It won't work out-of-the-box. If you include the spring-boot mongodb starter project, it will look for the property 'spring.data.mongodb.uri' to connect to a single database, and if it does not find it, it will try to connect to 'localhost:27017'. This single database will then be used for all Spring Data Repositories automatically.

You can add extra MongoClient beans for different databases, but you'll need some work to connect different Spring Data Repositories to those different beans.

And if you want to work with a dynamic set of databases, i.e. when you don't know which databases or how many, you can't work with fixed MongoClients as spring beans anyway. You'll need some sort of factory that creates multiple MongoClient beans based on the number of databases, and multiple SearchRepository beans connected to each of those MongoClient beans.

In depends on the complexity of your project, but for this usecase I would not use Spring Data at all, but stick to the MongoDB Java client API:

Map<String, MongoClient> clientsPerDatabase = new HashMap<>();

// Setup
MongoClient mongoClient = new MongoClient(/* default database */);
MongoCursor<String> dbsCursor = mongoClient.listDatabaseNames().iterator();
while(dbsCursor.hasNext()) {
  String databaseName = dbsCursor.next();
  if (!"admin".equals(databaseName)) {
    clientsPerDatabase.put(databaseName, new MongoClient(...+databaseName);
  }
}

// Query
for(MongoClient client: clientsPerDatabase.values()){
    ...   
}

I'd configure the Spring Boot application with admin access to the database and then use native queries to retrieve info on all existing databases, schemas, etc