There is no formal documentation on the Agentless method of Foglight implementation as the term "Agentless" actually refers to Remote Monitoring of the monitored host in question.
Currently, the Agentless method (Remote Monitoring) is a feature of both the new DB2, SQLServer, and Oracle database cartridges.
Basically, the term Agentless (remote monitoring as oppose to local monitoring) means the ability to monitor a database without the agent process (FglAM) being installed and running on the database monitored host.
This means that an FglAM and DB agent will still have to be deployed to monitor the databases, but you can choose where the FglAM and DB agent is deployed to (ie which box you want to use to monitor the database). In this way, the actual DB server is not touched by adding the Foglight Middleware (FglAM) or agents.
The FglAM would be installed as per our current documentation for FglAM.
The DB agents would be deployed (potentially remotely) as per the DB2, Oracle, and SQL Sever guides.
In summary, Foglight for Oracle and Foglight for SQL Server are 100% remote database monitoring solutions - they are not agentless. In Foglight, whenever you monitor something, an agent is created; for those who were accustomed to previous database monitoring cartridges, a local agent was required on the monitored database server, so one of the more attractive aspects of the new cartridges is that they are 100% remote.
The advantage is that you do not have to put another process on the DB boxes themselves.
In the future, it is expected that other agents will have the full ability to do remote monitoring.