What Is the MetalMiner Model Context Protocol Server?

The MetalMiner Model Context Protocol (MCP) metal prices server gives AI agents and assistants - including Claude, Cursor, VS Code, ChatGPT, OpenAI, Grok, Perplexity and Windsurf - live industrial-metal and commodity data through natural language. Over a single remote MCP endpoint it exposes more than 20 tools for current prices, up to 10 years of history, support/resistance levels, trend and correlation analysis, 12-month price forecasts, what-if scenario simulation, production-cost (should-cost) data and MetalMiner's proprietary analyst research.

Support/resistance levels are the price levels where a market has historically stopped falling or stopped rising, and should-cost data is an estimated cost to produce a part or product, built from material, energy, labor and conversion inputs. Coverage centers on industrial metals: base metals quoted on the London Metal Exchange (LME), the steel complex, stainless and battery/specialty metals, plus energy and agricultural markets, with prices normalized across exchanges so LME, Shanghai Futures Exchange (SHFE) and COMEX (CME Group metals exchange) values can be compared directly. To connect, add the endpoint to your AI client and sign in with your MetalMiner account (OAuth) - there is no key to copy.

Unlike a raw price API, the server lets an agent ask questions the way a buyer or trader would - "is copper near resistance?", "forecast stainless for Q3", "compare the cost to produce nickel by country" - and return analyst-grade answers with the numbers behind them. Model Context Protocol (MCP) is the open standard, originated by Anthropic, that lets AI assistants call external tools; any client that speaks MCP can use MetalMiner without custom integration.

MetalMiner MCP at a Glance

What

Industrial-metals & commodity intelligence delivered as an MCP server.

Endpoint

https://mcp.metalminer.com/mcp

Transport

Streamable HTTP, the transport used by the remote MCP endpoint and supported by the MCP-compatible clients listed below - one remote server, no local install or process.

Clients

Claude Desktop, Claude Code, Cursor, VS Code (Cline), Windsurf, ChatGPT, OpenAI, Grok, Perplexity - any MCP client.

Data

LME base metals, steel, stainless, battery & specialty metals, plus energy & agriculture.

Beyond prices

Forecasts, support/resistance, correlations, scenario simulation, should-cost, research.

Auth

OAuth 2.0 - sign in with your MetalMiner account when your client connects (no key in the URL).

Premium

Forecasts, scenario simulation and production-cost data; enabled per account.

What Data and Tools Does the MetalMiner MCP Server Provide?

MetalMiner's MCP server provides more than 20 tools spanning pricing, technicals, forecasting, cost modeling and research. Agents call them in plain English; the client selects the right tool automatically. The table below lists the core capabilities, what each returns, and an example prompt.

Which Metals and Commodities Does the MetalMiner MCP Server Cover?

MetalMiner's MCP server covers the metals that procurement and trading teams actually price: LME base metals, the full steel complex, stainless and its inputs, and battery/specialty metals - plus energy and agricultural markets.

How Does MetalMiner MCP Compare With Generic Price APIs and Precious-Metals MCP Servers?

MetalMiner's MCP server tells an agent where industrial metals are heading, what's driving them, and what it costs to make them - the reasoning a buyer or trader needs - while a precious-metals spot API can only tell you what gold costs. Most commodity MCPs return only a current or historical price; most metal MCPs available today are precious-metals spot servers.

What Can Teams Build With the MetalMiner MCP Server?

• Procurement & should-cost agents - benchmark supplier quotes against live prices and production cost, and flag when to buy.
• Hedging & trade timing - watch support/resistance and forecasts to time purchases or hedges across the metals complex.
• Forecasting copilots - drop 12-month outlooks and what-if scenarios straight into Claude or a custom agent.
• Market monitoring - daily briefings that pull prices, trends and the latest analyst research automatically.
• Risk & scenario analysis - model tariff, sanction and supply-shock impacts before they hit the cost base.
• Custom metals agents - one MCP endpoint and 20+ tools as the data layer for any commodity AI workflow.

How Do You Add the MetalMiner MCP Server to Your AI Client?

Add the MetalMiner MCP metal prices endpoint https://mcp.metalminer.com/mcp to your AI client. The first time it connects, your browser opens to sign in to MetalMiner and approve access using OAuth with PKCE (Proof Key for Code Exchange), the OAuth flow used for browser-based clients - no key to copy or paste. It's one remote server: no install, no local process. Setup for each supported client is below.

Claude (Desktop & Web)

Open Settings → Connectors → Add custom connector, paste https://mcp.metalminer.com/mcp, and click Connect. A browser window opens to sign in to your MetalMiner account and approve access - then Claude is connected. No config file or key needed.

Claude Code (CLI)

claude mcp add --transport http metalminer https://mcp.metalminer.com/mcp

Cursor

{
  "mcpServers": {
    "metalminer": {
      "url": "https://mcp.metalminer.com/mcp"
    }
  }
}

Cursor prompts you to sign in (OAuth) the first time it connects.

VS Code (Cline)

{
  "mcpServers": {
    "metalminer": {
      "url": "https://mcp.metalminer.com/mcp"
    }
  }
}

Approve the sign-in prompt (OAuth) when it first connects.

Windsurf

{
  "mcpServers": {
    "metalminer": {
      "serverUrl": "https://mcp.metalminer.com/mcp"
    }
  }
}

Grok

Go to grok.com/connectors and click New Connector → Custom connector.

1. Name it MetalMiner and set the Server to https://mcp.metalminer.com/mcp, then continue to the next screen.
2. Fill in the OAuth details:
   • Client ID: grok-connector
   • Client Secret: leave empty
   • Authorization Endpoint: https://auth.metalminer.com/api/oidc/auth
   • Token Endpoint: https://auth.metalminer.com/api/oidc/token
   • Scopes: openid email mcp:tools (press Enter after each one instead of typing commas)
   • Token Auth Method: none (PKCE only, recommended)
3. Save, then sign in with your MetalMiner account when Grok opens the browser.

Perplexity

In Perplexity, click your profile → Settings → Connectors (custom connectors require a paid plan).

1. Click + Custom connector, name it MetalMiner and paste the Server URL https://mcp.metalminer.com/mcp.
2. Set Transport to Streamable HTTP, Authentication to OAuth, and tick the acknowledgment checkbox.
3. Save, then sign in with your MetalMiner account. If the OAuth setup fails, expand Advanced and set the Client ID to perplexity-connector with no secret.

ChatGPT (OpenAI)

In ChatGPT, open Settings → Connectors → Advanced settings and enable Developer mode (available on Plus and Pro; on Business or Enterprise an admin must enable it).

1. Back in Connectors, click Create.
2. Name it MetalMiner, paste the MCP Server URL https://mcp.metalminer.com/mcp and choose OAuth authentication. If it asks for a Client ID, use chatgpt-connector with no secret.
3. Accept the custom-connector notice, create it, and sign in with your MetalMiner account.

Any MCP Client (Streamable HTTP)

# Endpoint
https://mcp.metalminer.com/mcp

# Auth: OAuth 2.0 (Authorization Code + PKCE)
# Point your client at the endpoint above; it discovers the
# authorization server automatically and opens a browser for
# sign-in. The token is sent as Authorization: Bearer - you
# never copy a key.

# Programmatic / non-interactive access (optional):
# send your key as  X-API-Key: YOUR_KEY

Then just ask, for example: What's the LME aluminum price and 3-month forecast? Your client discovers the available tools automatically.

What Should Users Know Before Connecting to the MetalMiner MCP Server?