Log inSubmit server
Redacta MCP logo

Redacta MCP

Locally pseudonymises clinical text, restores identifiers from token maps, and supports HIPAA Safe Harbor processing.

InstallRepositoryWebsite
Full README Compare alternatives Embed widget Claim this listing

Install Redacta MCP

Pick your client, then add it in one click or paste the config.

STDIONode.js
~/Library/Application Support/Claude/claude_desktop_config.json
claude_desktop_config.json
{
  "mcpServers": {
    "redacta-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "redacta-mcp"
      ]
    }
  }
}

Settings → Developer → Edit Config, merge this into mcpServers, then fully quit and reopen Claude Desktop.

We couldn’t confirm it starts correctly

The install command started, but didn't respond the way we expected when we tried to talk to it.

npx -y redacta-mcp

initialize succeeded but no response to tools/list.

This is an experimental automated check and can have false negatives — missing environment variables, a slow cold install, etc. It doesn’t necessarily mean something’s wrong. Last checked Today.

Overview

  • healthcare
  • privacy
  • pii-redaction
  • mcp

The redacta-mcp MCP server pseudonymises medical and clinical text locally before it is sent to an AI tool. It replaces detected identifiers with labelled tokens, returns a redaction report, and can restore original values from a token map after processing. Deterministic patterns cover common UK and US identifiers, while Safe Harbor mode applies stricter HIPAA-oriented removal. Reach for it when identifiable clinical text must stay within the local processing boundary.

Use cases

  • Redact patient identifiers before sending clinical text to an AI tool
  • Restore original identifiers after an AI workflow completes
  • Apply HIPAA Safe Harbor processing to US healthcare text
  • Review detected identifiers through the generated redaction report

Key features

  • Local pseudonymisation of clinical text
  • Reversible token mapping for re-identification
  • Deterministic UK and US identifier detection
  • HIPAA Safe Harbor mode
  • Redaction report generation
  • Self-check for identifiers left in output

Security & health

What it can reach and how reliably it runs — from our own checks, not the maintainer’s claims.

ActiveLast checked Today

Recent health check succeeded.

Supply-chain signal

No high-severity advisories surfaced by our automated scan.

Critical 0High 0Medium 0Low 0

Scanned 11d ago via OSV.dev · redacta-mcp (npm)

Adoption & maintenance

Factual signals from GitHub, npm, and our automated checks — not a rating.

GitHub stars8Stargazers on the source repository.
npm downloads233Package downloads in the last 30 days.
Last commit16d agoMost recent push to the default branch.
Install checkInconclusiveStarted but did not complete the MCP handshake in our test.
Directory activity1 viewsConfig copies, upvotes, and views on AllMCPs.

How Redacta MCP works

What redacta-mcp MCP server does

The redacta-mcp MCP server provides local redaction and re-identification for medical and clinical documents. It replaces patient and other personally identifiable information with stable, labelled placeholders such as [PATIENT_NAME_1], [NHS_NUMBER_1], and [DATE_OF_BIRTH_1]. The cleaned text preserves the surrounding clinical meaning, and the result includes a redaction report for review.

Redaction is designed as the first step in a round trip: redact text before sending it to another AI tool, then restore the original values locally after processing. Restoration uses the token map produced during the initial operation, so the original identifiers do not need to cross the privacy boundary.

The server uses the shared Redacta TypeScript detection engine. It is one of several Redacta surfaces, alongside libraries, a command-line tool, an iOS app, an agent skill, a self-hosted HTTP service, and a FigJam plugin.

How it works

The deterministic layer identifies fixed-format values using local processing and no network access. Supported patterns include validated UK NHS numbers, UK National Insurance numbers, dates of birth, UK postcodes, phone numbers, email addresses, hospital or MRN numbers, US Social Security numbers, and US ZIP codes.

The broader Redacta project also describes a reasoning layer for identifiers that patterns cannot reliably classify, including patient names, relatives, carers, postal addresses, and identifying ages. A final self-check looks for identifiers that may have remained in the output before the report is produced.

Safe Harbor mode applies a stricter pass for US de-identification. In that mode, all dates and specific ages are handled, along with additional HIPAA identifier categories such as fax numbers, certificate or licence numbers, device serial numbers, VINs, and health-plan beneficiary numbers.

Setup and configuration

Install and run the redacta-mcp MCP server with the package command shown in the project documentation:

Terminal
npx -y redacta-mcp

The README identifies Claude Desktop and Cursor as MCP clients for this surface. The processing model is local; organisations needing an HTTP deployment can use the separate gateway service and its Kubernetes manifests instead.

Tools and capabilities

The redacta-mcp MCP server supports the following operations:

  • Redact clinical or medical text into labelled pseudonymous tokens.
  • Produce a redaction report alongside the cleaned text.
  • Reinstate original identifiers from a prior token map.
  • Apply the stricter HIPAA Safe Harbor or US de-identification mode.
  • Keep identifier handling on the local machine during the redact-process-restore workflow.

Limitations and notes

Redacta is not presented as a complete guarantee that every identifier will be found. Its documentation describes it as a first line of defence and advises reviewing the redaction report before sharing text. It should not replace formal data-protection processes.

Detection coverage varies by identifier type. The deterministic layer handles specified structured patterns, while less structured information such as names and addresses depends on the broader reasoning workflow.

The redacta-mcp MCP server is therefore most suitable when local pseudonymisation and reversible token mapping are required, provided the resulting report and output are reviewed before disclosure.

Read the full README View source on GitHub

Reviews

No reviews yet — be the first to share how it worked for you.

Frequently Asked Questions about Redacta MCP

redacta-mcp is an MCP server that runs local redaction and reinstatement for medical and clinical text, connecting MCP clients such as Claude Desktop and Cursor to tools that replace identifiers with labelled tokens and restore originals from a token map. It supports deterministic identifier patterns and a HIPAA Safe Harbor mode.

Install it with `npx -y redacta-mcp`.

Yes, redacta-mcp is listed for Cursor.

Yes, it can restore original values from the token map created during redaction. This supports a local redact, process, and re-identify workflow.

Yes, Safe Harbor mode applies stricter US de-identification rules, including handling for all dates, specific ages, and additional HIPAA identifier categories.

Related MCP servers

All in Biology & Bioinformatics All alternatives
Add a badge to your READMEShow it’s listed — live SVG, Markdown or HTML.

AllMCPs Directory Badge

Full Badge Customizer

Showcase your server listing on GitHub or your project documentation. Embed this dynamic SVG badge to highlight official listing status and live engagement.

Badge Style:
Live Dynamic SVG PreviewRedacta MCP AllMCPs Directory Badge
Markdown (GitHub README)
[![AllMCPs](https://allmcps.com/api/badge/redacta-mcp?style=directory)](https://allmcps.com/mcp/redacta-mcp)
HTML Embed
<a href="https://allmcps.com/mcp/redacta-mcp"><img src="https://allmcps.com/api/badge/redacta-mcp?style=directory" alt="Redacta MCP on AllMCPs" /></a>