While reviewing documentation, I found that the Reinforcement Learning Section is written to accommodate Eagle, which is now decommissioned. At a glance, all of the documentation appears like it would work on Kestrel, but it might be best to remove mentions of Eagle.