Dxpert MCP server
{
"mcpServers": {
"dxpert-mcp": {
"command": "npx",
"args": [
"-y",
"@dxpert/mcp@latest"
],
"env": {
"DXPERT_API_KEY": "<dxpert_api_key>"
}
}
}
}{
"mcpServers": {
"dxpert-mcp": {
"command": "npx",
"args": [
"-y",
"@dxpert/mcp@latest"
],
"env": {
"DXPERT_API_KEY": "<dxpert_api_key>"
}
}
}
}{
"mcpServers": {
"dxpert-mcp": {
"command": "npx",
"args": [
"-y",
"@dxpert/mcp@latest"
],
"env": {
"DXPERT_API_KEY": "<dxpert_api_key>"
}
}
}
}{
"servers": {
"dxpert-mcp": {
"command": "npx",
"args": [
"-y",
"@dxpert/mcp@latest"
],
"env": {
"DXPERT_API_KEY": "${input:dxpert_api_key}"
}
}
},
"inputs": [
{
"id": "dxpert_api_key",
"type": "promptString",
"description": "Dxpert api key",
"password": true
}
]
}{
"mcpServers": {
"dxpert-mcp": {
"command": "npx",
"args": [
"-y",
"@dxpert/mcp@latest"
],
"env": {
"DXPERT_API_KEY": "<dxpert_api_key>"
}
}
}
}file .mcp.json (in your project root)file claude_desktop_config.jsonfile ~/.cursor/mcp.jsonfile .vscode/mcp.jsonfile ~/.codeium/windsurf/mcp_config.jsoninstall npx -y @dxpert/mcpversion 0.1.1
What it does
This MCP server exposes the dxpert.ai API to local MCP clients such as Claude Code or Codex, enabling industrial AI-readiness diagnostics, Unified Namespace (UNS) design, and OT/IT data architecture advisory. It targets engineers and architects working on industrial data infrastructure who want to query dxpert's advisory and agent capabilities from their local tooling using a dxp_ API key.
- Ask advisory questions about AI-readiness, UNS design, and OT/IT data architecture and receive answers grounded in dxpert's knowledge base.
- Run industrial agents (shift report, OEE narrator, alarm triage, maintenance copilot, root cause) over a data bundle you supply, without connecting to a live plant system.
- Generate a namespace design with starter HighByte, MaestroHub, and Node-RED exports by invoking the architect agent with a plain-text message.
Tools this server exposes
| Tool | Does |
|---|---|
| ask_dxpert | Sends a question to the dxpert chat API for advisory and assessment responses about AI-readiness, UNS design, and industrial data standards. |
| run_agent | Runs a named industrial agent (shift-report, oee-narrator, alarm-triage, maintenance-copilot, root-cause, or architect) over a caller-supplied data bundle. |
| run_diagnostic | Submits a 16-field intake to the diagnostic API and returns a preliminary AI-readiness verdict. |
| csv_to_bundle | Converts a shift-report or OEE CSV export into a bundle suitable for passing to run_agent. |
| get_runtime_manifest | Fetches the current runtime version, artifact hashes, and changelog from the dxpert API. |
| get_storefront | Returns what the current account owns and is eligible to purchase. |
| start_purchase | Initiates a Stripe checkout session for a named product and returns a URL the user must open in a browser to complete payment. |
| start_agents_purchase | Initiates checkout for a new site subscription with a selected set of agents and returns a fee breakdown and checkout URL. |
| add_agents | Adds agents to an existing site subscription immediately and charges the prorated amount; requires an account login token. |
| remove_agents | Removes agents from an existing site subscription immediately and credits the prorated amount; requires an account login token. |
Before you enable it
An MCP server runs with whatever access you give it. Start with a development account rather than production, and check which tools the server exposes before you let an agent call them unattended.