# machine-readable domain observability
A governed surface for seeing what machines try to discover about an organization, then publishing and monitoring the metadata they need.
Agents now read a domain before humans do. What they find is scattered, stale, and unowned — instagrate-me makes it a product with an owner.
A domain's machine-readable truth leaks across /llms.txt, .well-known, OpenAPI specs, MCP endpoints, schema.org markup, and security.txt — published by different teams, drifting independently, and increasingly acted on by agents that don't ask twice.
Treat the machine-readable surface as a governed product: one place to observe what machines request, publish what they should find, and monitor the drift between the two.
surfaces
What you want models to know, in the place they now look first.
The protocol-level front door: policies, identities, capabilities.
The executable surface — what agents can actually call.
Names, claims, delegations, and the people allowed to change them.
mechanics
See what crawlers and agents actually request from your domain, including the 404s — demand is data.
Every published machine-readable artifact, with an owner, a version, and a review trail.
Alert when the published surface and the operational truth diverge — the machine-readable equivalent of a broken-link checker, but for facts.
Scoped authority over slices of the surface, so the API team owns the OpenAPI without owning the brand claims.
artifacts
instagrate-me.sudoscience.dev
what machines ask domains for, ranked
published surface vs. operational truth
edges
Started from watching agent traffic in server logs ask for files nobody owned.