The standard way to sinalize the ability to resize is with the operating system's own mouse pointer, the double-pointed arrow in the direction you are to resize (horizontal, vertical or diagonal). This happens when the mouse hovers the edge.
My application, a google chrome extension, consists of a right-handed panel that is inserted as a side bar into a given webpage. When the user hovers over it's edge a handle appears for the panel, and in this handle there is an icon for locking the panel in place, and clicking the handle hides/shows the panel.
So the edge is the only and thus a NECESSARY path between the web content and my app (which is the panel), since the edge separates the two. So given the user will be hovering over this edge multiple times, is it good interaction design to hide the handle and only show on hover? It would hide unnecessary elements most of the time, but also hide the affordance in a static situation.
As for resizing of the panel, should the handle as a whole allow resizing when dragged, or should I limit it to icons or visual areas in the handle? What visual cues would work best in this situation?
Should these visual cues light up on hover?
Should these visual cues change state when you have them 'grabbed'?
Also is it good interaction design to use a hand cursor (for 'grab') instead of the standard double-headed arrow cursor for resizing?