Redaction

Redacta Connector Risk

Assess and monitor the risk of the Redacta connector in Anthropic Claude, OpenAI Codex, and Microsoft Copilot.

Pseudonymise patient identifiers and PII in text, and restore them — on-device, nothing leaves your machine.

Viewing Details For

Connector Details

Redacta replaces patient identifiers and personal data (NHS numbers with Modulus-11 validation, names, dates of birth, MRNs, emails, phones, postcodes, plus general PII like URLs, IPs, payment cards and account numbers) with labelled tokens, so text can be safely shared or processed by AI. It includes a HIPAA Safe Harbor mode and is fully reversible via a token map. All processing happens locally in the server process — it makes no network calls and stores nothing.

Type: Local
Works With: claude-desktop
Categories: redaction, pseudonymisation, de-identification, pii, phi, healthcare, nhs, hipaa, privacy
Visibility: commercial
Added: 6/19/2026

Documentation Support

Tools the Redacta Connector Exposes to Anthropic Claude

redact

reinstate

self_check

Want a live walkthrough of this connector?

PromptArmor tracks connector risk across Claude, Codex, Copilot, and other MCP-enabled platforms. Book a demo to see the live connector risk dashboard for your environment.

Book a demo