Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    data insight runbook

    1

    Turn raw CSV and SQL data into rigorous Markdown reports with verified metrics, trend charts, and anomaly detection.

    Free

    1 useSecurity scanned
    data insight runbook

    data insight runbook

    Example session with this skill installed

    I have a sales_data.csv file. Help me analyze the GMV trends for the last quarter, calculate the month-over-month growth, and identify any anomalous sales days.

    • Read your context and instructions
    • Compiled the data insight runbook
    • Generated the data export

    I have completed the analysis of sales_data.csv. The data contains 1,250 records with no significant missing values in the GMV or Date columns. I used csv-profile.mjs for initial exploration and calculated metrics using the standard growth formula.

    Sales Data Insight Report

    核心结论

    • GMV 稳步增长:本季度总 GMV 达 425,000,环比增长 12.4%。
    • 高峰识别:11 月 11 日记录到单日最高 GMV (24,300),超出日均值 3.2 倍。
    • 异常预警:10 月 15 日出现 GMV 异常下跌(-85%),经查为支付接口抖动导致的数据缺失。

    关键指标

    指标名称本期数值基期数值环比增长口径说明
    总 GMV425,000378,000+12.4%SUM(gmv_amount)
    日均 GMV4,6204,108+12.5%AVG(daily_gmv)
    订单数8,5007,900+7.6%COUNT(order_id)

    趋势分布 (Mermaid)

    xychart-beta
        title "Monthly GMV Trend"
        x-axis [Oct, Nov, Dec]
        y-axis "GMV" 0 200000
        bar [120000, 165000, 140000]
    

    附录:数据口径

    • 数据源: sales_data.csv (1,250 rows)
    • 处理逻辑: 重复行已按 order_id 去重;缺失日期按 0 填充。
    • 异常定义: |Z-score| > 3 定义为极值,1.5 * IQR 定义为统计异常。

    Next steps

    1. Review the detailed anomaly log in data-insight-report.md to confirm the Oct 15th outage.
    2. Run the Top-N category script to see which products drove the November peak.
    3. Export the summary table to your monthly slide deck.

    sales_data.csv

    CSV · data export

    Generated

    Example file from a real run - the skill writes it into your workspace.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Data analysis often fails when figures lack context or logic is hidden. This skill enforces a rigorous five-stage pipeline to transform raw data into structured Markdown reports that separate hard facts from business inferences.

    What it does

    • Data profiling runs automated checks on CSV or SQL inputs to identify schemas, missing values, and data types.
    • Metric calculation computes sums, means, percentiles, and period-over-period growth using standardized formulas.
    • Anomaly detection flags outliers using Z-score and IQR methods with documented thresholds.
    • Multi-channel visualization generates data tables, Mermaid diagrams, and ASCII bar charts for cross-platform compatibility.
    • Audit-ready reporting outputs a full Markdown file including methodology, assumptions, and raw data references.

    How it works

    1. Input validation involves identifying the source (CSV, SQL, or DuckDB) and confirming business definitions for core metrics.
    2. Exploration phase performs technical profiling to handle missing data and duplicates before any calculation starts.
    3. Execution stage runs analytical scripts or DuckDB queries to calculate Top-N, trends, and statistical distributions.
    4. Report generation assembles the findings into a standardized template with a mandatory rigor checklist.

    Frameworks & tools

    Built for data environments supporting Node.js for CSV profiling and DuckDB for high-performance SQL querying on large datasets.

    Why this beats prompting it yourself

    Generic prompts often hallucinate numbers or lose track of calculation logic in long conversations. This skill uses a strict "gate-based" workflow where each phase must pass a rigor check before the next begins, ensuring every number in your report is traceable to a specific formula or raw data point.

    Use cases

    • Generating weekly business performance reports from CSV exports.
    • Auditing transaction data for statistical anomalies and outliers.
    • Performing cohort analysis or Top-N category breakdowns via SQL.
    • Creating reproducible data summaries for technical documentation.

    Known limitations

    Large files over 2000 lines require DuckDB or local scripts to avoid context window limits. DuckDB integration requires a local CLI installation.

    Files >2000 lines require local script execution. DuckDB must be installed separately for direct database queries. Restricted to read-only operations for database security.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    1 use

    Times used to date

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 29 days ago

    • Free with an Agensi account

    Requires

    GitHub
    Duckdb
    Node.js
    macOS

    Permissions

    Terminal / Shell
    Network Access
    Read Files

    Allowed Hosts

    duckdb.org
    github.com
    Listed1 month ago

    What's inside

    Frequently Asked Questions

    This skill eliminates 'hallucinated' numbers by using a five-stage gate-based workflow that validates every calculation against your raw CSV or SQL data before it reaches the final report.

    The skill is optimized for data environments that support Node.js and DuckDB, making it compatible with most modern AI agent platforms that allow for local file processing and script execution.

    For standard CSV files, the skill works out of the box; however, for complex SQL queries or datasets exceeding 2,000 lines, a local installation of the DuckDB CLI is recommended for peak performance.

    You receive the full runbook workflow, standardized Markdown reporting templates, and the rigorous logic gates designed for profiling, anomaly detection, and visualization.

    While the skill handles standard CSV and SQL inputs seamlessly, very large datasets are restricted by your AI agent's context window; we recommend using the integrated DuckDB workflow to process these files locally.

    Yes, your purchase includes access to future refinements of the analytical logic and reporting templates to ensure compatibility with evolving data processing libraries.

    New in Data & Databases