PII detection
Metacrafter tags PII using rules in the pii context (and language-specific
packs such as en, ru, fr). Combine context filters with a confidence
threshold so the report stays useful.
File
metacrafter scan file users.csv \
--contexts pii \
--langs en \
--confidence 20.0 \
--format full \
-o pii_scan.json
JSON Lines:
metacrafter scan file users.jsonl \
--contexts pii \
--langs en \
--confidence 20.0 \
--format full
Database
metacrafter scan sql "sqlite:///users.db" \
--contexts pii \
--langs en \
--confidence 20.0 \
--format full \
-o pii_scan.json
Limit columns when you already know the likely PII fields:
metacrafter scan file users.csv \
--fields email,phone,name,address \
--contexts pii \
--format full
What you get
Each matching field includes:
- rule id and datatype key
- confidence (0–100)
- registry URL, for example
https://registry.apicrafter.io/datatype/email
Rich table output highlights PII matches when writing to a terminal.
Next steps
- Export the JSON report to DataHub or OpenMetadata
- Add country-specific PII with
pip install 'metacrafter[rules]' - Use hybrid LLM mode for columns rules miss