Yore builds a timeline of a website. You choose the pages, the languages and the years. Yore asks the Internet Archive for every capture it holds of those pages, fetches each distinct one once, and shows what the page said at each moment and what changed between two of them, word by word.
Everything Yore shows rests on a capture you can open yourself: the archive's own address for it, its time in UTC, the status the site answered with, and the SHA-256 of the bytes Yore fetched.
What you do with it
- Make a plan for a website: its paths, its languages and a window of time.
- Read a page as versions, compare two moments, and trace one sentence from the capture that first shows it to the capture that first lacks it.
- See which pages changed, appeared or were removed, month by month.
- Take it out as an export anyone can check without asking Yore, as a printed exhibit, or as a read-only link.
- Watch a page, so a change from today on is captured at the next daily check that sees it.
The site, the MCP server and the yore command give the same answers: all three ask one API.
Where the past comes from
The past comes from the Internet Archive's Wayback Machine, and from nowhere else. Yore reads the archive's index to learn which captures exist, and fetches the raw bytes of a capture, without the archive's banner or rewritten links. It names itself in every request and keeps to the archive's pace: one request every 1,5 seconds, shared by every plan.
From the day you watch a page, Yore also checks the page itself once a day and keeps what it answers.
What Yore is not
- Not a law firm. Yore states what the captures show. It gives no legal advice and draws no conclusion about what a change means.
- Not a way past a door. Yore reads public pages. It doesn't sign in anywhere, doesn't pass a paywall, and doesn't get around bot protection. A site that refuses Yore's own request is recorded as blocked.
- Not a replay. Yore keeps a page's HTML and the images in its content. It doesn't keep stylesheets, scripts or fonts, so it shows what a page said, not how it looked. The capture's address in the web archive shows that.
- Not a guess. The text of a page is derived from its bytes by fixed rules. No model reads or rewrites it, so the same bytes always give the same text.
- Not a continuous record. The archive visits a page when it visits. Between two captures nobody looked, and Yore never claims a page said something on a day without a capture.