Mars Ind - Relay

Synthetic signals through the netcode

Status – Awaiting Steam’s final review for release scheduling. Build is complete and ready for launch.

Mars Ind Relay is the hardline between your AI backend and the outside world; a unified outbound tunnel that binds inference, TTS, image, video, and 3D workloads into a single, disciplined signal stream. It connects to every core you run, establishes a secure tunnel to your gateway, and pushes synthetic signals through the netcode with zero drift.

Relay tracks your inference host automatically, maintaining a live link even as processes restart, ports shift, or workloads migrate. If the gateway drops, Relay doesn’t. It keeps probing the line until the grid comes back online, restoring the tunnel without human intervention.

A built‑in API Key Firewall protects your gateway by enforcing key‑bound access at the relay layer. Every request is filtered, validated, and proxied through a hardened channel, ensuring that only authorized signals reach your site or application nodes.

Relay also performs automatic memory cleanup, clearing stale sessions, freeing residual RAM, and preventing runaway processes from degrading your AI stack. It’s built for continuous operation; the kind of reliability you need when your nodes behave like semi‑autonomous constructs, and your workloads never sleep.

Core Features

• Outbound tunnel to your gateway for all AI workloads
• Auto‑tracking of inference core host
• API Key Firewall with secure proxy routing
• Auto‑reconnect when the gateway goes down
• Automatic stale‑session cleanup and RAM recovery
• Unified handling of inference, TTS, image, video, and 3D pipelines
• Chaos‑resistant throughput for unstable networks

Relay is the signal spine of Mars Ind systems; the conduit that keeps synthetic workloads flowing, even when the grid fractures.