Show HN: GraphFlow – A lightweight Rust framework for multi-agent orchestration

github.com

8 points by alonagmon a day ago

It all started with a conversation among friends about limitations in current multi-agent orchestration frameworks. We faced issues like limited control over agent memory and state, complicated persistence, scaling problems, and lack of type safety in Python-based tools. These challenges inspired us to try something different. The result was GraphFlow, a Rust-based lean framework for orchestrating multi-agent workflows that's simple, scalable, and robust. Its key features include: Graph-based orchestration: Easily define workflows using nodes and edges. Lean Execution Engine: A minimal and efficient graph executor / state machine implementation. Clear Memory Management: Direct and transparent handling of agent states. Simple DB Schema: Easy-to-understand schema for persistence and state tracking. High Performance: Native Rust performance with low overhead and easy scaling. Type Safety: Rust's type system reduces runtime errors. GraphFlow is open-source ofc and aims to solve real-world problems we've experienced firsthand. We'd love your feedback!

0rzech 6 hours ago

Great initiative!

A quick web search revealed at least 4 other agent orchestration frameworks written in Rust [1][2][3][4]. Just being curious: did you decide against contributing to existing ones for some reason?

[1] https://github.com/kevin-rs/autogpt

[2] https://github.com/The-Swarm-Corporation/swarms-rs

[3] https://github.com/M4n5ter/rigs

[4] https://crates.io/crates/kowalski (for some reason GitHub link gives 404 and seems to be Ollama-only anyways)

  • alonagmon 4 hours ago

    Thanks for the comment! Yes, the idea was to create something much leaner and simpler that is not tightly coupled to any framework, yet provides robust graph execution capabilities. In fact, for this reason, GraphFlow can server as a graph execution framework for Rust application that are not AI oriented at all.