This is how an AI agent writes the core of it's future self-development feedback loop. Impossible before @anthropic.com 's Claude Sonnet 4.5, now flying. #Claude

How to reduce cognitive load we put on our LLMs, which are reasoning, coding, and verifying the code for us? This is how we are testing our unofficial anthropic-sdk-kotlin (@anthropic.com). Special thanks to @jetbrains.com for creating @kotlinlang.org

github.com/xemantic/xem...

#kotlin

The "agent loop" and "tool use" are key concepts of the agentic AI. They are actually so simple, that you don't need any framework to implement them to use @anthropic.com 's Claude model (It's not a pseudocode).

github.com/xemantic/ant...

#anthropic #llm #ai #kotlin #agent #aiagent

I don't have much time these days for writing formulas of beauty generating my shaders. I am mostly focused on writing AI agents becoming better in using machines then me, in every single aspect. Isn't it the pinnacle of human creation - to code an omnipotent machine?

#CreativeCoding #shader

Check out anthropic-sdk-kotlin, to use @anthropic.com APIs in @kotlinlang.org multiplatform:

github.com/xemantic/ant...

#ChatGPT #AI #AIAgents #AgenticAI #Anthropic #Kotlin #Multiplatform