portable-datascout

name	portable-datascout
description	Use this when a user asks to inspect, summarize, profile, sample, or infer schema for local CSV/TSV-like data using moonrun and the bundled portable datascout WASM artifact from this skill.

Portable Datascout

Use this skill to turn a local delimited data file into a compact schema and sample report. It is useful before sending data to an agent because it exposes column names, inferred simple types, empty counts, max lengths, and a bounded sample without loading the full file into the conversation.

Quick Start

moonrun /path/to/portable-datascout/assets/datascout.wasm \
  --file data.csv \
  --sample 5

Write JSON for automation:

moonrun /path/to/portable-datascout/assets/datascout.wasm \
  --json \
  --file data.csv

For TSV or semicolon-delimited files:

moonrun /path/to/portable-datascout/assets/datascout.wasm \
  --delimiter tab \
  --file data.tsv

Workflow

1. Run datascout before pasting or packing large data files.
2. Use --sample N to keep examples small.
3. Use --max-rows N to cap scan work for large files.
4. Use --json when another tool should consume the schema.
5. Use --no-header when the first row is data.

Useful Options

• --file PATH: read a guest-visible data file.
• --json: write compact JSON.
• --delimiter auto|comma|tab|semicolon|CHAR: select delimiter.
• --sample N: number of sample rows to include.
• --max-rows N: maximum rows to scan.
• --no-header: synthesize column_1, column_2, ...
• --output PATH: write report to a guest-visible path.

Caveats

This is a lightweight profiler, not a full dataframe engine. CSV quote handling is enough for common files, but it does not stream rows, validate encodings, or run SQL-like queries.