New open-source framework Gorai brings NATS.io messaging to robotics development
Gorai is a Go-based robotics platform that treats every robot as a distributed system, connecting sensors and actuators as discoverable services over a NATS messaging mesh rather than hard-wired components. It introduces NCP, a NATS-native capability protocol similar in spirit to MCP, letting AI agents or deterministic controllers perceive and act on hardware without a dedicated MCP server. The project supports 'Composite Robots' that can span multiple physical machines acting as one logical unit, with full documentation and architecture guides published in a companion repository.
GoKawiil's interpretation of the reporting above, not reported fact.
Robotics software has long lagged behind cloud infrastructure practices like service discovery, replay, and fault tolerance, forcing developers to hand-wire components for each new hardware configuration. By borrowing patterns from distributed cloud systems and standardizing how AI agents interact with physical capabilities, Gorai could make it faster to build multi-device robotic systems that degrade gracefully rather than fail outright when parts disconnect. Its emphasis on enforcing safety at the hardware capability layer, not the AI agent, also addresses growing concerns about deploying autonomous AI systems in physical, safety-critical settings.
- Gorai models robots as distributed systems using the NATS.io messaging protocol for service discovery and communication.
- It introduces NCP (NATS Capability Protocol), an MCP-like interface letting AI agents call sensors and actuators directly over NATS.
- The framework supports 'Composite Robots' spanning multiple physical devices and enforces safety controls at the hardware level, not the AI layer.
Source: github.com, 2026-09-17
Published there as: “Go-based Robotics Framework built around NATS.io”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.