Tell your AI assistant to run your cloud
Connect Claude, ChatGPT, Codex, Cursor, or any MCP-compatible assistant to CloudCenmax and deploy servers, restart them, check your balance, and top up — in plain language, without leaving your editor. Your API key decides exactly what the assistant is allowed to touch.
Connected in three steps
No SDK to install and no glue code to maintain. It's your existing API key and one URL.
Issue an API key
Create a key in your dashboard and choose its scope — read-only for safe browsing, or read + write to let the assistant make changes.
Add the MCP endpoint
Drop the CloudCenmax MCP URL and your key into your client config — Claude Desktop, Claude Code, ChatGPT, Codex, Cursor, and any other MCP-aware tool.
Talk to your agent
“Deploy a 4 GB server in Frankfurt and tell me my balance.” Your assistant calls the right tools and reports back.
What your assistant can do for you
Every action maps to the same governed API that powers the dashboard — so the assistant can only do what your account already can.
Deploy servers
Spin up a VPS from any catalog plan by name — “launch a high-performance box in Singapore.”
Control the lifecycle
Start, stop, reboot, or tear down servers on command, with status read back to confirm.
Browse the catalog
Search plans by location, type, and price before committing to a deploy.
Watch your spend
Pull live balance, ledger activity, and deposit history without opening the billing page.
Top up the account
Start a deposit so a provisioning request never stalls on a low balance.
Run provider actions
Open a VNC console, reinstall, enter rescue mode, take snapshots, set hostnames, enable IPv6.
Manage object storage
Create and delete S3-compatible buckets in any region and list them — “make a bucket called backups in Frankfurt.”
Giving an AI the keys shouldn't mean giving up control
The server never stores your credentials. Each request carries your API key, and that key decides what's possible. Hand your assistant a read-only key to explore safely, or a write key when you're ready to let it act.
Read-only keys can browse and report but cannot change anything. Write access is opt-in, per key.
Keys are displayed a single time and stored only as a hash — revoke and rotate any time.
Tools that terminate a server are clearly marked so your assistant treats them with care.
Requests are throttled and locked to your organization — no cross-account access, ever.
Add CloudCenmax to your AI client
The hosted endpoint is https://mcp.cloudcenmax.com/mcp. Pick your client below — every option below works against the same live server today.
No config file. You'll be redirected to cloudcenmax.com to log in and approve.
- claude.ai → Settings → Connectors → Add custom connector.
- Paste this URL: https://mcp.cloudcenmax.com/mcp
- Click Connect. A new tab opens on cloudcenmax.com — log in, pick the organization, tick the AI-risk acknowledgement, and Approve.
- Back in Claude, enable the CloudCenmax connector in a chat and ask: "what's my balance?"
Requires a paid Claude plan (custom connectors are not on Free).
Edit your claude_desktop_config.json and restart the app.
- macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
- Windows: %APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"cloudcenmax": {
"type": "http",
"url": "https://mcp.cloudcenmax.com/mcp",
"headers": {
"Authorization": "Bearer ck_your_api_key"
}
}
}
}Replace ck_your_api_key with a real key from your API keys page.
One command in your terminal:
claude mcp add cloudcenmax --transport http \
--header "Authorization: Bearer ck_your_api_key" \
https://mcp.cloudcenmax.com/mcpThe MCP server appears under /mcp inside Claude Code and tools become available immediately.
ChatGPT supports custom MCP connectors on Pro / Business / Enterprise plans.
- chatgpt.com → Settings → Connectors → Create.
- Connector type: MCP server. URL: https://mcp.cloudcenmax.com/mcp
- Authentication: OAuth — ChatGPT discovers CloudCenmax as the auth server automatically. Approve the consent on cloudcenmax.com.
- Enable the connector in any chat to use the CloudCenmax tools.
If your tenant restricts custom connectors, ask your workspace admin to allow CloudCenmax.
Cursor reads ~/.cursor/mcp.json (or the same JSON shape via Settings → MCP).
{
"mcpServers": {
"cloudcenmax": {
"type": "http",
"url": "https://mcp.cloudcenmax.com/mcp",
"headers": {
"Authorization": "Bearer ck_your_api_key"
}
}
}
}Replace ck_your_api_key with a real key from your API keys page.
The official MCP Inspector is the quickest way to verify the connection and explore the tools without touching a client.
npx @modelcontextprotocol/inspector- Transport: Streamable HTTP.
- URL: https://mcp.cloudcenmax.com/mcp
- Header: Authorization: Bearer ck_your_api_key
- Hit Connect — you should see the full tool list.
Questions, answered
Is it safe to let an AI deploy and delete servers?
You decide. Issue a read-only key and the assistant can only look. Grant write access only when you want it to act — and revoke any key instantly.
Does the MCP server cost extra?
No. The server itself is free to use. You only pay the same hourly rate for the servers you actually run.
Do I need to install anything?
No SDK or agent to host. Add one URL and your API key to your MCP client config and you are connected.
Which assistants can connect?
Any client that speaks MCP over Streamable HTTP — including Claude Desktop, Claude Code, ChatGPT, OpenAI Codex, Cursor, and custom agents built on an MCP SDK.
Can the assistant see other customers’ data?
Never. Every request is scoped to your organization and rate limited. Keys cannot reach across accounts.
What if I want to undo write access later?
Rotate or delete the key from your dashboard. Access stops immediately on the next request.
Put your cloud on autopilot
Create an account, generate an API key, and connect your assistant in minutes. Hourly billing, no contracts, cancel anytime.