Product · PII redaction
Find personal data or secrets before they spread.
Scan text or files for personal identifiers or leaked credentials. Redact every finding. Use it as a CI gate or a command-line tool. Or run it as an HTTP service yourself.
See it work
Try it from wherever you work.
1corpusforge scan ./data # exit 2 if anything is found
2corpusforge redact ./data/notes.txt -o notes.redacted.txt
What you get
What PII redaction does.
Checksum-validated identifiers
Emails, IPs, phone numbers, credit cards (Luhn), Australian TFN, ABN and Medicare, plus national IDs and IBANs from over twenty countries.
Secrets, not just people
Private-key blocks, provider API keys, JWTs and credential assignments, with the same detectors behind the CI scan.
Your own names
Add an operator list of people, handles and hosts to redact on top of the built-in detectors.
Optional language model
A contextual model finds names and addresses that patterns cannot, and falls back to the deterministic passes if it cannot load.
How it works
Three steps.
- 1
Detect
Patterns and checksums, secret detectors, your denylist, and optionally a model.
- 2
Merge
Overlapping findings resolve to the longest span.
- 3
Redact
Each span is replaced with its entity type, and offsets are returned.
Limits up front.
Know the limits
- The contextual model inherits its training bias and misses novel, domain-specific and non-English entities. Health, justice and heritage data needs a person in the loop.
- You can self-host the service, or call the hosted endpoints on the Webfold API (
/v1/pii/redactand/v1/pii/scan, in early access). Hosted calls send your text to the API; self-hosting keeps it on your network. - Findings are a first pass, not a guarantee. Offsets are byte offsets into the original text.
Keep reading.
FAQ
Questions.
Does the text leave my network?
Not if you self-host. It is a single static binary with no managed cloud and no data egress. If you call the hosted API instead, the text is sent to it.
Is it a replacement for a secrets scanner?
It includes secret detectors in the same class as trufflehog and gitleaks, shape-only with no live verification, and a scan command for CI.
What licence?
Apache 2.0.
Use PII redaction with Webfold.
Run the command-line tool today. Join the waitlist for the hosted API.