Reported issues for umbryn-mcp
Pod holds 10 of 10 problems reported by people outside the maintainer team. Issues filed by the project's own owners, members and collaborators are excluded entirely — a maintainer's release checklist is not a warning to a prospective user.
Back to umbryn-mcp.
Most discussed
Add a US Passport number recognizer (context-gated, no checksum)
Add a recognizer: US Passport number
Great first contribution! This is one of the friendliest recognizers to add: there's no check-digit math to implement, so it's mostly a regex + a context gate + tests. Everything you need to implement it is spelled out below — no further research required.
What & why
A US passport number is a nine-character government identifier that shows up in travel records, KYC/onboarding forms, visa paperwork, and support tickets — exactly the kind of PII t
Read the thread · 2026-07-06 · open · 3 comments
Add an Australian Tax File Number (TFN) recognizer (mod-11 weighted checksum)
Add an Australian Tax File Number (TFN) recognizer
Great first issue! This is a self-contained recognizer with a real (and satisfying) check digit, so it follows the exact same shape as the UK NHS and Canadian SIN recognizers already in the tree. If you've been looking for a first PR here, this is a good one to claim.
What & why
A Tax File Number (TFN) is the personal identifier the Australian Taxation Office issues to individuals and organisations — the Australian equivalent o
Read the thread · 2026-07-06 · open · 2 comments
Add an IBAN (International Bank Account Number) recognizer
What & why
The default (zero-dependency) engine detects clinical identifiers and common PII, but it does not detect IBANs (International Bank Account Numbers). IBANs are financial PII that regularly appear in healthcare billing, insurance, and patient-payment text, so they belong in the redaction set. Presidio's engine already ships an IBAN_CODE recognizer, but the default regex engine has no equivalent — so a user on the zero-dependency default gets no IBAN coverage. This issue cl
Read the thread · 2026-07-04 · closed · 1 comment
Add a US Driver's License number recognizer (context-gated starter)
What & why
Add a recognizer for US Driver's License (DL) numbers to the default (zero-dependency) engine. A DL number is a direct HIPAA identifier (45 CFR §164.514(b)(2)(i)(J) — "certificate/license numbers") and shows up constantly in intake forms, incident notes, and identity-verification text, so redacting it before text reaches an LLM is high-value.
This is deliberately a starter recognizer. US DL formats vary by state and there is **no single nationwide format and no universa
Read the thread · 2026-07-04 · closed · 1 comment
Add a Legacy Medicare HICN (Health Insurance Claim Number) recognizer
What & why
Add a recognizer for the legacy Medicare HICN (Health Insurance Claim Number) — the SSN-based Medicare identifier that was printed on Medicare cards before the MBI transition completed on 2020-01-01. HICNs no longer appear on new cards, but they are still all over historical records — old claims, EOBs, redetermination notices, legacy EHR exports, spre
Read the thread · 2026-07-04 · closed · 1 comment
Add a US ITIN (Individual Taxpayer Identification Number) recognizer
What & why
An ITIN (Individual Taxpayer Identification Number) is a US tax ID the IRS issues to people who need a taxpayer identifier but are not eligible for an SSN. It is a direct-identifier under HIPAA's Safe Harbor "account/certificate/license number" and "any other unique identifying number" clauses, and it appears in exactly the kinds of intake, billing, and tax documents phi-redact-mcp is meant to scrub. We currently detect SSNs but not ITINs, so a document like `Taxpayer ITIN
Read the thread · 2026-07-04 · closed · 1 comment
Add a Canadian Social Insurance Number (SIN) recognizer
What & why
Add a recognizer for the Canadian Social Insurance Number (SIN) — the 9-digit identifier Service Canada issues for tax, benefits, and employment. It is squarely PII (and PHI in Canadian clinical/insurance contexts), so it belongs in the default engine alongside US_SSN. It has a Luhn mod-10 check digit, so we can validate it and give it a high, precise score rather than relying on shape alone.
Good news for the implementer: the SIN's check digit is a **plain Luhn over all
Read the thread · 2026-07-04 · closed · 1 comment
Add a UK NHS Number recognizer (10 digits, mod-11 check digit)
What & why
The NHS Number is the unique patient identifier used across the UK's National Health Service. It is a direct patient identifier and clearly PHI/PII, so it belongs in the default recognizer set alongside our other clinical identifiers (NPI, DEA, MBI, CLIA).
Crucially, it carries a mod-11 check digit, which puts it in our high-precision "checksum-backed" tier: a match that fails the check digit is discarded outright, not merely down-scored. This lets us score it high on shape
Read the thread · 2026-07-04 · closed · 1 comment
Most recent
Add a Brazilian CPF recognizer (11 digits, two mod-11 check digits)
Add a Brazilian CPF recognizer (checksum-backed)
Good first issue · Adds one new checksum-backed recognizer. No changes to the engine or public API — just a regex, a small validator, an entity constant, and tests. A great way to learn the recognizer pattern end to end.
What & why
The CPF (Cadastro de Pessoas Físicas) is Brazil's individual taxpayer registry number, issued by the Receita Federal (the federal tax authority). It's the primary national identifier for a private in
Read the thread · 2026-07-06 · open · 0 comments
Add a Spanish DNI/NIE recognizer (mod-23 check letter)
What & why
The Spanish DNI (Documento Nacional de Identidad) is the national ID number for Spanish citizens; the NIE (Número de Identidad de Extranjero) is its equivalent for foreign residents. Both are ubiquitous personal identifiers across Spanish healthcare, tax, and banking records, and both are exactly the kind of national-ID PII this project aims to redact. They're a great fit for our precision-first approach because each carries a mod-23 check letter: we can run a vali
Read the thread · 2026-07-06 · open · 0 comments
The remaining reports are on the project's issue tracker.