csv-stats-reporter

name	csv-stats-reporter
description	Compute statistical summaries (mean, median, p95, p99, min, max) for numeric columns in a CSV file and output a JSON report.
version	1.0.0

CSV Stats Reporter

Skill Summary

Reads a CSV file and computes descriptive statistics for all numeric columns. Produces a JSON report containing per-column metrics that is useful for monitoring, anomaly detection, and dashboarding.

Inputs

Parameter	Flag	Required	Description
input	-i	Yes	Path to the input CSV file
output	-o	Yes	Path to the output JSON report
columns	--columns	No	Comma-separated list of columns to analyse (default: all)
group-by	--group-by	No	Column name to group statistics by

Computed Metrics

For each numeric column the following metrics are computed:

• count — number of non-null values
• mean — arithmetic mean
• median — 50th percentile
• std — standard deviation
• min / max — extremes
• p95 — 95th percentile
• p99 — 99th percentile

Output

A JSON file structured as:

{
  "source": "<input_file>",
  "total_rows": 1234,
  "columns_analysed": ["response_time_ms", "status_code"],
  "stats": {
    "response_time_ms": {
      "count": 1200,
      "mean": 245.3,
      "median": 180.0,
      "std": 312.7,
      "min": 12,
      "max": 4500,
      "p95": 890.0,
      "p99": 2100.0
    }
  }
}

When --group-by is used, stats becomes a nested dict keyed by group value.

Implementation

python3 ./skills/csv-stats-reporter/stats_reporter.py -i <input.csv> -o <report.json> \
    [--columns col1,col2] [--group-by col_name]

Dependencies

• Python 3.10+ (standard library only)