liquidos-ai/AutoAgents
A multi-agent framework written in Rust that enables you to build, deploy, and coordinate multiple intelligent agents
What it solves
AutoAgents provides a production-grade framework for building multi-agent systems. It addresses the need for a high-performance, type-safe environment where multiple AI agents can coordinate, use tools, and maintain memory across different LLM backends, whether hosted in the cloud or running locally on the edge.
How it works
Built in Rust, the framework uses a modular architecture that decouples the agent's reasoning logic from the underlying model. It employs a type-safe agent model with structured tool calling and a pluggable system for memory and LLM providers. It supports various execution patterns like ReAct and Chain-of-Thought, and includes a sandboxed WASM runtime for executing tools securely.
Who it’s for
It is designed for developers who need to build scalable, safe, and performant AI agents that can run on both servers and edge devices, and those who prefer the safety and speed of Rust (though Python bindings are provided).
Highlights
- Broad LLM Support: Unified interface for cloud providers (OpenAI, Anthropic, Google, etc.) and local runtimes (Ollama, LlamaCpp, Mistral-rs).
- Production Tooling: Includes derive macros for structured tool inputs/outputs and a WASM sandbox for tool execution.
- Agent Orchestration: Features typed pub/sub communication for multi-agent coordination.
- Built-in Guardrails: Integrated LLM guardrails for input/output sanitization and auditing.
- Multimodal Capabilities: Support for vision, multimodal inputs, and local TTS/STT speech processing.
- Observability: OpenTelemetry integration for tracing and metrics.