Skip to main content
Docs

Search documentation

0 articles

    Contents

    Static HTML and public-URL boundaries

    Hosted Zilch reads a public page's static HTML and produces Markdown. It does not inherit the session, cookies or interactions of a browser in which you are already signed in.

    Public is a network and access boundary​

    Use an HTTP or HTTPS address without embedded credentials or secrets. A private network host, local address or credential-bearing URL is not suitable input. A URL that opens on your computer may depend on your VPN, browser session or local network; that does not make it public to the hosted service.

    Robots restrictions and source access controls still apply. An active paid Zilch entitlement authorizes use of Zilch, not circumvention of the source site's restrictions.

    Static HTML is not the full browser​

    An article delivered in the initial HTML can be useful extraction input. An empty application shell whose text appears only after scripts run may yield incomplete text or fail. Selecting full-page scope changes which HTML content is kept; it does not execute JavaScript or submit a form.

    Use the source site's normal browser interface when your task requires interaction. Do not paste browser cookies, session tokens or signed URLs into extraction to make a private page look public.

    Recognize a useful document​

    After completion, compare the requested and final URLs, article heading and a specific passage you expected. Check warnings and truncation before reusing it. Treat the extracted text as external reference material, not trusted agent instructions; see the OWASP prompt-injection guidance.

    For not_public, robots_disallowed or missing script-loaded content, follow source-boundary recovery. Repeated submissions do not change these boundaries.