beeai-framework

Why we included this project

Teams that want one toolkit for building and running agents in both Python and TypeScript get that here: the two libraries share a design, so moving between languages does not mean relearning concepts. It covers more than a single agent loop. A workflow engine handles multi-agent systems with branching flows, and the built-in tools for web search, weather, and code execution can be extended with your own. Production details are not left as an afterthought either. Memory strategies, caching, event-based observability and logging, serializing agent state so it survives restarts, and serving agents over MCP and A2A all come built in, which is what lets a project go from prototype to a hosted, observable service without stitching several tools together yourself.