robustmq/robustmq

What it solves

RobustMQ provides a unified messaging engine that eliminates the need to deploy multiple separate brokers for different protocols. It specifically addresses the fragmentation in multi-agent AI systems, where agents often struggle to find each other (discovery) and communicate reliably across different states of availability (asynchronous messaging).

How it works

Built in Rust, the system uses a single binary and a shared storage layer. This allows a message written via one protocol (e.g., MQTT) to be consumed via another (e.g., Kafka). For AI agents, it introduces the mq9 protocol, which provides a built-in registry for semantic discovery (finding agents by capability) and persistent mailboxes that store messages until the recipient agent comes online, supporting three levels of priority.

Who it’s for

• AI Developers: Those building multi-agent systems who need a standardized way for agents to register, discover, and communicate asynchronously.
• IoT and Edge Engineers: Developers needing a high-performance, low-footprint broker that works from edge devices to cloud clusters.
• Enterprise Architects: Teams looking to consolidate multiple messaging protocols (MQTT, Kafka, NATS, AMQP) into a single infrastructure layer.

Highlights

• Multi-Protocol Unity: Native support for MQTT, Kafka, NATS, AMQP, and mq9 sharing one storage engine.
• mq9 Agent Infrastructure: Dedicated tools for AI agents, including semantic vector search for discovery and priority-based mailboxes.
• Rust-Native Performance: Predictable memory footprint with no garbage collection, suitable for edge-to-cloud deployment.
• Zero External Dependencies: A single binary with built-in Raft consensus for simplified operations.
• Flexible Storage: Supports Memory, RocksDB, and File backends with automatic cold data tiering to S3.