Privacy & security
What happens to a page you extract
Section titled “What happens to a page you extract”- Sent to Analog: Analog receives the URL and rendered page content for extraction. The structured records come back.
- Stored on your machine: the records and the page’s markdown, under a local handle. Raw HTML is never written to disk.
- Not logged: Extraction URLs are never logged. What you read
through Analog is your business. A quality report includes the URL
you explicitly submit; it never includes page content. See
analog feedback. mode="local"sends nothing — fetch and markdown conversion happen entirely on your machine; no extraction, no backend call, no account needed.
Analog fetches as an unauthenticated visitor by default: it sees what
any visitor sees. If you hand it authenticated content (your own
html= or a logged-in custom fetcher), that content is sent to the
extraction API just the same — point Analog at content that is
acceptable to send, or use mode="local".
Credential storage and scope
Section titled “Credential storage and scope”Your API key lives in an owner-only local file — the
~/.aws/credentials posture. Any process running as you can read it,
including the agent driving your session. A leaked key is bounded by
server-side revocation
(analog logout), origin binding (the key refuses to travel to
any host but the one it was minted against), and narrow scope
(extraction and feedback only).
Fetching identity and boundaries
Section titled “Fetching identity and boundaries”Analog’s default fetcher is a local headless browser with a stable
identity that respects robots.txt — including for the fit check’s
single feed-verification request. No proxies, no fingerprint spoofing,
no CAPTCHA solving, ever. A site declining automated visitors is a
result Analog reports, not a problem it engineers around.
Public web addresses only, by default
Section titled “Public web addresses only, by default”get and assess decline local and private-network hosts and
non-web schemes like file: — on the initial URL, on every redirect,
and on every pagination hop — so a page can’t steer a fetch at
localhost, an internal address, or the local disk. Reaching a dev
server or intranet host on purpose is an explicit opt-in:
--allow-private (allow_private=True). Paginated sweeps follow
only links on the page’s own host.