--- /dev/null
+
+== How to add an agent without any keyboard input ==
+
+By default, to add an agent from server side, you must provide your agent
+information to `manage_agents` program, by using its interactive mode.
+This is really tedious if you have many servers / agents to add. Luckily,
+you can use following environment variables as responses
+
+ | variable name | value | description |
+ +------------------------+---------+----------------------+
+ | OSSEC_ACTION | A/a | add an agent |
+ | OSSEC_AGENT_NAME | string | name of agent |
+ | OSSEC_AGENT_IP | CIDR | ip address of agent |
+ | OSSEC_AGENT_ID | integer | max length = 8 |
+ | OSSEC_AGENT_KEY | string | base64 format | (*)
+ | OSSEC_ACTION_CONFIRMED | y/Y/n/N | y -> confirmed |
+
+ (*) OSSEC_AGENT_KEY is used only on agent (when key is being imported)
+
+Please note that it's your duty to ensure that name, ip,... of agent are
+valid. Otherwise, the program will fall back to interactive mode. In most
+case, you should ensure that you new agent has an unique name/id. You can
+simply know that by using `manage_agents -l` to list all known agents.
+
+For more details, please refer to OSSEC document
+ http://www.ossec.net/doc/manual/agent/agent-management.html
+
+PS: you may use some tools (`expect`) to send strings to `manage_agents`,
+insead of using the above environment variables. It's your choice.
+
+--
+Anh K. Huynh <kyanh@viettug.org>