Fundamental Analysis

fundamental-analysis-mcp

An MCP server that gives an AI assistant the calculators from fundamental-analysis: P/E, DCF, Graham Number, Altman Z-Score, Piotroski F-Score, return on equity, profit margins, and liquidity and debt ratios, plus one tool that runs every ratio your figures allow in a single call.

Every result returns four things: the value, the formula that produced it, the inputs it used and the caveats you need before acting on it. The assistant brings the figures. The server fetches nothing, needs no API key and has no dependencies.

Tools

Tool	What it returns
pe_ratio	Price / EPS, with a note when EPS is zero or negative
dcf_valuation	Enterprise value from projected free cash flow, the terminal value's share of it, and an implied share price when you pass debt, cash and shares
graham_number	sqrt(22.5 x EPS x book value per share), and the price against it
altman_z_score	Z (public manufacturers) or Z'' (everyone else), with the zone
return_on_equity	Net income / equity, null when equity is zero or negative
piotroski_f_score	The 0 to 9 score with each of the nine signals, and which signals a missing prior year leaves unscored
profit_margin	Gross, operating, net or free cash flow margin as a share of revenue
liquidity_ratio	Current ratio, or quick ratio without inventory
leverage_ratio	Debt to equity, or interest coverage (EBIT / interest expense)
analyze_company	Every ratio the figures allow, grouped as valuation, profitability, liquidity, solvency, efficiency, risk and quality, with the metrics that came back null named. Add withInsights for a status and a one-line reading per metric

Every tool is read-only: it computes and answers, and changes nothing.

Set up

Claude Code:

claude mcp add fundamental-analysis -- npx -y fundamental-analysis-mcp

Claude Desktop (claude_desktop_config.json) and Cursor (.cursor/mcp.json):

{
  "mcpServers": {
    "fundamental-analysis": {
      "command": "npx",
      "args": ["-y", "fundamental-analysis-mcp"]
    }
  }
}

Example

Ask: "Graham Number for EPS 4 and book value 25 a share, price 40?" The graham_number tool answers:

{
  "value": 47.43,
  "priceVsGraham": 0.84,
  "formula": "Graham Number = sqrt(22.5 * EPS * book value per share)",
  "inputs": { "eps": 4, "bookValuePerShare": 25, "price": 40 },
  "caveats": ["22.5 is 15 times earnings by 1.5 times book, Graham's limits for a defensive investor. It ignores growth, so it is strict on growth companies and on asset-light businesses."]
}

Build and test

npm run build   # one file, dist/server.js, with the library bundled in
npx vitest run mcp   # from the repository root

MCP registry

server.json describes the server for the official MCP registry as io.github.boffincoders/fundamental-analysis-mcp, and mcpName in package.json carries the same name so the registry can check the npm package. Publish to npm first, then run mcp-publisher publish from this folder.

Licence

MIT. Maintained by Boffin Coders.