Use a stealth Playwright browser to search 18+ search engines (Google, Bing, Baidu, DuckDuckGo, etc.) or crawl any URL. Invoke when the user asks you to search the web, look something up online, fetch a webpage, or when web search results would help answer a question.

Anytime Search Skill

Stealth browser search and web crawling tool using Playwright. Supports 18+ search engines with anti-bot detection evasion.

Setup (one-time)

cd D:/dev/github/anytime-search-skill
pip install -r requirements.txt
playwright install chromium

Search

python D:/dev/github/anytime-search-skill/search.py -q "<query>" [options]

Key options

Option	Default	Description
`-q` / `--query`	—	Search query (required for search mode)
`-e` / `--engine`	`google`	Engine name: `google`/`g`, `bing`/`b`, `baidu`, `duckduckgo`/`ddg`, `yahoo`, `yandex`, `brave`, `sogou`, `360`, `naver`, etc.
`-n` / `--num-results`	`10`	Max results to return
`--json`	off	Output as JSON array `[{title, url, snippet}]`
`--deep`	off	Crawl each result URL and return full page content instead of snippet
`--proxy`	—	Proxy URL, e.g. `http://127.0.0.1:7890` or `socks5://user:pass@host:port`
`--no-headless`	off	Show browser window (useful for manual CAPTCHA solving)

Examples

# Default Google search
python D:/dev/github/anytime-search-skill/search.py -q "Python asyncio tutorial"

# Use Bing, return 5 results as JSON
python D:/dev/github/anytime-search-skill/search.py -q "latest news" -e bing -n 5 --json

# Search Baidu (Chinese)
python D:/dev/github/anytime-search-skill/search.py -q "Python 异步编程" -e baidu

# Deep search: get full page content for each result
python D:/dev/github/anytime-search-skill/search.py -q "openai api docs" --deep

Crawl a URL

python D:/dev/github/anytime-search-skill/search.py -u <URL> [--wait-for "<CSS_SELECTOR>"]

Returns cleaned HTML body (scripts, styles, images removed). Use --wait-for for SPAs.

# Crawl a page
python D:/dev/github/anytime-search-skill/search.py -u https://example.com

# Wait for dynamic content to load
python D:/dev/github/anytime-search-skill/search.py -u https://spa-site.com --wait-for ".main-content"

Supported engines

Global: google, bing, duckduckgo, yahoo, yandex, ecosia, startpage, brave, ask, dogpile, searx China: baidu, sogou, 360, shenma East Asia: naver, yahoo_jp Privacy: metager, swisscows Russia: mail Shortcuts: g=google, b=bing, ddg=duckduckgo

Run python D:/dev/github/anytime-search-skill/search.py --list-engines for the full list.

Output format

Text (default):

[1] Page Title
    URL: https://...
    Snippet text...

JSON (--json):

[{"title": "...", "url": "https://...", "snippet": "..."}]

CAPTCHA handling

If Google returns a CAPTCHA, the script exits with code 2 and prints suggestions. Try:

Switch engine: -e bing or -e duckduckgo
Run with --no-headless to manually solve it
Run --clear-session to reset cookies and retry