Python SDK

pip install webx-sdk

Zero dependencies — stdlib urllib only. Works in sandboxes, lambdas, and anywhere you can’t install heavy clients. Import as webx_sdk:

from webx_sdk import WebX

wx = WebX("http://localhost:8080")           # api_key="webx_..." if the server requires one

Scrape

doc = wx.scrape("https://example.com")
print(doc["data"]["markdown"])

# render + act on JS pages
doc = wx.scrape("https://spa.example.com", render=True,
                actions="click:.accept | scroll | screenshot",
                formats=["markdown", "a11y"])

Search

res = wx.search("go generics tutorial", limit=5)

Crawl (async job — polls until done)

job = wx.crawl("https://docs.example.com", goal="authentication docs",
               limit=20, semantic=True)
for page in job.data:
    print(page["url"])

Extraction — deterministic CSS or LLM schema

data = wx.extract("https://shop.example.com/p/1",
                  css={"title": "h1", "price": ".price"})

Tenancy (admin key)

key = wx.create_key("agent-1", rpm=120, monthly_units=5000)
print(key["key"])          # webx_... — shown once
print(wx.usage(days=7))

wx.schedule("nightly docs crawl", "0 3 * * *", "crawl",
            {"url": "https://docs.example.com", "limit": 50},
            webhook_url="https://me.example/hook")

Errors

Raise WebXError with .status, .code (payment_required, quota_exceeded, concurrency_limit, …), and .payment metadata for 402-gated targets — upgrade paths are data, not dead ends.

Source: sdk/python/ — PyPI: webx-sdk.