RelayCloud
Get started
AI Infrastructure Platform

Remote AI
Agents.
Local Feel.

Deploy OpenCode on powerful remote servers, manage all your AI assistants from one dashboard, and connect securely through our Relay—your existing local clients just work, no VPN needed.

Your Machine any local client 127.0.0.1:4090 RelayCloud Secure Relay Outbound Only opencode #1 opencode #2 + provision... Remote Server Auto-provisioned request response routed reply Your tools, unchanged Zero exposed ports Powerful hardware

How it works

Three steps to
remote AI power.

01

Provision a server

Pick a region, click deploy. RelayCloud spins up a VPS, installs the runtime, and launches your OpenCode instance automatically—no SSH required.

02

Relay connects securely

Your server opens a persistent outbound WebSocket to our Relay. No inbound firewall rules, no open ports, no VPN.

03

Work locally, as always

RelayCloud runs a local proxy on your machine. Point any OpenCode-compatible client at 127.0.0.1:4090—it tunnels through the Relay transparently.

Automated Provisioning

Spin up servers.
Not headaches.

One command provisions a cloud VPS, installs all dependencies, deploys OpenCode, and registers it with the Relay. Your assistant is online in under two minutes.

  • Choose region, size, and runtime version
  • Automatic updates and health monitoring
  • Multiple instances across different regions
Terminal

~ $ relaycloud server provision --region us-east --auto

Allocating VPS ............... 2.1s

Installing runtime ............. 9.3s

Deploying OpenCode ............. 12.1s

Registering relay .............. 1.4s

✓ Server prod-01 is ready.

Connect via 127.0.0.1:4090

~ $

Secure Relay & Management

Built secure
from the ground up.

Your remote instances connect outbound to our Relay—no inbound ports, no firewall rules, no VPN. Manage all your OpenCode workspaces from a single dashboard: status, logs, usage, and configuration.

  • Outbound-only connections—nothing listens on the internet
  • No firewall changes or open ports required on your server
  • Unified dashboard: health, logs, sessions, API tokens
Relay Instances 3 online
Instance Region Status Uptime
prod-01 us-east-1 Online 14d 6h
dev-02 eu-west-1 Online 3d 12h
staging-01 ap-south-1 Idle 7d 0h

Local Proxy & Onboarding

Keep your tools.
Lose the limits.

Already running OpenCode locally? Onboard it in minutes. The local proxy exposes your remote instances over the Relay so any client connecting to 127.0.0.1:4090 is actually talking to your remote agent.

  • Onboard existing local OpenCode instances in one step
  • Zero changes to your editor or IDE configuration
  • Switch between local and remote instances instantly
Local proxy

~ $ relaycloud proxy serve

Listening on http://127.0.0.1:4090

Onboard existing instance

~ $ relaycloud onboard --detect

Found OpenCode at 127.0.0.1:4096

Instance registered as local-01

Conversation Sync

Pick up where
you left off.

Every conversation is synced to your account in real time. Switch devices, swap instances, or move to a different server—your full history is always there, exactly as you left it.

  • Access any conversation from any device, any time
  • History persists across server replacements and reprovisioning
  • Not tied to any single instance—your work belongs to you
Conversations Synced
P

Refactor auth middleware

prod-01 · us-east-1 · 2h ago

24 msgs
D

Debug flaky CI pipeline

dev-02 · eu-west-1 · Yesterday

11 msgs
L

Write OpenAPI spec for v2

local-01 · onboarded · 3 days ago

38 msgs

Organization Access

Your team's work,
shared by default.

Every OpenCode instance in your organization is accessible to your teammates. Reference a colleague's conversation for context, or pick up a task exactly where they left off—without any handoff.

  • Access any instance in your organization, not just your own
  • Browse and continue conversations started by teammates
  • Shared context means less repeated work across the team
Organization Instances acme-corp
JL

Jamie's workspace

prod-01 · Active now

Online
MR

Morgan's workspace

dev-02 · 12 min ago

Idle
AK

Alex's workspace

staging-01 · 2h ago

Idle

MCP Server

Agents that talk
to each other.

RelayCloud provides a MCP server that any compatible client can connect to. Use the prompt-assistant tool to start a new conversation in any workspace or continue an existing one—with full control over model, agent, and run variant.

  • Target any workspace by ID to open a new conversation
  • Or pass a conversation ID to continue an existing thread
  • Configure agent, model provider, model, and run variant per call
RelayCloud MCP · prompt-assistant

# Start a new conversation in a workspace

prompt-assistant({

workspace_id: "wsp_a1b2c3",

prompt: "Review PR #142 and summarize

the security implications."

agent: "build",

model_provider_id: "OpenAI",

model_id: "gpt-5.5",

variant: "high",

})

# Or continue an existing conversation

prompt-assistant({

conversation_id: "conv_x9y8z7",

prompt: "What was your conclusion?",

})

Your AI agents.
Ready when you are.

Get started in minutes. Provision your first remote server, onboard existing instances, or just explore the dashboard.