Add browser-use integration: browser_use_agent and BrowserUseTools

[wu-changxing]

Summary

• browser_use_agent(task) — AI-driven browser automation using ConnectOnion's LLM proxy (co/gemini-2.5-flash by default). Pass any co/* model via the model param.
• BrowserUseTools — auto-generates ALL browser-use registry actions as ConnectOnion tools (navigate, click, scroll, screenshot, search, evaluate JS, etc.). New browser-use actions appear automatically without code changes.
• Watchdog events (downloads, browser errors) are bridged to the agent via get_events() queue.

Usage

# Simple: one-shot task
from connectonion import browser_use_agent
result = browser_use_agent("Go to github.com and get the trending repos")

# Advanced: agent controls individual browser actions
from connectonion.useful_tools import BrowserUseTools
browser = BrowserUseTools()
agent = Agent("researcher", tools=[browser])
browser.close()

Test plan

• browser_use_agent("search for connectonion on google") returns a result
• BrowserUseTools().navigate("https://example.com") navigates successfully
• BrowserUseTools().get_content() returns page text with [N] indices
• BrowserUseTools().close() cleans up Chrome process