You run agents for other people. Run them from one place.
Clients, teams, projects, models and servers multiply fast. OPAYA gives you one workspace to deploy, clone, update, back up and fix every agent on every VPS and Docker host you look after, and to move skills, MCP servers, secrets and memory between them.
See Opaya in action
See OPAYA in a 30-minute walkthrough with n8n Lab: your agents on this computer, on servers and in Docker, plus answers to your questions.
Group agents per clientEvery VPS and containerUpdates everywhereSecrets never shown
01 / YOUR FLEET
Every client. Every agent. One view.
Group agents per client or team, tag them, and filter by status, place and tag. See what is connected, what needs attention and what runs where.
Groups and tags such as Client: Acme, #prod or #support
Filters by status, this computer, VPS or Docker, with live counts
Connect all, Check & fix all, Update all agents from the Workspace
Local agents, remote agents and API connections side by side
OPAYA
Home: every agent grouped in the sidebar, and fleet-wide actions one click away.
02 / DEPLOY
Build it once. Roll it out in minutes.
Perfect an agent on your own machine, then clone it to the client's server with its settings, skills, memory and instructions. OPAYA installs the agent there first if needed.
As a regular install, a Docker container, or a profile inside a container that already runs there
New VPS makes an SSH key and walks you through adding the server
Logins and keys only when you choose, chat history never
Redeploy the same parts again when you improve the original
OPAYA
A client's support agent cloned to a staging server in Docker.
03 / MAINTAIN
Updates checked everywhere. Applied the right way.
OPAYA checks every agent and tool on every machine every hour, knows how each one was installed, and updates them after one approval. What fails, the Opaya Agent finishes.
Per machine: installed and latest versions, and how each tool was installed
Update selected or Update all, across every server
Docker agents get the new image and keep their data, port and token
Automatic fixes when an agent will not connect because something is too old
OPAYA
Eight updates on two machines, ticked and run with one approval.
04 / MACHINES
Know every server before the client calls.
Test every machine at once: online or not, how fast it answered, its system, uptime, free disk and memory, the tools installed and the agents running there.
Terminal, files and versions on any server in one click
Import ~/.ssh/config or add a server by address
Install Docker there when it is missing
OPAYA
Three servers, tested in one click.
05 / SHARE WHAT WORKS
Skills, MCP, secrets and memory. Between any agents.
A skill that works for one client can work for the next. Move skills, API keys, MCP servers and gateway tokens between agents, frameworks and machines, and keep a global skills library to install anywhere.
Across Hermes, Claude Code, Codex and OpenClaw
Keys land where each agent reads them, values never shown
MCP servers written into each agent's own config after you approve
Memory and personality travel with a clone
OPAYA
Three skills, every key and two MCP servers going to another client agent.
06 / PROTECT
Backups and clean exits. For every agent.
Back up any agent, on any server or in any container, to a private archive on your computer. When a project ends, uninstall removes everything the way it was installed.
With or without chat history and keys, readable only by your user
Gateways and services stopped and removed on uninstall
Hermes profiles and container profiles removed without touching their neighbours
OPAYA
Everything about one client agent, including its backups.
07 / FIRST-LINE SUPPORT
Something broke overnight? It is probably fixed.
Every agent error goes to the Opaya Agent with a first diagnosis: a gateway missing or down, onboarding unfinished, SSH trouble, a dropped connection. It checks, repairs and tells you what it did.
Restarts gateways and containers, finishes onboarding, updates what is too old
Takes new keys safely: it only ever sees a reference
Every command in a terminal you can read, with your approval or iTrust
OPAYA
It fixed the gateway, stored the new key and spotted three updates.
08 / ON SCHEDULE
Routine work runs itself.
A support digest every weekday, an update check every Monday, a backup every night. Opaya sends the message to the agent on time, and each schedule keeps its own chat.
Any agent or the Opaya Agent, daily, weekly, on weekdays or on a cron line
Run now and pause without deleting anything
The agents' own cron jobs listed next to yours, read where each agent runs
OPAYA
A weekday support digest for a client and a weekly update check on every machine.
09 / WHY TEAMS SWITCH
Built for more than one agent at a time.
For people testing agents, models and providers across many projects and organizations.
Every framework
Hermes, Claude Code, Codex, OpenClaw, OpenCode, Goose and any OpenAI-compatible API in the same workspace.
Compare side by side
The Playground asks two agents the same question, so you pick the right one for each client.
Every model
Set the model and reasoning effort per agent or per chat, across providers.
Remote work on local code
An agent on a client server works on your project on its own branch; you review before anything lands.
One vault
Tokens, API keys and MCP secrets encrypted on your computer, shared between agents without being shown.
No new cloud
Agents keep running on your machines and your clients' servers. OPAYA connects over SSH.