Data scientist
Agent Plugins 1.0.0 compatible
Turn data into decisions with sound exploratory analysis, hypothesis tests, A/B test readouts, precise metrics, and clear charts.
Version 1.0.2Updated Oct 20, 2026Source
First time? Install with npm i -g harnesstap
Download as Agent Plugins zipWhat's inside
Skills (6)
ab-test-readoutAnalyze a finished A/B test and write a decision memo: sanity checks, primary metric effect with confidence interval, guardrails, segments, and a ship or no-ship recommendation. Use when an experiment reaches its planned duration or sample size, or when someone asks \"did the test win?\".decision-chartDesign one chart that answers one stakeholder question clearly, with the right chart type, honest scales, direct labels, and a takeaway title. Use when presenting analysis results, building a slide or report figure, or when a chart is being misread.exploratory-analysisRun a focused exploratory data analysis that starts from a question, checks data quality, and ends with findings and next questions. Use when given a new dataset, a vague question like \"what is going on with churn?\", or before modeling. Produces a notebook and a short findings summary.hypothesis-testingChoose and run the right statistical test for a question, check its assumptions, and report effect sizes with confidence intervals in plain language. Use when comparing groups, testing whether a difference is real, or validating a finding from exploration.metric-definitionWrite an unambiguous metric definition covering purpose, formula, grain, filters, time window, edge cases, owner, and reference SQL, so every dashboard computes the same number. Use when introducing a KPI, when two teams report different numbers for the same metric, or before an experiment uses a new metric.notebook-hygieneKeep analysis notebooks reproducible, readable, and reviewable: ordered cells, pinned environment, parameters at the top, no hidden state, and clean diffs. Use when starting a notebook, before sharing one, or when a notebook only works if cells are run in a special order.
Rules (3)
chart-accessibilityMake every chart readable for colorblind and low-vision readers.data-ethicsUse personal data responsibly and report results honestly.reproducible-seedsMake every analysis reproducible.
Commands (2)
export-chartExport a chart from a notebook as a publication-ready figure with alt text and caption.render-notebookRe-run a notebook from a clean kernel and render it to HTML for sharing.
Plugins (3)
fullstack-dev-skills@fullstack-dev-skillscontext7@claude-plugins-officialdrawio-diagramming@claude-orchestration