Rigor statistics MCP server

Rigor is a small helper program that gives your AI actual statistical math. Instead of your AI recalling a half-remembered number from training, it computes the test, checks the assumptions, and hands back a cited answer. It is handy for anyone who runs experiments, surveys, or A/B tests and needs numbers they can trust.

What this MCP server does

You ask your AI a statistics question in plain words, like whether two groups differ or how many people you need in a study. Your AI passes the numbers to Rigor. Rigor runs the actual test from scratch, checks the assumptions, and picks a citation. Then your AI shows you the result in the chat: the statistic, the p-value, a confidence interval, and any warnings.

Click to zoom

What you can do with it

• Run t-tests, z-tests, chi-squared, Fisher's exact, McNemar, and ANOVA
• Run non-parametric tests like Mann-Whitney, Wilcoxon, and Kruskal-Wallis
• Compute effect sizes such as Cohen's d, Hedges' g, and Cramér's V
• Work out power and required sample size for a planned study
• Correct p-values for multiple comparisons with Bonferroni or Benjamini-Hochberg
• Run pairwise comparisons across many groups and correct the whole batch at once
• Check a running experiment repeatedly without inflating false positives

Try asking your AI

• “I have two groups of test scores, 12 people each. Are they significantly different?”
• “How many people per group do I need to detect an effect size of 0.46 with 85% power?”
• “Run a chi-squared test on this 2x3 table of survey answers and tell me if the rows are independent.”
• “I checked my A/B test dashboard 10 times and it just hit p=0.04. Is that real?”

What it gives back to you

You get back a clear answer in the chat: the test statistic, the p-value, a confidence interval, and a citation for the method used. Rigor also flags assumption warnings, like when a sample is too small for the test to be reliable. For sample size questions, you get a number to round up. For multiple comparisons, you get corrected p-values for each test.

Before you start

Who it's for

Researchers, analysts, product managers, and anyone who runs experiments or surveys and wants trustworthy statistics without doing the math by hand.