# Normality Test Engine MCP for AI Agents AI Agent Connect > Normality Test Engine. Verify if your data follows a normal distribution before running parametric tests like t-tests or ANOVA. This Connector uses local Skewness and Kurtosis analysis to give your AI agent a definitive pass/fail verdict on your data's normality, ensuring your results aren't based on guesses. ## Overview - **Category:** developer-tools - **Price:** Free - **Endpoint:** https://edge.vinkius.com/vk_preview_WUlfcwTW8ijdRBrYjBDTpDiFw1dcKsTz5HdufzbG/ai-agent-connect - **Tags:** statistics, data-science, normality-test, mathematical-analysis, data-validation ## Description Normality Test Engine lets you verify if your data follows a normal distribution before you run any parametric tests. When you're doing serious data analysis, you can't just hope your data behaves. If you run a t-test or a linear regression on non-normal data, your results are basically junk. Most people try to ask an AI to eyeball the distribution, but that's a huge mistake. LLMs aren't calculators; they're word predictors. They'll see a list of numbers and guess at the shape, often getting it wrong and leading to flawed research. This MCP changes that by giving your agent actual math muscles. It pulls your raw numbers and runs a Jarque-Bera test locally on your machine. This means the AI gets a real p-value and a clear normal or not normal answer based on exact coefficients. You get the mean, standard deviation, skewness, and kurtosis without the AI making up any numbers. It's a huge step up for anyone who needs accuracy over vibes. You can find this and thousands of other specialized tools in the Vinkius catalog to make sure your data work stays accurate. ## Tools ### test_normality Runs a Jarque-Bera test on your numbers to check for normality. It gives you a p-value and a clear pass/fail verdict for your data. ## Prompt Examples **Prompt:** ``` Check if these residuals are normal for my regression. ``` **Response:** ``` I've run the Jarque-Bera test on your residuals. The p-value is 0.42, which means we fail to reject the null hypothesis. Your residuals appear normally distributed, so you're good to proceed with your regression model. ``` **Prompt:** ``` Is this sensor data normal or should I use a non-parametric test? ``` **Response:** ``` The data shows significant positive skewness (1.8) and high kurtosis (5.2). The Jarque-Bera p-value is 0.001. Because the data is not normally distributed, you should use a non-parametric test instead of a standard t-test. ``` **Prompt:** ``` Run a normality test on the Revenue column. ``` **Response:** ``` Normality test passed (p = 0.61). The skewness is 0.3 and kurtosis is 2.9, both of which are close to normal parameters. Your confidence intervals will be valid for this data. ``` ## Capabilities ### Run Jarque-Bera tests Perform a deterministic normality check on your numeric data to get a p-value. ### Get skewness and kurtosis Calculate exact coefficients to see how much your data leans or tails. ### Receive pass/fail verdicts Get a clear interpretation of whether your data is normal enough for parametric tests. ### Pull descriptive statistics Get mean and standard deviation alongside your normality metrics. ### Keep data local Ensure your research numbers never leave your machine during the calculation. ## Use Cases ### Regression Analysis A researcher wants to check if their residuals are normal before trusting their linear model. ### Survey Validation A sociologist needs to know if survey scores are normally distributed before running an ANOVA. ### Sensor Data An engineer wants to see if noisy sensor data is normal enough for standard deviation-based alerts. ### Finance A quant wants to check if a set of stock returns is normal before applying standard risk formulas. ## Benefits - Stop AI hallucinations by using local math instead of LLM guesswork for your normality tests. - Get a definitive pass or fail verdict so you know exactly when to use parametric tests. - Keep your sensitive research data private because all calculations happen on your local machine. - Get precise skewness and kurtosis values to understand the exact shape of your data distribution. - Save time on manual calculations by letting your agent handle the Jarque-Bera test automatically. ## How It Works The bottom line is you get a mathematically sound go or no-go for your statistical tests instead of an AI guess. 1. Provide your agent with a list of numeric data points. 2. The agent calls the normality check to run the Jarque-Bera test. 3. You get a report containing the p-value, coefficients, and a clear normality verdict. ## Frequently Asked Questions **What is the Normality Test Engine MCP for?** It helps your AI agent check if your numeric data follows a normal distribution. This is a required step before you can use many standard statistical tests like ANOVA or t-tests. **Does the Normality Test Engine MCP use my data for training?** No. All calculations are performed locally on your machine. Your research data stays private and is never sent to an external server. **How does the Normality Test Engine MCP handle math?** It uses the simple-statistics library to calculate exact coefficients. This prevents the AI from hallucinating numbers and gives you a mathematically accurate result. **Can I use the Normality Test Engine MCP for non-numeric data?** No, this tool is specifically for numeric data. It calculates skewness and kurtosis, which require numbers to function correctly. **What is a Jarque-Bera test?** It's a statistical test used to determine if data has the properties of a normal distribution. This Connector runs that test for you to give a clear pass/fail verdict. **Why can't I just ask the AI if the data looks normal?** LLMs are not calculators. They might guess correctly sometimes, but they often get it wrong. This Connector provides a deterministic, local calculation you can actually trust. **Is this the Shapiro-Wilk test?** This engine implements the Jarque-Bera normality test, which uses Skewness and Kurtosis. It is highly effective for medium-to-large samples and avoids the Shapiro-Wilk implementation gaps in JavaScript. **How many data points do I need?** The Jarque-Bera test works best with 30 or more samples. For very small samples (n < 20), consider using visual QQ-plot analysis as a complement. **What does a 'not normal' result mean for my analysis?** If your data is not normally distributed, parametric tests like t-tests and ANOVA may be unreliable. Consider using non-parametric alternatives like Spearman correlation or Mann-Whitney U tests.