What is the crosstabs MCP server?

An open-source Model Context Protocol server that gives AI assistants like Claude 40+ rigorous statistical tools for contingency-table analysis: chi-square, G-test, Fisher's exact, McNemar's, odds ratios, Cramér's V, kappa, CMH, correspondence analysis, power analysis, and more. Install it with pip install crosstabs.

Why use an MCP server instead of letting the AI compute statistics itself?

Language models approximate arithmetic and frequently get p-values and test statistics wrong. The MCP server executes the actual numerical routines, so the assistant reasons about your question while the math is computed exactly.

Does the MCP server send my data anywhere?

No. It runs locally on your machine over stdio. Your tables are only shared with the AI assistant you connect it to.

For AI assistants

Crosstabs MCP Server

Give Claude, Cursor, or any MCP client 40+ exact statistical tools for contingency tables. The assistant does the reasoning; the server does the math — locally, with real numerical routines instead of LLM arithmetic.

Reviewed by the crosstabs.com methods team · Last updated June 10, 2026

Install

pip install crosstabs

Claude Code

claude mcp add crosstabs -- crosstabs

Claude Desktop

{
  "mcpServers": {
    "crosstabs": {
      "command": "crosstabs"
    }
  }
}

In claude_desktop_config.json (Settings → Developer → Edit Config).

Cursor / other MCP clients

Add a stdio server with command crosstabs. Any client that speaks the Model Context Protocol works.

What the assistant can do with it

Tests

Chi-square (with Yates), G-test, Fisher's exact, McNemar's, CMH stratified test, Breslow-Day, Monte Carlo chi-square

Effect sizes

Cramér's V (bias-corrected), phi, odds ratio with CI, relative risk, risk difference, attributable risk

Ordinal measures

Spearman's rho, Kendall's tau, gamma, Somers' D, Stuart's tau-c

Agreement

Cohen's kappa, weighted kappa

Planning

Power analysis, sample-size calculations, Bonferroni and FDR corrections

Exploration

Correspondence analysis, expected counts, residual diagnostics

Example prompt: “Here's my 2×3 table of plan type by churn status — is the association significant, and how strong is it?” The assistant picks the right tests, runs them through the server, and interprets the results.

Open source

MIT-licensed, Python 3.10+. Source lives in the crosstabs-lite repository and releases ship to PyPI. The statistical methods match the methods documented for the web calculators.

Prefer plain HTTP?

The free JSON API exposes chi-square, Fisher's exact, odds ratios, and 20+ measures without installing anything.

Frequently asked questions

What is the crosstabs MCP server?: An open-source Model Context Protocol server that gives AI assistants like Claude 40+ rigorous statistical tools for contingency-table analysis: chi-square, G-test, Fisher's exact, McNemar's, odds ratios, Cramér's V, kappa, CMH, correspondence analysis, power analysis, and more. Install it with pip install crosstabs.
Why use an MCP server instead of letting the AI compute statistics itself?: Language models approximate arithmetic and frequently get p-values and test statistics wrong. The MCP server executes the actual numerical routines, so the assistant reasons about your question while the math is computed exactly.
Does the MCP server send my data anywhere?: No. It runs locally on your machine over stdio. Your tables are only shared with the AI assistant you connect it to.