If your server publishes room directory over federation and has public rooms within the directory,

they will appear in the search index after the next full reindex process (should be run daily).

Please keep in mind that to have a room indexed, you have to:

1. Explicitly mark a room federatable when creating it
2. Explicitly mark a room as public in room settings
3. Explicitly publish a room in the room directory
4. Explicitly publish your room catalog over federation

When you do all of those steps, you clearly understand the consequences of your decisions, i.e. a particular room that was made public, federatable, published in room catalog and then shared room catalog over federation will be accessed over federation.

MRS can't index a room if you didn't explicitly allow a room to be visible over federation. You either publish something over federation, or not. MRS is not a special thing, it uses the same API and the same set of rules as any other matrix server does.

unpublish your room directory from the federation

If your server is indexed, that means you explicitly published your public rooms directory over federation. If you don't like that the information you explicitly published over federation is accessed over federation, you should consider unpublishing it.

For synapse, you need to add the following config options in the homeserver.yaml:

allow_public_rooms_over_federation: false

in case of etke.cc/ansible and mdad, add the following to your vars.yml:

matrix_synapse_allow_public_rooms_over_federation: false

However, if you think that "publishing over federation, but not for that particular member of the federation" segregation is a good thing for Matrix protocol, MRS has several options to unlist/unpublish/block/remove your server and its rooms from indexing.

(specific room) include special string in room topic

You could add the following string into your room topic/description to prevent it from indexing (exact match):

matrixrooms.info: noindex

(deprecated) robots.txt

While robots.txt is not part of the Matrix protocol, that's the easiest and convenient way to prevent your server from indexing. MRS supports several made-up endpoints to prevent indexing your server.

NOTE: due to the fact robots.txt is not part of the Matrix protocol, MRS treats it a bit differently than usual bots. User-agent: MRSBot MUST be present in your robots.txt file to make it work (User-agent: * will be ignored!)

User-agent: MRSBot
Disallow: /_matrix/federation/v1/publicRooms

User-agent: MRSBot
Disallow: /_matrix/federation/v1/publicRooms/!yourRoomID:example.com