CSV Profiler Free Claude Code skill, copy & paste

${CLAUDE_SKILL_DIR} resolves to the skill's own folder, so the script is found wherever the skill is installed.

---
name: csv-profiler
description: Profile a CSV file — row count, column types, missing values, distinct counts and numeric ranges. Use when the user shares or asks about a CSV or TSV file.
allowed-tools: Bash(python *)
---

To profile a CSV, run the bundled script rather than writing new code:

```bash
python "${CLAUDE_SKILL_DIR}/scripts/profile.py" path/to/file.csv
```

Then summarise for the user:
- What each column appears to hold
- Columns with missing values, and how many
- Anything suspicious: a numeric column stored as text, a column that
  is all one value, an id column with duplicates

Do not load the whole file into the conversation.

Standard library only, so it runs anywhere Python 3.9+ does — no pandas to install first.

import csv
import sys
from collections import Counter


def as_number(value):
    try:
        return float(value.replace(",", ""))
    except ValueError:
        return None


def main(path):
    with open(path, newline="", encoding="utf-8-sig") as handle:
        sample = handle.read(4096)
        handle.seek(0)
        dialect = csv.Sniffer().sniff(sample, delimiters=",;\t|")
        rows = list(csv.DictReader(handle, dialect=dialect))

    print(f"{path}: {len(rows)} rows")
    if not rows:
        return

    for column in rows[0].keys():
        values = [(row.get(column) or "").strip() for row in rows]
        present = [v for v in values if v]
        numbers = [n for n in map(as_number, present) if n is not None]
        distinct = Counter(present)

        kind = "number" if present and len(numbers) == len(present) else "text"
        line = (f"- {column}: {kind}, {len(values) - len(present)} missing, "
                f"{len(distinct)} distinct")

        if kind == "number":
            line += f", min {min(numbers):g}, max {max(numbers):g}"
        elif distinct:
            top, count = distinct.most_common(1)[0]
            line += f", most common {top!r} ({count})"

        print(line)


if __name__ == "__main__":
    if len(sys.argv) != 2:
        sys.exit("usage: profile.py <file.csv>")
    main(sys.argv[1])

Keep the folder layout — SKILL.md at the top, the script under scripts/.

mkdir -p ~/.claude/skills/csv-profiler/scripts
mv SKILL.md ~/.claude/skills/csv-profiler/
mv profile.py ~/.claude/skills/csv-profiler/scripts/

# then, inside Claude Code:
#   what's in sales-2026.csv?      (Claude loads the skill itself)
#   /csv-profiler sales-2026.csv   (or call it directly)