Getting started

1. Install

git clone https://github.com/YOUR_ORG/metrc-mcp.git
cd metrc-mcp
npm install

2. Credentials

Create a .env file in the repo root:

METRC_API_URL=https://sandbox-api-co.metrc.com
METRC_VENDOR_API_KEY=your-vendor-key
METRC_USER_API_KEY=your-user-key

For production, point METRC_API_URL to the appropriate state METRC API and use production keys.

3. Add to MCP (e.g. Cursor)

In Cursor: Settings → MCP (or .cursor/mcp.json):

{
  "metrc": {
    "command": "node",
    "args": ["/absolute/path/to/metrc-mcp/server.js"],
    "cwd": "/absolute/path/to/metrc-mcp"
  }
}

To pass keys via config instead of .env, add an env block to the server entry. Scripts (e.g. npm run populate-simulated-year) also read from MCP config: they try .env first, then .cursor/mcp.json in the repo, then ~/.cursor/mcp.json.

{
  "metrc": {
    "command": "node",
    "args": ["/path/to/metrc-mcp/server.js"],
    "cwd": "/path/to/metrc-mcp",
    "env": {
      "METRC_VENDOR_API_KEY": "your-vendor-key",
      "METRC_USER_API_KEY": "your-user-key"
    }
  }
}

Restart the MCP server (or Cursor) after pulling new tools.

4. Use tools and skills

• Tools: In Cursor or any MCP client, call tools by name (e.g. metrc_get_facilities, metrc_get_packages). See Tools.
• Skills: In Cursor with this repo open, ask “What needs attention?”, “Summarize the facility”, “Audit in a week — check risk areas”, “What to pull for samples?” — the rules route to the right skill. See Skills.
• External LLMs: Call resources/list and resources/read with URIs like metrc://skills/needs-attention to load skill text into context, then call the METRC tools as described in the skill.

5. Test

npm test

Runs an integration test (initialize, list tools, call a few tools). Requires no credentials for structure checks; with valid .env, API calls are exercised.