Category

data Agent Skills

9 open data skills for AI agents. Each package is a portableSKILL.md folder with trigger guidance, instructions, and a benchmark suite — installable into Claude Code, Codex, Cursor, and OpenCode.

  • Analyze A/B and experiment results with correct metrics, uncertainty, and caveats. Use whenever the user shares experiment data, asks if a variant won, or needs help interpreting lift, p-values, confidence intervals, CUPED, or sample ratio mismatch.

    Category
    data
    Context
    377
    Updated
    2026-07-26
  • Investigate metric and log anomalies with structured hypotheses and next checks. Use whenever the user sees a spike, drop, or weird pattern in metrics or logs and asks what is going on, whether it is real, or how to investigate.

    Category
    data
    Context
    257
    Updated
    2026-07-26
  • csv-analysisCommunity

    Analyze CSV, TSV, and tabular data files — profile the data, answer questions with verifiable numbers, and flag quality issues. Use whenever the user uploads or pastes tabular data, asks "what does this data show", wants totals, trends, comparisons, or charts from a spreadsheet-like file, or asks any question that must be answered from rows and columns.

    Category
    data
    Context
    961
    Updated
    2026-07-26
  • data-cleaningCommunity

    Profile and clean messy tabular datasets with reproducible transformations and explicit handling of types, missing values, duplicates, and outliers. Use whenever the user has inconsistent CSV or spreadsheet data, asks to prepare data for analysis, merge records, standardize columns, remove duplicates, or fix data-quality problems.

    Category
    data
    Context
    540
    Updated
    2026-07-11
  • Define producer-consumer data contracts with schema evolution, compatibility, and SLAs. Use whenever the user designs event schemas, warehouse contracts, schema registry rules, or breaking-change policy between data producers and consumers.

    Category
    data
    Context
    308
    Updated
    2026-07-26
  • Choose and produce clear, honest charts from tabular or summarized data. Use whenever the user asks for a chart, dashboard, plot, visual comparison, trend view, or advice on how to visualize data, including requests to improve an existing graph or select the right chart type.

    Category
    data
    Context
    579
    Updated
    2026-07-11
  • Design ETL and ELT pipelines with clear contracts, quality checks, and failure modes. Use whenever the user plans data ingestion, transformations, warehouse loads, dbt-style models, batch/stream pipelines, or data pipeline architecture.

    Category
    data
    Context
    300
    Updated
    2026-07-26
  • Define product and business metrics with precise numerators, denominators, and edge cases. Use whenever the user asks what to measure, how to define a KPI, North Star, activation, retention, churn, or event taxonomy for analytics — even for a rough product idea.

    Category
    data
    Context
    378
    Updated
    2026-07-26
  • sql-analyticsCommunity

    Write analytical SQL for warehouses and BI questions with correct grain and caveats. Use whenever the user asks for warehouse SQL, BigQuery, Snowflake, Redshift, DuckDB analytics, funnel metrics in SQL, or reporting joins — as opposed to OLTP query performance tuning alone.

    Category
    data
    Context
    294
    Updated
    2026-07-26