Changelog

What shipped, and when. The dates are the point. For what the collectors are doing right now, see status.

  1. New

    Public status page and changelog

    Every source now reports what it collected and when, including the ones that are built but idle and the ones not built yet. Freshness is judged on how much of the list a source actually read, so a single domain checked by hand cannot make a stopped pipeline look current.

  2. Changed

    Types, tests and a production build run on every push

    The two test suites already existed and passed. Nothing ran them automatically, so nothing stopped them rotting. The build is included because it is the only check that catches a bad import reaching the edge bundle.

  3. New

    Raw upstream responses are archived

    Every reading now keeps the exact bytes the source returned, gzipped to object storage beside the parsed number. A parse we later find to be wrong can be redone from the original. The archive can never fail a reading: if storage is down, the reading still lands and only the copy is lost.

  4. Security

    Outbound fetches are checked against internal addresses

    Any source that fetches a host you name now resolves it first and refuses loopback, private, link-local and cloud metadata addresses, on the first request and on every redirect after it. The public lookup reaches a TLS handshake with no account, which made this the sharpest edge on the site.

  5. New

    MCP server, so agents can read your domains

    Nine tools over a bearer key, covering the domain list, current metrics and history. Keys are hashed, scoped to one account, and an untracked domain reports as untracked rather than as missing.

  6. New

    AI crawler policy tracking

    One read of robots.txt per domain per day, parsed for the eight crawlers that matter to generative search, plus a check for llms.txt. Each agent records allowed, blocked or unspecified, because a file that never names a crawler can still block it with a catch-all.

  7. Fixed

    Popularity rank updates daily instead of freezing

    Rank was read once when a domain was added and never again, so every series was flat from its add date. Ranks now come from the daily list, which also means one download covers every domain instead of one request each.

  8. New

    Accounts, dashboard and domain grouping

    Sign in, add domains without a cap, and group them with tags. Tags rather than folders, because a domain can be both a competitor and something you own, and a folder forces one home.

  9. New

    Enrichment beyond Domain Rating

    Adding a domain now reads every cheap source at once: rating, popularity rank, registration date, certificate, DNS. Previously it read the rating alone and left the rest blank.

  10. New

    First release

    Domain Rating tracking with history, an embeddable badge, and a public lookup that needs no account.