⚠
High Risk:This skill has significant security concerns. Review the findings below before installing.

sheetsmith

⚠Caution·Scanned 2/17/2026

Sheetsmith is a pandas-based CLI for inspecting, filtering, transforming, and converting CSV/Excel files. It executes user-supplied pandas --expr/--query using df.eval(..., engine="python") and df.query(..., engine="python"), which allows arbitrary Python expression execution.

from clawhub.ai·vb74f617·22.6 KB·0 installs
Scanned from 1.0.1 at b74f617 · Transparency log ↗
$ vett add clawhub.ai/crimsondevil333333/sheetsmithReview security findings before installing

Sheetsmith

Sheetsmith is the pandas-based CSV/TSV/Excel assistant for OpenClaw. It gives you one CLI for:

  • Inspecting spreadsheets (summary, describe, preview)
  • Filtering or sampling rows via pandas queries (filter)
  • Transforming columns with expressions, renames, drops, and safe writes (transform)
  • Converting between CSV, TSV, and Excel formats (convert)

The script lives at skills/sheetsmith/scripts/sheetsmith.py and is the single source of truth for all operations, so you never have to re-write pandas boilerplate.

Dependencies (already installed)

Sheetsmith relies on the Debian-packaged stack:

  • python3-pandas (dataframes + Excel/CSV readers)
  • python3-openpyxl + python3-xlrd (Excel file support)
  • python3-tabulate (pretty Markdown previews)

Because these are installed system-wide, you can run the CLI without building a virtualenv.

Usage

  1. Place the file somewhere under the workspace (e.g., workspace/inputs/my-data.xlsx).
  2. Run a command: python3 skills/sheetsmith/scripts/sheetsmith.py <command> <path> plus any flags listed below.
  3. Inspect the output (Markdown tables are provided for previews) and, when you write data, use --output new.xlsx or --inplace to persist it.

Common commands

CommandDescriptionExample
summaryPrint shape/dtypes, missing-value info, and a Markdown preview of the first/last rows... summary inputs/sales.csv --rows 5 --tail
describeRun DataFrame.describe() with optional --include/--percentiles... describe data.xlsx --include all --percentiles 10 90
previewShow only head/tail rows without any analysis... preview report.tsv --tail --rows 3
filterApply a pandas query string and optionally sample/write results... filter data.csv --query "state == 'CA'" --output outputs/ca.csv
transformAdd/rename/drop columns via pandas expressions, then preview or save... transform data.csv --expr "density = population / area" --rename active:is_active --output outputs/with-density.csv
convertRe-export to CSV/TSV/XLSX by specifying --output with the desired extension... convert raw.xlsx --output clean.csv

Filtering & transforming tips

  • Query strings use pandas syntax: "region == 'EMEA' and sales >= 1e5".
  • Use --sample <n> on filter to inspect a random subset without overwhelming the session.
  • Provide --expr expressions (star column = formula) to create calculated columns, then rename/dropping as needed before writing.
  • Write to a new file with --output; --inplace only overwrites when you explicitly request it.

Automation & repeated work

You can chain commands manually:

  1. filter the rows you need
  2. transform to add/rename/drop columns
  3. convert to output the final format

Every step shares the same CLI, so scripts and workflows stay consistent.

Handling files from humans/bots

  1. Receive the attachment (CSV/Excel) and save it into the workspace (I usually place it under workspace/inputs/).
  2. Run Sheetsmith pointing to that path. For example: 
    python3 skills/sheetsmith/scripts/sheetsmith.py summary workspace/inputs/inbox.xlsx --rows 5
  3. Share results back in chat. If a modified file is needed, use --output to write it into workspace/outputs/ and upload that file (I can send it back via Telegram or WhatsApp).

If you want me to keep a log of every dataset I touched, I can update memory entries as part of the workflow.

Testing

Run the unit tests with:

python3 -m unittest discover skills/sheetsmith/tests

They exercise the summary/preview workflows, filter, and transform commands using tests/data/test.csv, so you can trust the CLI on small but representative data.

Publishing & development notes

  • Skill metadata: SKILL.md explains triggers and workflows.
  • Additional reference: references/usage.md contains a cheat sheet plus troubleshooting notes.
  • Packaging script: python3 $(npm root -g)/openclaw/skills/skill-creator/scripts/package_skill.py skills/sheetsmith creates sheetsmith.skill for ClawHub or release bundles.

Links