Dark Matter Docs

Connect an LLM to Dark Matter via the Model Context Protocol: install MCP server, configure client, query positions via natural language.

What you will build

You will connect an LLM (e.g., Claude, ChatGPT) to Dark Matter using the Model Context Protocol (MCP). The LLM will be able to query positions, search inventory, and answer questions about your private market data using natural language.

Prerequisites

A DMTR API key
An MCP-compatible client (e.g., Claude Desktop, Cursor)
Node.js 18+ or Python 3.10+ (depending on MCP server runtime)

Step 1: Install MCP server

Install the DMTR MCP server from npm or the GitHub releases. Follow the installation instructions for your environment. The server exposes DMTR API capabilities as MCP tools and resources.

npm install -g @dmtr/mcp-server

Step 2: Configure client

Add the DMTR MCP server to your client configuration. Provide your API key (or path to credentials) in the config. The server will authenticate to Dark Matter on your behalf when the LLM invokes tools.

{
  "mcpServers": {
    "dmtr": {
      "command": "npx",
      "args": ["@dmtr/mcp-server"],
      "env": {
        "DMTR_API_KEY": "your-api-key-here"
      }
    }
  }
}

Step 3: Query positions via natural language

Start a conversation with your LLM and ask questions like "What positions do I have?" or "Show me SPV LLC positions over $1M." The LLM will use MCP tools to call the DMTR API and return formatted results. Experiment with filters, aggregations, and follow-up questions.

What you learned

You installed the DMTR MCP server, configured your LLM client, and queried positions via natural language. The MCP integration enables AI assistants to act on your Dark Matter data with appropriate tool use.

Next steps

MCP Tools — Full tool reference
MCP Resources — Exposed resources
First API Call — Direct API usage

Connect via MCP