Show HN: Maki, an open-source multi-agent LLM framework (local or hosted) (github.com)

Maki, an innovative open-source framework for developing multi-agent large language model (LLM) applications, has been unveiled. This Python-based framework allows developers to utilize multiple LLM backends, including those from OpenAI, Anthropic, and HuggingFace, and features a rich plugin system comprising 16 built-in tools. Notably, Maki incorporates a workflow engine that supports dependency resolution and parallel execution, enabling efficient orchestration of agents through a distributed layer that serves agents over HTTP. The significance of Maki lies in its modular design, facilitating the creation of complex LLM applications with ease. It includes a robust agent system capable of memory management, reasoning, and collaborative task execution, allowing for seamless multi-agent interactions. The inclusion of a fail-safe plugin security model and detailed logging mechanisms enhances reliability and safety, critical aspects for AI/ML developers. This framework not only accelerates development but also caters to a broad range of use cases, from web scraping to trading strategies, positioning Maki as a valuable tool for the evolving AI and ML landscape.