Crosstabs
Run contingency-table statistics from your agent when you need to test survey splits, funnel segments, or A/B-style category counts without opening SPSS or R.
Overview
io.github.barangaroo/crosstabs is a Grow-phase MCP server that runs chi-square, Fisher’s exact, odds ratio, and effect-size analysis on categorical contingency tables inside your agent.
What is this MCP server?
- Chi-square and Fisher’s exact tests for 2×2 and larger contingency tables
- Odds ratios and effect-size summaries alongside p-values
- PyPI package `crosstabs` with stdio MCP transport (v1.0.2)
- Suited to survey exports, support tags, and lightweight experiment tables
- Local stdio server—no hosted remote URL in server metadata
- Server version 1.0.2
- PyPI identifier crosstabs, stdio transport
- Repository subfolder mcp-server-python on barangaroo/crosstabs-lite
What problem does it solve?
You have segment or survey counts in a table but no fast, trustworthy way to test whether differences are statistically meaningful while you stay in the agent workflow.
Who is it for?
Solo builders analyzing small-to-medium categorical datasets from surveys, funnels, or support taxonomies who want MCP-native stats without R or Jupyter.
Skip if: Teams needing regression, time-series forecasting, big-data warehouses, or production BI dashboards—this server is contingency-table statistics only.
What do I get? / Deliverables
Your agent can return formal crosstab test results and effect summaries you can paste into validation memos, growth dashboards, or investor one-pagers.
- Chi-square or Fisher’s exact test results for your table
- Odds ratio and effect-size style summaries for interpretation
- Agent-ready statistical narrative tied to your segment breakdown
Recommended MCP Servers
Journey fit
Solo builders validate growth and product decisions with real response data; formal crosstab stats belong where you interpret metrics and cohort behavior, not where you write UI code. Chi-square, Fisher’s exact, odds ratios, and effect sizes answer “is this segment difference real?”—the core question in analytics and experiment readouts.
How it compares
A statistics MCP tool for crosstabs, not a full data-science agent skill or spreadsheet replacement.
Common Questions / FAQ
Who is io.github.barangaroo/crosstabs for?
Indie and solo builders who use Claude Code or Cursor and need chi-square and related tests on category cross-tabs without switching to a separate stats app.
When should I use io.github.barangaroo/crosstabs?
Use it when you have two (or more) categorical variables in a table—segments, answers, or labels—and you want significance tests and effect measures before acting on the pattern.
How do I add io.github.barangaroo/crosstabs to my agent?
Install the PyPI package crosstabs (v1.0.2), configure an stdio MCP server entry pointing at that package’s server command, and restart your MCP client.