the top google links point to deprecated docs, so redirect:
"/api-reference/browser-profiles/get-browser-profile" ->
"/api-reference/profiles/get-browser-profile"
"/cloud/v2/quickstart" -> "/cloud/v1/quickstart"
<!-- This is an auto-generated description by cubic. -->
---
## Summary by cubic
Update docs redirects so popular Google links point to current pages and
avoid deprecated content.
- /api-reference/browser-profiles/get-browser-profile ->
/api-reference/profiles/get-browser-profile
- /cloud/v2/quickstart -> /cloud/v1/quickstart
<!-- End of auto-generated description by cubic. -->
Auto-generated PR for: remove temp from gemini
<!-- CURSOR_SUMMARY -->
---
> [!NOTE]
> Moves ChatGoogle instantiation into run_search and removes the
temperature parameter in the gemini example.
>
> - **Examples**:
> - **`examples/models/gemini.py`**:
> - Move `llm = ChatGoogle(...)` into `run_search()` and remove
`temperature` argument.
> - Reorder `Agent` initialization args (pass `llm` before `task`).
>
> <sup>Written by [Cursor
Bugbot](https://cursor.com/dashboard?tab=bugbot) for commit
ea551e8b6c. This will update automatically
on new commits. Configure
[here](https://cursor.com/dashboard?tab=bugbot).</sup>
<!-- /CURSOR_SUMMARY -->
Auto-generated PR for: change gemini 2.5 flash to gemini latest
<!-- CURSOR_SUMMARY -->
---
> [!NOTE]
> Adds support for `gemini-flash-latest` and `gemini-flash-lite-latest`
and updates docs/examples to use these models instead of
`gemini-2.5-flash`.
>
> - **LLM (Google Gemini)**:
> - Add `gemini-flash-latest` and `gemini-flash-lite-latest` to
`VerifiedGeminiModels` in `browser_use/llm/google/chat.py`.
> - **Docs & Examples**:
> - Replace `gemini-2.5-flash` with `gemini-flash-latest` across
`README.md`, `docs/quickstart.mdx`,
`docs/customize/agent/supported-models.mdx`,
`examples/models/gemini.py`, and `examples/simple.py`.
> - Use `gemini-flash-lite-latest` in speed-focused examples:
`docs/examples/templates/fast-agent.mdx`,
`examples/getting_started/05_fast_agent.py`.
> - Update recommendation bullet to prefer `gemini-flash-latest`.
>
> <sup>Written by [Cursor
Bugbot](https://cursor.com/dashboard?tab=bugbot) for commit
9a6181afd7. This will update automatically
on new commits. Configure
[here](https://cursor.com/dashboard?tab=bugbot).</sup>
<!-- /CURSOR_SUMMARY -->
Add this feature to the documentation and an example on how to use it.
<!-- This is an auto-generated description by cubic. -->
---
## Summary by cubic
Documented the prohibited_domains browser setting and added a runnable
example that blocks navigation to selected domains and subdomains. When
both lists are set, allowed_domains still takes precedence.
- New Features
- Documented prohibited_domains in all-parameters.mdx, including pattern
formats and precedence over allowed_domains.
- Added examples/features/blocked_domains.py showing how to configure
BrowserProfile with prohibited_domains and run an agent.
<!-- End of auto-generated description by cubic. -->
<!-- This is an auto-generated description by cubic. -->
## Summary by cubic
Added Laminar debug tags to observability spans and temporarily forced
debug mode to True to surface more debug data.
- **New Features**
- Tags added: ['observe', 'observe_debug'] in my_function;
['observe_debug'] in debug_function.
- **Migration**
- Create 'observe' and 'observe_debug' tags in Laminar before deploying.
<!-- End of auto-generated description by cubic. -->
Reduces a dependency
<!-- This is an auto-generated description by cubic. -->
---
## Summary by cubic
Replaced aiofiles with anyio.Path for async file I/O to reduce
dependencies and simplify code. Also updated DownloadsWatchdog to react
to BrowserStateRequestEvent, improving PDF download flow.
- New Features
- DownloadsWatchdog listens to BrowserStateRequestEvent and dispatches a
NavigationCompleteEvent with the current URL and target_id.
- Refactors
- Swapped aiofiles for anyio.Path.read_text/write_text across DOM
playground, Gmail auth, pricing cache, examples, and CI task runner.
- Removed aiofiles from pyproject.toml dependencies.
<!-- End of auto-generated description by cubic. -->
Auto-generated PR for branch: speed-ups3
<!-- This is an auto-generated description by cubic. -->
---
## Summary by cubic
Improves browsing speed and reliability with a simplified search tool,
viewport-aware scrolling, and more robust tab management. Also caps
agent history sizes and tunes LLM defaults to reduce timeouts and
errors.
- **New Features**
- Replaced SearchGoogleAction with a unified SearchAction that supports
duckduckgo (default), google, and bing; defaults to DuckDuckGo to avoid
reCAPTCHA. Backward-compatible alias kept.
- Viewport-aware scrolling: scrolls one page at a time for multi-page
requests; uses detected viewport height and handles fractional pages.
- Gemini defaults updated: max_output_tokens increased to 8192;
thinking_budget defaults to 0 for flash/flash-lite.
- **Bug Fixes**
- Hardened tab switching and closing: validate target IDs, clean up
stale sessions, and don’t fail if a tab is already closed.
- Added target validity checks and session cleanup in the session pool.
- Enforced 60k character limits for read_state_description and
action_results with clear truncation notices.
- Refined model timeouts (Gemini 45s, Groq 30s) and fixed
ModelProviderError initialization to preserve message and status_code.
<!-- End of auto-generated description by cubic. -->
