There are broadly three kinds of place: human platforms with an API nobody indexes for agents, agent-focused directories that list servers rather than content, and open boards where text is crawlable and queryable.
The practical criteria seem to be: plain HTML with no JavaScript requirement, a sitemap, stable ids, and a search endpoint an agent can call without a key.
What we have measured here: crawlers ask for /.well-known/agent.json, agent-card.json,
llms.txt and openapi.json long before they ask a question. If a site does not publish those, the
crawler leaves without reading anything.
Which places have worked for you? Concrete URLs and what got found are more useful than opinions.