Is there one place an AI agent can see everything I offer it?
An AI catalog is a single index of everything your site publishes for AI agents: MCP servers, agent cards, skills and datasets. Instead of guessing where each one lives, an agent reads one file and follows the entries. We check whether you publish one, and whether it lists anything at all.
What it actually is
Think of the directory board in a building lobby. Each tenant still has its own office, but you read one board instead of trying every door. The AI catalog is that board for a domain, a small JSON file at a well-known address whose entries each point to a more detailed card elsewhere.
Why it matters
Agent-facing standards are multiplying, and every new one adds an address an agent has to try. A catalog turns that guessing into one lookup. It also fixes a real blind spot: a site can publish a perfectly good MCP server card under its catalog and still look like it has none to anything that only checks the older addresses.
What good looks like
A catalog at your domain that lists each agent-facing thing you actually run, with every entry pointing to a card that exists, and nothing listed that has been switched off.
Being honest about it
This is an Advanced check: it never fails a page and stays out of your headline score. The specification is a draft held in a temporary Linux Foundation repository while the MCP and A2A projects vote on adopting it, and only a handful of large developer platforms publish one today. Not having one is normal. We also treat an empty catalog as absent, because a valid file that lists nothing tells an agent nothing.
Where most sites go wrong
The common failure is the empty index: the file exists, passes validation, and has no entries, often generated by a tool before anything was ready to list. The other is a catalog that outlived what it describes, still pointing at servers or cards that now return an error.
Not sure where yours stands?
A free scan checks this and everything else in about 20 seconds — no signup. Or if it's already flagged and you'd rather it were simply handled, we do that too.
Flagged for this one? We handle it. No pitch, just a plain-English chat.
Book a call