Connect an AI assistant to blitz.cloud over MCP

blitz.cloud runs an MCP server at https://mcp.blitz.cloud/<your name>. Connect an AI assistant to it and the assistant can do what you can do in the dashboard: put apps online, create databases, change settings, read logs and add DNS records. You can then ask it things like "put Uptime Kuma online at uptime" or "why did my app stop?".

There are two ways to connect. Most assistants sign in with your blitz.cloud account, the same way you connect them to other services: you add the address, a browser window opens, you sign in and allow it. Scripts and clients that can't sign in use an API key instead.

Your server address is on the AI tools page in the dashboard.

Sign in (recommended)

Claude (web and desktop): Settings, Connectors, Add custom connector. Paste your server address, click Connect and sign in.

Claude Code

claude mcp add --transport http blitz https://mcp.blitz.cloud/<your name>

Then type /mcp in Claude Code, pick blitz and sign in.

Cursor (~/.cursor/mcp.json)

{
  "mcpServers": {
    "blitz": { "url": "https://mcp.blitz.cloud/<your name>" }
  }
}

VS Code (.vscode/mcp.json)

{
  "servers": {
    "blitz": { "type": "http", "url": "https://mcp.blitz.cloud/<your name>" }
  }
}

ChatGPT and other clients: add a connector or MCP server with your server address and sign in when asked.

When you sign in, blitz.cloud shows which app is asking before you allow it. If the app's address belongs to a publisher blitz.cloud recognises (Claude Code and VS Code), the page names that domain. Any other app registered itself, and the page says its name is its own claim. Only allow an app you just connected yourself. blitz.cloud never asks you to do this from an email or a message.

Each app you sign in appears under Connected apps on the AI tools page. Disconnect stops it immediately, not when its sign-in would have run out. Signing it in again later works.

With an API key

On the AI tools page, give the key a name (the computer or tool that will use it) and click Make a key.

• The key is shown once, right after you make it. Copy it then. blitz.cloud only keeps a hash of it and can't show it again.
• You can have up to 5 keys. Make one per computer or tool, so you can take one back without touching the others.
• Rotate gives a key a new value under the same name. The old value stops working at once.
• Revoke deletes a key. Anything still using it is refused from then on.

Send the key with every request in the Authorization: Bearer <your key> header. For Claude Code:

claude mcp add --transport http blitz https://mcp.blitz.cloud/<your name> --header "Authorization: Bearer <your key>"

The AI tools page has the same setup for Cursor and VS Code, with a new key already filled in. Treat a key like a password. Anyone who has it can manage your apps.

What an assistant can do

The full list, with a description of each tool, is on the AI tools page and comes straight from the server. In short:

Restarting, pausing, deleting and removing ask you first. If your client can show a question from the server (VS Code can, for example), blitz.cloud asks you there directly, and the assistant can't answer for you. Other clients have to call again with confirm: true, so the assistant should check with you before it does.

Connecting a Tailscale or WireGuard network is not a tool. It takes a secret or a config file, and those belong in the dashboard rather than in a chat.

Limits

• Each key or connected app can make 120 calls a minute, and 20 of those can change something. A call over the limit is refused with how many seconds to wait.
• Your plan's limits are the same as in the dashboard. An assistant can't put more apps online than you could.
• Changes need a confirmed email address, just like in the dashboard.
• A key or a sign-in for the MCP server works only there. It doesn't open the dashboard or its API.

Keeping track

Everything an assistant changes is recorded in your account's audit log under the key or app that did it. On an app's Activity tab it appears as done by you.