AI assistant for writing/modifying scripts in Grasshopper (Kea plug-in)

Got it ,thank you sir

xAI can’t create a GH battery.

[17:28:48] EXCEPTION - Error in PopulateFromTemplate | ExceptionMessage=Value cannot be null. (Parameter ‘type’) ExceptionType=ArgumentNullException
at System.ArgumentNullException.Throw(String paramName)
at System.Activator.CreateInstance(Type type, Boolean nonPublic, Boolean wrapExceptions)
at Kea.ScriptComponentUtilities.<>c__DisplayClass10_0.b__0()

[17:28:48] EXCEPTION - Error in CreateOrUpdateScriptComponent | ExceptionMessage=Object reference not set to an instance of an object. ExceptionType=NullReferenceException
at Kea.ScriptComponentUtilities.CreateOrUpdateScriptComponent(GH_Document doc, String scriptCode, List1 newParamSpecs, String scriptDescription, ILogger logger, Nullable1& newInstanceGuid, String connString, Nullable`1 isRhino8Arg, ContentFormat cf, IGH_Component existingComponent, String nickName, String usingsCode, String additionalCode, Single x, Single y)

Try different models, especially the main ones you might find on the web chats, and they should return better quality answers that Kea can reliably process to create or update scripts.

Cheers for the log lines though, I’ll have a look into it.

Kea now supports Azure OpenAI - but this is unlocked per company, so if you work in a company that has Azure OpenAI deployments, send me the domain ID you see in the About pane of Kea and I’ll remotely unlock that for your company.

More details here: #grasshopper #computationaldesign #ai #azureopenai #plugin #generative #aec | Nicolaas Burgers

1 Like

can you explain why you’ve listed OpenAI’s gpt-5 codex or pro. Seems they should not work, at least i get

Thanks for flagging this - I’ll have a look into what I might need to add to Kea to handle these new models.

Hi @crz_06 I’ve just released a new version (0.0.11-alpha) that should fix the issue with OpenAI gpt-5xx models.

It also stores chat histories locally, too, by the way.

2 Likes

Thanks! Could you also comment on why it seems that the replies (the “thinking” phase) take (about five times) longer when using Kea compared to the ChatGPT web interface? Not meant as criticism — just feedback for UX flow. Sometimes the fastest way is to stop Kea, move to web and rerun question, copy the snippet from the browser, stop Kea while it’s still thinking, and paste the results into that window so it can create the node. Is it related to using specific models? Only tested openai gpt5 thus far