AIOZ AI Platform MCP Server

The AIOZ AI Platform MCP server gives your coding agent direct access to the AIOZ AI Platform: run inference on models, manage models and datasets, upload files, check balances and task history, and work with organizations and repositories.

For the full tool list, see the MCP Catalog.

[!NOTE] You need an editor or app that supports MCP servers. Verified with VS Code, Cursor, Trae, Antigravity, and Claude Desktop.

Prerequisites

• An AIOZ AI Platform account.
• An API key. Log in to AIOZ AI Platform, go to Settings → API Keys, and create one. Public tools (browsing public models and datasets) work without a key.

Install

The server runs locally over stdio. Add it to your client's MCP config with your API key in the env block, and every tool call is authenticated automatically.

The npx command below downloads the right binary for your platform on first run and caches it. It needs Node.js. If you don't have Node, use a pre-built binary instead.

VS Code (.vscode/mcp.json, or run MCP: Open User Configuration for all workspaces):

{
    "servers": {
        "aioz-ai-platform": {
            "type": "stdio",
            "command": "npx",
            "args": ["-y", "@aiozai/mcp-server"],
            "env": {
                "AIOZ_AI_API_KEY": "xxxxxxxxxxxxxxxx"
            }
        }
    }
}

Cursor (.cursor/mcp.json), Claude Desktop (claude_desktop_config.json), Trae (mcp.json), Antigravity (mcp_config.json) all use the same shape, with mcpServers instead of servers:

{
    "mcpServers": {
        "aioz-ai-platform": {
            "command": "npx",
            "args": ["-y", "@aiozai/mcp-server"],
            "env": {
                "AIOZ_AI_API_KEY": "xxxxxxxxxxxxxxxx"
            }
        }
    }
}

Where to find the config in each app:

| App | How to open it | |---|---| | VS Code | ⌘/Ctrl + Shift + P → MCP: Open User Configuration (global) or MCP: Open Workspace Folder MCP Configuration | | Cursor | Settings → Cursor Settings → Tools & MCP → + Add Custom MCP | | Trae | Settings → MCP → Add → Add Manually | | Antigravity | Side panel → Agent session → "..." → MCP Servers → Manage MCP Servers → View raw config, then Refresh | | Claude Desktop | Settings → Developer → Edit Config |

Pre-built binary (no Node.js)

Download your platform's binary from the Releases page:

| Platform | Binary | |---|---| | Linux x86-64 | aioz-ai-mcp-linux-amd64 | | Linux ARM64 | aioz-ai-mcp-linux-arm64 | | macOS Intel | aioz-ai-mcp-darwin-amd64 | | macOS Apple Silicon | aioz-ai-mcp-darwin-arm64 | | Windows x86-64 | aioz-ai-mcp-windows-amd64.exe |

Make it executable (chmod +x, macOS and Linux), then use the same config as above with the npx command swapped for the binary's path:

"command": "/path/to/aioz-ai-mcp",
"args": [],

Agent skills (recommended)

The tools alone don't teach an agent the inference lifecycle. Two skills do:

npx skills add AIOZNetwork/aioz-ai-developer/skills/aioz-ai-run-inference
npx skills add AIOZNetwork/aioz-ai-developer/skills/aioz-ai-upload-file

See skills/ for what they cover.

API key

Set AIOZ_AI_API_KEY in the env block of your config, as shown above. The server attaches it to every request, so tools work without passing a key per call.

To use a different key for one specific call, pass x_api_key on that call. It overrides the configured key for that call only.