Over here I asked Claude in its own editor to connect to Rhino 9 and execute exercises from a training manual about ship hull modeling that I wrote many years ago.
I get this feedback:
I’m stopping here because the Rhino window state has become unclear, and I don’t want to make changes blind.
What happened
Rhino 9 accepted the access approval. I turned on Record History and started the regular Loft command, and the Loft Options dialog opened with Style set to Normal.
The desktop shell then took focus, so my clicks on the Style dropdown were refused. Bringing Rhino 9 forward with the app-open tool made things worse: a Rhino start screen (“Rhinoceros”) now sits over a blank window. It’s a separate window from the working session. My scripting connection still reports the working document, with 408 objects and the nine curves selected, so that session is still alive.
The Loft command I sent timed out. It is probably still waiting on the Loft Options dialog in the working session, which is hidden behind the start screen.
Yesterday, Claude opened constantly new Rhino instances to continue after asking me questions. Then it was not able to continue in the instance it was already working in.
Also I still have not solved connecting the Rhino internal AI editor to Claude again. It worked for a while, although I did nothing with it, and now I am not succesful connecting it again.
You’re trying to use Rhino 9 from outside via the MCP. Can you share the prompts you used with Claude and what you were trying to do? Did Claude tell you any errors it came across?
What worked, and then what stopped working? Did it stop replying or could it not be found?
I had one issue with Rhino 9 that was not happening with Rhino 8
I asked my DeepSeek agent to model a simple house and send me screenshots via Telegram while I was away, when I did that using Rhino 8 it was able to keep improving no problem. Deleted a body and remade it with the adjustments and just kept going.
however, in Rhino 9 it started using python scripts instead of commands and saving each file with modifications after it, at the end I had 15 3dm files with the improvements of each iteration and a bunch of .py files saved on the same folder.
The agent is the same, there was no update on it or anything, the only difference relies on the Rhino MCP skills, it reported to me that my R8 version has 32 skills while the R9 version has 44 skills.
Could the skill update changed something by accident or something like that?