A time machine for your own website

Record your pages. Revisit any run. See what changed.

Your verified domains. History starts with your first run.

See it working

Real job runs of four public sites. Each picture opens the same view in the live demo.

Timeline of Hacker News job runs in the Sitehistory.fyi web app

Every job run of a page, in order

The Hacker News home page, desktop: 33 job runs on 10 UTC days, newest 2026-11-22 04:03 UTC. Every row opens its job run.

Job run 2026-11-22 04:03 UTC

Open this in the demo
Job run view of BBC News in the Sitehistory.fyi web app

One job run, with everything it stored

The BBC News home page as recorded in the job run of 2026-11-22 04:00 UTC, with the files stored from that visit and a hash for each.

Job run 2026-11-22 04:00 UTC

Open this in the demo
Network waterfall of the BBC News home page

The network of one page load

Every request of the BBC News home page of 2026-11-22 04:00 UTC as a waterfall, with size and timing per request and third parties marked.

Job run 2026-11-22 04:00 UTC

Open this in the demo
Comparison of two Wikipedia job runs

What changed between two job runs

The Wikipedia main page, desktop: the job run of 2026-11-21 22:08 UTC against the one of 2026-11-22 04:13 UTC.

Job run 2026-11-22 04:13 UTC

Open this in the demo
Answer card built from MCP tool results about BBC News

What an AI agent gets back

What an AI agent gets back from the MCP server: the newest job runs of BBC News and the latest change events, each with its job run IDs and hashes. Built from the data of the live demo.

Job run 2026-11-22 04:00 UTC

Open this in the demo

The demo is read only and shows public pages of four well-known sites, recorded by us for demonstration; customers record only domains they verify. It records daily at 04:00 UTC, plus 10:07, 16:07 and 22:07 UTC. The demo has no job runs yet.

What you get

Revisit any moment

Screenshots, HTML and network from the same run.

See what changed

Choose two runs. Compare the evidence.

Keep the full record

Desktop and smartphone, on your schedule.

One property. Every page. A history.

  1. Web property

    bbc.com

  2. Job

    bbc.com/news

    Desktop, banner accepted

  3. Job runs

    Every run keeps the screenshot, the HTML, the network log and a SHA-256 hash for each file.

A property holds jobs. A job is one page on one device. Each run of a job is one recording.

For AI agents, too.

Ask your website's history.

  • MCP
  • CLI
  • REST API

Try it with your own agent, no sign-up

A shared read-only token for the public demo. One line connects Claude Code; snippets for Cursor, Claude Desktop and curl are on the demo page.

claude mcp add --transport http sitehistory https://staging.sitehistory.fyi/mcp --header "Authorization: Bearer 755043e670885d20b21b5539b6dc49265a67cb63bc3ba556"

Start free. Build your history.

Free

EUR 0 / month

  • 1 URL
  • Daily runs
  • 16 months

A verified domain is required. You can track only domains you verify as yours; it is not a competitor watcher.