METRC MCP
MCP (Model Context Protocol) server that exposes METRC cannabis tracking API tools and analysis skills for LLMs. Use it from Cursor, Claude Desktop, or any MCP client to query and act on facilities, harvests, packages, plants, transfers, and more.
Documentation
| Page | Description |
|---|---|
| Getting started | Setup, credentials, Cursor MCP config |
| Setup (GitHub + Vercel) | Deploy with gh and vercel CLIs — Pages + API |
| Tools | All MCP tools (50+) by category |
| Skills | Analysis skills (needs-attention, facility summary, audit snapshot, FIFO, etc.) and MCP resource URIs |
| Master Framework | Product/feature framework mapped to MCP capabilities (bulk, aging, compliance, yield, reports, etc.) |
| Framework Q&A | Example questions and expected answers for each framework item (demos, testing) |
| Adam Freed Q&A | Questions from Adam Freed and how this MCP answers them; add example answers from chat as you run them |
| Sandbox view | Colorado sandbox: everything referenced and cross-linked (entities, lifecycle, tools, skills) |
| Chat | Chat UI with facility dropdown (GitHub Pages; point API URL at your Vercel deployment) |
What’s included
- Tools: Read and write METRC data (facilities, strains, items, locations, packages, harvests, plants, transfers, lab tests, waste, processing). Colorado sandbox; Basic Auth.
- Resources: Skills are exposed as MCP resources (
metrc://skills/<name>). Any client can callresources/listandresources/readto load skill text and pass it to the LLM. - Cursor: Project rules in
.cursor/rules/route questions to the right skill when this repo is open. - Framework: Master Function & Feature Framework maps 14 feature areas (bulk, aging, compliance, yield, tolling, reports, etc.) to what to build in MCP/skills vs in your app.
Quick start
- Clone the repo, run
npm install, add.envwithMETRC_VENDOR_API_KEYandMETRC_USER_API_KEY. - Add the server to your MCP config (e.g. Cursor) with
command: node,args: ["/path/to/metrc-mcp/server.js"],cwd: "/path/to/metrc-mcp". - Use tools (e.g.
metrc_get_facilities,metrc_get_packages) or ask in natural language; with Cursor rules, questions like “What needs attention?” or “Audit readiness” trigger the right skill.
See Getting started for full setup.