Add Live Oil Prices to Claude, Cursor & AI Agents (MCP Server Tutorial)

Why Give Your Agent Commodity Data?

AI agents are great at reasoning and terrible at knowing today's Brent price. The OilPriceAPI MCP server (open source, MIT) fixes that through the Model Context Protocol: instead of pasting prices into prompts, your agent calls typed tools and gets source-timestamped JSON back.

The server runs locally via npx and talks only to api.oilpriceapi.com. No conversation content leaves your machine — only the structured tool inputs (commodity codes, time periods). It even works with no API key at all in demo mode, serving live prices for a limited commodity set.

Every install below is quoted from the server's own README and was verified against the published oilpriceapi-mcp v2.4.2 package. You can also find it on Smithery, Glama, and our MCP page.

Install in Claude Desktop

Option A: One-Click .mcpb Bundle

Download oilpriceapi-2.4.2.mcpb from the latest GitHub release and double-click it. Claude Desktop installs the server and prompts you for an optional API key.

Option B: JSON Config

Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

{
  "mcpServers": {
    "oilpriceapi": {
      "command": "npx",
      "args": ["-y", "oilpriceapi-mcp"],
      "env": {
        "OILPRICEAPI_KEY": "your-api-key-here"
      }
    }
  }
}

Restart Claude Desktop. You should see the oilpriceapi server with 26 tools in the tools menu.

Install in Claude Code

One command:

claude mcp add oilpriceapi -e OILPRICEAPI_KEY=your-api-key-here -- npx -y oilpriceapi-mcp

Or, to share the server with your whole team, commit a .mcp.json at your project root (this is the config from the server's README):

{
  "mcpServers": {
    "oilpriceapi": {
      "command": "npx",
      "args": ["-y", "oilpriceapi-mcp"],
      "env": {
        "OILPRICEAPI_KEY": "your-api-key-here"
      }
    }
  }
}

Verify with /mcp inside Claude Code — the server reports as oilpriceapi v2.4.2.

Install in Cursor

Add to .cursor/mcp.json in your project root:

{
  "mcpServers": {
    "oilpriceapi": {
      "command": "npx",
      "args": ["-y", "oilpriceapi-mcp"],
      "env": {
        "OILPRICEAPI_KEY": "your-api-key-here"
      }
    }
  }
}

Cursor picks the server up on the next window reload (Settings → MCP shows it as connected). VS Code + Cline and Windsurf use the same npx command — see the README for those config paths.

Add Your API Key

Demo mode covers Brent, WTI, diesel, gasoline, natural gas, gold, heating oil, and two forex pairs. For the full 70+ commodity catalog, historical data, futures curves, and persistent alerts, sign up for a free API key — 7-day free trial, then a free tier of 200 requests per month. No credit card required.

Put the key in the OILPRICEAPI_KEY env entry of whichever config you used above. The key is stored locally in your MCP client's config and sent only to api.oilpriceapi.com.

Real Prompts to Try

These exercise different tools. Each answer comes back with the source timestamp of the underlying quote, so the agent can tell you how fresh the number is.

Production: Alerts & Price Watches That Outlive the Chat

A one-off price check is a demo. The production pattern is to have your agent set up monitoring that keeps running after you close the window. With an API key, two tool families make that possible:

That's the difference between an agent that can quote a price and an agent with a standing market desk: the data pipeline runs on a schedule whether or not anyone is chatting.

Frequently Asked Questions

Give Your Agent a Market Desk

npx -y oilpriceapi-mcp