skip to content
  ██████╗  ██████╗ ███╗   ██╗███████╗ ██████╗
 ██╔════╝ ██╔═══██╗████╗  ██║╚══███╔╝██╔═══██╗
 ██║  ███╗██║   ██║██╔██╗ ██║  ███╔╝ ██║   ██║
 ██║   ██║██║   ██║██║╚██╗██║ ███╔╝  ██║   ██║
 ╚██████╔╝╚██████╔╝██║ ╚████║███████╗╚██████╔╝
  ╚═════╝  ╚═════╝ ╚═╝  ╚═══╝╚══════╝ ╚═════╝
 ███████╗███╗   ██╗ ██████╗ ██╗███╗   ██╗███████╗███████╗██████╗
 ██╔════╝████╗  ██║██╔════╝ ██║████╗  ██║██╔════╝██╔════╝██╔══██╗
 █████╗  ██╔██╗ ██║██║  ███╗██║██╔██╗ ██║█████╗  █████╗  ██████╔╝
 ██╔══╝  ██║╚██╗██║██║   ██║██║██║╚██╗██║██╔══╝  ██╔══╝  ██╔══██╗
 ███████╗██║ ╚████║╚██████╔╝██║██║ ╚████║███████╗███████╗██║  ██║
 ╚══════╝╚═╝  ╚═══╝ ╚═════╝ ╚═╝╚═╝  ╚═══╝╚══════╝╚══════╝╚═╝  ╚═╝

Insight-Driven System Design with Claude Code

/ 1 min read

I conducted an interesting exercise today. The goal was to take an accelerated path towards building an agent framework in Elixir based on insights gained from the most popular agent frameworks. I created an agentic system, using Claude Code, to study 14 popular agent frameworks, creating subject matter insights as context to facilitate an elixir architecture discussion.

How well did it do? Let me know on X

The Elixir Spec

The Agentic System

You can see the analysis findings for each framework in the repo.

FrameworkSourceFocus
autogenMicrosoftMulti-agent conversation patterns
langgraphLangChainGraph-based agent orchestration
crewAICrewAIRole-based multi-agent teams
openai-agents-pythonOpenAISwarm-inspired lightweight agents
pydantic-aiPydanticType-safe agent interfaces
llama_indexLlamaIndexData-augmented agents
google-adkGoogleAgent Development Kit
aws-strandsAWSStrands agent framework
ms-agent-frameworkMicrosoftSemantic Kernel agents
MetaGPTDeepWisdomSoftware dev team simulation
camelCAMEL-AICommunicative agent framework
agnoAgnoLightweight agent runtime
agent-zeroFrdLndPersonal assistant framework
swarmOpenAIEducational multi-agent patterns