Data analysis skills
Data Analysis Skills-md for AI Agents
Data analysis skills are SKILL.md files that help Claude, Codex and other AI agents understand a dataset before anyone acts on it: profiling its structure and quality, summarising variables, measuring relationships and reading distributions correctly. They give the agent decision rules, thresholds and verification steps, not vague advice.
6 of 6 skills published, version 1.0.0.
Where to start: exploratory-data-analysis, then correlation-analysis, descriptive-statistics and data-distributions.
Exploratory Data Analysis
Runs a structured first pass over a new tabular dataset so an AI agent understands its shape, quality and main patterns before any modelling, reporting or decision.
Correlation Analysis
Measures and interprets the relationship between two variables correctly: it plots the pair first, picks the right coefficient (Pearson, Spearman or Kendall) from the data's shape, reports the strength with a sample size and a confidence interval or p-value, checks for confounders and Simpson's paradox, and states in plain words what the number does and does not mean.
Descriptive Statistics
Summarises one variable honestly: it reads the shape, then reports the centre and spread that fit (median and IQR for skewed data, mean and standard deviation only when symmetric), with the sample size, missing count and five-number summary.
Data Distributions
Characterises the shape of one numeric variable: it plots the distribution with a bin rule that fits, reads modality, skew and tails, quantifies skewness and excess kurtosis correctly (kurtosis is tails, not peakedness), and judges normality from a QQ plot rather than a large-sample p-value.
Grouping and aggregation
Compares a metric across groups honestly: it ties the aggregate to the question, computes a rate as a ratio of sums rather than an average of averages, reports every group with its size, weights overall figures correctly, and checks for Simpson's paradox before trusting the comparison.
Time-series exploration
Runs a careful first look at data indexed by time before any forecast: it parses the timestamp, confirms the real frequency and finds gaps, plots the raw series, resamples with the aggregate that fits a flow or a stock, reads the trend with rolling windows, separates trend from seasonality from noise, and watches timezone and daylight-saving traps.
No skill matches that search. Clear it or choose All.
What data analysis skills do for an AI agent
A general model can discuss data analysis, but left to itself it will reach for the mean on skewed data, trust a correlation it never plotted, or read a normality test on fifty thousand rows as if the p-value settled the question. These skills give the agent the discipline a careful analyst uses: look at the shape before choosing a statistic, plot a relationship before trusting its coefficient, and state plainly what a number does and does not support. Each one is a plain Markdown file the agent loads when the work calls for it.
Where to start
Begin with exploratory data analysis, the structured first pass over any new dataset. Add correlation analysis for two-variable work, descriptive statistics for summarising a single variable, and data distributions for reading its shape. Together they cover the first questions you ask of almost any dataset, and each one hands off to the next instead of repeating it.
What the agent needs from you
Point the agent at the data and say what one row represents, the type of each variable, and the decision behind the question. The skills ask for anything missing rather than guessing, because the grain of the data and the type of each variable change which method is correct.
Frequently asked questions
What are data analysis skills for AI agents?
They are SKILL.md files that turn a general model into a careful analyst for one task each. The agent loads the file, follows its workflow and decision rules, and runs the checks it names before answering.
Do I need to know statistics to use them?
No. The skills carry the judgement, such as when the median beats the mean or when a correlation is misleading. You supply the data and the question; the agent applies the method and explains what it found.
Which skill should I install first?
Exploratory data analysis, because it profiles any new dataset and points you to the next step. Install the others alongside it so the handoffs between them work.
Do they work in Claude, Cursor and Codex?
Yes. Each is a plain SKILL.md you drop into the tool’s skills folder. The description decides when it loads, so once installed the right skill fires from everyday wording.
Sources and more data skills
Each skill lists its own sources with the date it was last checked, drawn from the official documentation of the tools it uses, such as pandas, NumPy and SciPy, and from standard statistical references. Browse the other Data subcategories as they are published, or see the full Skills-md library.
