Your Own Agent

Bring Your Own Agent

If you already work with an AI agent, it can create and manage RepoCloud servers for you. This is not the DevOps AI Agent. Your agent runs in your own tools and with your own AI provider, and RepoCloud gives it a way to create a server, a way to manage that server, and the server itself.

The place to start is the AI Agent Portal, which is the API / MCP tab of your dashboard at repocloud.io/mcp.

What the Portal Gives You

• API Keys

  Press Generate New Key to make a key for your agent. The portal shows the full key once: “Copy it now — it won’t be shown again”

• Connection Settings

  Copy-and-paste settings for several MCP clients, and a JSON form for others

• Agent Skill File

  A SKILL.md file that your agent can follow without MCP: Copy it or Download as SKILL.md

• REST API Reference

  The endpoints behind it all, for scripts and for agents that call the API directly

Three Ways to Connect

The skill file in the portal calls the REST API with curl, so it works without MCP. The copy at https://mcp.repocloud.io/SKILL.md is the version for an agent that has the MCP server connected: it tells the agent to call the MCP tools, so it needs the MCP connection. The full addresses are https://mcp.repocloud.io/mcp for the MCP server and https://repocloud.io/api/v1/ for the REST API.

What Your Agent Can Do

With a key, your agent can:

• Deploy an Ubuntu VPS
• Check a server’s status and get its connection details
• Power a server on and off, and reboot it
• Resize it to another tier
• Delete it

If your agent can run shell commands, it can then log in to the server over SSH with the details the API returns and install and configure an application: the skill file describes that workflow. Every action uses your key and your balance, so the agent acts on your behalf.

What the API Does Not Cover

The API supports Ubuntu VPS servers only, and it sees only the servers created through it with a key. It has no endpoints for snapshots, SSH keys, custom domains, billing, Container Apps, or AI Deploy: those stay in the dashboard. A server created through the API has no DevOps AI Agent chat.

Private Repositories

AI Deploy takes public GitHub repositories. For a private one, use your own agent: it creates the server, logs in, and sets the application up, and you give it access to the repository yourself, with a key or token you can revoke. The AI Deploy form points to the AI Agent MCP for this.

Costs and Requirements

• To create a key, the account needs a payment of $1 or more.
• To deploy a server, the balance must be above $3.
• The server is billed by the hour at its tier’s price, like any VPS App.
• The AI model your agent uses is billed by its own provider, not by RepoCloud.

Keeping It Safe

• Treat the key as the agent’s authority. Keep it in an environment variable instead of a file you commit, and revoke it in the portal when you stop using it.
• The answers about a server carry its root password, so your agent’s conversation holds it.
• Keep your client’s approval prompts on for tools that resize or delete servers.

See Safety and Control for habits that apply to any agent working on a server.

Next Steps

• MCP Setup

  Connect Cursor, Claude Code, and other MCP clients

• API Introduction

  Authentication, limits, and the endpoints

• API Key Management

  Create, list, and revoke keys

• AI Agents Overview

  Compare every way to use an AI agent on RepoCloud