Supplement Advisor MCP Server

Evidence-based supplement recommendation MCP server. 17 supplements, 40+ conditions, medication interaction checking, form quality classification. Data sourced from NIH DSLD, PubMed, NSF, and USP.

Built for use with Claude, Cursor, Windsurf, and any MCP-compatible client.

Tools

Tool	Description
recommend_supplement	Get ranked supplement recommendations for a condition with clinical evidence, dosing, cost-per-dose, and purchase links
compare_forms	Compare different forms of a supplement (e.g., magnesium glycinate vs citrate vs oxide) with absorption data and warnings
check_interactions	Check if medications deplete nutrients — covers statins, metformin, PPIs, birth control, and more
get_dosage	Get evidence-based dosage recommendations by condition, with clinical trial references and timing guidance
classify_form	Paste any product name or ingredient list and get a quality verdict on the supplement form used

Supplements Covered

Magnesium, Vitamin D, Omega-3, Creatine, Iron, Vitamin B12, CoQ10, Collagen, Multivitamin, Protein, Biotin, Calcium Citrate, Vitamin C, Methylfolate, Probiotics, Ashwagandha, Electrolytes

Installation

git clone https://github.com/erinheit451/supplement-advisor-mcp.git
cd supplement-advisor-mcp
npm install
npm run build
npm start

Claude Desktop / Claude Code Configuration

Add to your ~/.claude/mcp.json (or Claude Desktop config):

{
  "mcpServers": {
    "supplement-advisor": {
      "command": "node",
      "args": ["/path/to/supplement-advisor-mcp/dist/index.js"]
    }
  }
}

Data Sources

• NIH Dietary Supplement Label Database (DSLD)
• PubMed — clinical trial references with PMIDs
• NSF International — third-party certification data
• USP (United States Pharmacopeia) — quality standards

Full comparison pages and product rankings: verifiedsupplementdata.com

How It Works

The server loads structured evidence data (clinical doses, form comparisons, medication-nutrient interactions) and exposes it through MCP tools. When an AI assistant receives a supplement question, it calls the appropriate tool and gets back formatted, citation-backed recommendations.

Product rankings are based on cost-per-effective-daily-dose, third-party certification status, and clinical evidence — not sponsorship.

License

MIT