An action cue is a cue given to the viewer which requires him/her to interact in some way with the target and report what happens. For example, the monitor might ask the viewer to "Mentally tap on the target and tell me what you hear."
It is important to understand that the monitor should only give an action cue after the viewer has identified a particular aspect of the target, or has provided information which would make the action possible and logical.
Example: Let's say that the viewer has said that there is some cubical thing at the site. The monitor can cue the viewer to touch it, thump on it, feel it, etc. But the monitor cannot suggest that the viewer "go inside" it until the viewer has identified it as hollow.
Action cues fall into two categories:
Normal action cues - where the viewer is suggested to interact with the target in some way, as in the example, above. The results of such commands are recorded in the session transcript, but the suggestion, itself, does not need to be.
Move commands - where the viewer is suggested to move in relation to the target. These are always recorded in the session transcript, so everyone can keep track of the viewer's position at the target and relationship with the target.
All action cues are given in the format of a "well formed-question to a psychic" (see definiton elsewhere). That is, the suggestion is made in two parts - a suggestion to the subconscious mind, followed by a present-tense request to the conscious mind.
See also, "Move Command" and "Well-formed question to a psychic".