Example and readme is very confusing

[whynames]

Hey, your repo seems very interesting, but very confusing. Could you only explain to me how AIAgent can be used independently of your sample, what is the exact difference between langchain and non-langchain version?
Thanks.

[IhateCreatingUserNames2]

Hello, yes, its a mess, im lazy =/

Well, the NOLangGraph Version, it just a framework that controls AI Agents, the system creates them, controls memory for each one of them.

The LangGraph Version, it's the same framework, except that instead of sending prompts directly to the LLM,
It goes thru first langGraph and comes back.
It's just a sample in LangGraph Integration.

The difference is that noLangGraph is a very basica Ai agent framework.
While LangGraph is a framework itself too, that can be used to create complex agents with CoT, Memory and etc.
IF you wanna simple bots that will respond to Prompts without any trigger or anything, go without lang graph
But if you wanna build complex bots with Cot, Triggers(trigger animation, actions, scripts...) u probably will be better with LangGraph which is a framework specially made to make AI Agents.

[whynames]

Thanks. So is Langchain version fully functional now? Are these triggered animations, actions, scripts are typescript or unity c# functions? Would you say it is usable in production code, or needs a lot of work?

[IhateCreatingUserNames2]

Needs lots of Work. But can be used. Depends on what is your objective to know how much Work it will need.
It needs 'polishing'

In this integration, i used PuerTS to run LangGraphJS thru a webpack.

You can call Any function using triggers , there is an sample of a LangGraph code with a 'dance' trigger in the sample demo

[IhateCreatingUserNames2]

I have asked claude to remake the ReadMe , i think now it is way better...
https://github.com/IhateCreatingUserNames2/MaiMai/tree/main?tab=readme-ov-file