modify requests after routing before sent to agents

[cash]

The motivation: a system developer can assign an arbitrary trigger word to an agent. That could break the parsing of the request. We need a way to strip the trigger word from the request or provide another mechanism to remove it.

Ideas:

• filter after routing (just like we have before routing). Then the filter has a list of trigger words to remove.
• modify the request in the router
• configure the agent with a trigger word