GORAI is a Go-based robotics platform built around a NATS.io mesh that treats every sensor and actuator as a discoverable service. It argues that real robots are distributed systems and applies cloud-era disciplines - service discovery, location transparency, health checks, fan-out, and replay - to robotic architectures. Capabilities are exposed as resources (sensors) and tools (actuators) over a NATS Capability Protocol (NCP) with no central MCP server in the path. A single logical robot can span multiple machines (rover, drone, mast, compute box) as a composite system that composes at runtime and degrades gracefully as parts join or leave. Action logs, state streams, and replay are first-class platform concerns, and safety enforcement happens at capability nodes so agents can be used without being blindly trusted.
Implementation details emphasize practicality and deployability: robots compile to one static Go binary that embeds the mesh and selected components via blank imports (the CADDY model); components self-register with init() calls, and Go modules provide dependency management. A gorai CLI supports validation, iterative run, cross-compilation, component discovery/adding, and live mesh introspection. Non-Go workloads run as external services connected over NATS. Target users are software-first teams building prosumer field robots (marine, surface, underwater, land) who want deterministic, operable, and componentized autonomy on single-board Linux hosts.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.