Skip to content

MCP Setup Guide

The OpexMX MCP (Model Context Protocol) server connects your AI assistant to OpexMX maintenance data: query work orders, search assets and parts, troubleshoot via manuals, and create tickets through natural language. It runs on every OpexMX instance over HTTP — nothing to download or install.

1. Create an API key

In your OpexMX instance, go to Settings → API Keys and create a key. The key carries your permissions — the AI can only do what you may do.

2. Point your client at the endpoint

Every OpexMX instance serves MCP at /api/mcp over Streamable HTTP. Authenticate with your API key as a Bearer header.

Claude Code

claude mcp add --transport http opexmx https://<your-instance-domain>/api/mcp \
  --header "Authorization: Bearer mx_your_api_key"

Cursor, VS Code, Cline, Roo, and other MCP clients

{
  "mcpServers": {
    "opexmx": {
      "url": "https://<your-instance-domain>/api/mcp",
      "headers": {
        "Authorization": "Bearer mx_your_api_key"
      }
    }
  }
}

3. Verify the connection

curl -X POST https://<your-instance-domain>/api/mcp \
  -H "Authorization: Bearer mx_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'

A healthy endpoint answers with a JSON-RPC result listing 69 tools: assets, tickets, parts, work orders, predictive maintenance, knowledge search, and more. Tools are permission-scoped to the key's owner, and new API capabilities appear automatically — no client updates needed.