Download latest version of anonymix

Features & Updates in v1.1.*

• Initial stable rollout for the v1 architecture.
• JSON & XML payload support
• Tier 1 regex patterns for sensitive data redaction.
• Tier 2 NER pipelines for sensitive data redaction in English.
• Customizable regex patterns for advanced users.
• New contextual detection tier: finds PII that carries no formatting of its own, such as a bare 5551234567 or 123456789, which the regex tier could only match when punctuated.
• Detections are confirmed by structural validation before redaction — Luhn checksum and issuer prefix for payment cards, issuing-authority rules for US Social Security Numbers, numbering-plan rules for phone numbers, and the ABA checksum for bank routing numbers.
• Ordinary business values are left alone: order numbers, batch references, timestamps, totals and trace ids keep the same shape as PII, so a detection also requires a cue word nearby or in the enclosing field name.
• Field and element names now inform detection, so values with no surrounding prose — such as phone_raw or the entries of a phones array — are recognised.
• New detectors for international phone numbers, bank routing numbers, dates of birth in non-ISO formats, device and advertising identifiers, and driver's licence numbers.
• New patterns for MAC addresses, GPS coordinate pairs and User-Agent strings.
• Fixed acronyms and sentence openers such as SSN, DOB and Contact being wrongly redacted as names or places by the NER tier.
• Fixed the NER tier reading text the regex tier had already rewritten, which shifted entity positions and caused both missed and spurious redactions. All tiers now analyse the original text and their results are combined in a single pass.
• Reduced memory allocations per redacted value, so the new tier adds detection coverage without adding overhead.
• New anonymix.contextual configuration block, with per-detector cue words for domain-specific vocabulary. Detectors are enabled and disabled through the existing built-in pattern lists.
• New contextual tier in the redaction metrics and the X-Anonymizer-Fields-Redacted response header.