The search engines and AI companies change their rules, crawlers and guidance quietly — a doc edited here, a bot token appearing there. We watch 19 first-party sources every day and publish the ledger below, dated and hash-verified. Nobody should have to take a vendor's word for "what changed". Not even ours.
Last sweep: pending — first sweep runs on the next site visit · Verified 0/19 · Unverified 0 (a failed fetch is never reported as "unchanged") · Raw ledger: JSON
Each source is fetched daily (browser-grade fetch, one vantage point). HTML is tag-stripped, JSON is key-sorted, whitespace is normalized, and the result is SHA-256 hashed. A hash change = "changed" — we report the fact and the size delta, and a human reads the diff before saying what it means. Registry and IP files (OpenAI's gptbot.json and friends) are where new crawler tokens usually appear first; schema.org and the llms.txt spec are watched because they version. Industry feeds are labeled as such — they are commentary, not first-party truth. Excluded after live verification on 2026-08-29: Perplexity (every page returns 403 to non-JavaScript readers — watched indirectly via registries), Yandex and Brave (no stable English doc URL that day). Exclusion is policy, not oversight; we re-test excluded sources quarterly.
Cite this freely. Journalists and researchers: the raw ledger is public, hash prefixes let you verify any snapshot, and we share data on request — hello@synapsein.com. Corrections are logged and dated, never silently edited.