Csv pii redactor

Skill sisodiabhumca/agent-skills/skills/csv-pii-redactor

Production-Ready Agent Skills : product analytics, growth experiments, CRM, research synthesis, postmortems, data contracts, SaaS spend, compliance, architecture maps, and LLM eval and many more.

Install

npx -y skills add sisodiabhumca/agent-skills --skill csv-pii-redactor

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Vendor-neutral skill to detect and redact common PII in CSV files and produce a redaction report.

SKILL.md

1.6 KB, as published. Nobody here has run it

When to invoke

You need to share CSV data externally but want to reduce privacy risk.
You want a quick, explainable scan for likely PII columns/values.

Inputs needed

--input_csv: Path to input CSV.
--output_csv: Path to write the redacted CSV.
--report_json: Path to write a JSON redaction report.
Optional:
- --redact_with: Replacement token (default: [REDACTED]).
- --min_match_rate: Minimum fraction of non-empty values in a column that must match a PII pattern before redaction is applied (default: 0.2).

Workflow

Load CSV and sample values per column.
For each column, run regex-based detectors on non-empty values:
- email addresses
- phone numbers (basic international/US patterns)
- US SSN-like patterns
- IPv4 addresses
- credit card-like numbers (with Luhn check)
Decide whether to redact a column based on match rate threshold.
Produce:
- Redacted CSV (only columns that exceed threshold are redacted).
- JSON report with detectors triggered, match counts, and examples.

Output format

Redacted CSV with same headers/row count as input.
JSON report:
- columns[].name
- columns[].redacted (boolean)
- columns[].detectors[] with type, matches, total_non_empty, match_rate, examples

Guardrails

This is heuristic detection; it may miss PII or flag false positives.
Do not treat output as compliance certification.
Avoid printing raw PII to stdout; report examples are truncated.

Reference code

csv_pii_redactor.py