Cameron White

I would assume that if Agent Vellum Unpack doesn't see the agent having any clips, then the same would be true for any other crowd nodes... Have you inspected the 'agentclipcatalog' primitive intrinsic for the agent to see what clips it has in the agent definition?

I'd also suggest filing a bug with support if you have an example file where the Agent Clip SOP is slower that you would expect.

You can set Clip Assignment to None on the crowd state DOP, which will avoid assigning a new clip to the agent when it enters the state. However, the crowd solver will still keep adjusting the clip times etc to play back the clip, which sounds like it might not be what you want if you are trying to use custom poses from SOPs? In that case you can set the `crowdanimated` point attrib to 1 and specify the Target Geometry path on the crowd object DOP. In that scenario, the crowd solver will import the agents updated transforms from that path instead of actively solving the agent, but the agent can still turn into a partial ragdoll etc

I think that node will look at the `clipchannels` detail attribute (https://www.sidefx.com/docs/houdini18.5/nodes/sop/kinefx--characterblendshapes#attributes) when adding blendshape channels to the agent's rig.