02:03 -08:00
749fa56d1a
improve text field handling
2026-02-03 12:52:56 -08:00
38fb5b12df
Merge remote-tracking branch 'origin/main' into feat/smart-pdf-reading
2026-02-03 14:15:49 -05:00
da5b4237ba
fixed lint issue
2026-02-02 23:01:52 -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
65405d8a79
Merge branch 'main' into align-agent-success-with-judge
2026-02-02 21:50:02 -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
dae6dc6d0e
update agent success with judge verdict
2026-02-02 18:34:11 -08:00
023367052c
in flash mode disable planning
2026-02-02 18:28:28 -08:00
Laith Weinberger