Quickstart

2. curl

curl -X POST https://eu-pii-redaction.p.rapidapi.com/redact \
  -H "content-type: application/json" \
  -H "X-RapidAPI-Host: eu-pii-redaction.p.rapidapi.com" \
  -H "X-RapidAPI-Key: $RAPIDAPI_KEY" \
  -d '{"text": "Dear Mr Hansen, please call me on +45 32 12 34 56."}'

Response:

{"redacted": "Dear Mr [PERSON_1], please call me on [PHONE_1].",
 "entities": [{"type": "PERSON", "start": 8, "end": 14, "placeholder": "[PERSON_1]"},
              {"type": "PHONE", "start": 34, "end": 49, "placeholder": "[PHONE_1]"}]}

3. Python

pip install eu-pii-redact

from eu_pii_redact import Redactor

with Redactor() as r:                     # reads RAPIDAPI_KEY
    redaction = r.redact(ticket_text)
    answer = ask_llm(redaction.text)      # the model only sees placeholders
    print(redaction.restore(answer))      # real values back in the answer

Texts longer than 50,000 characters are split automatically, with consistent placeholders. Package: eu-pii-redact on PyPI.

4. JavaScript / TypeScript

// Node 18+ (or any runtime with fetch). No package needed.
const HOST = "eu-pii-redaction.p.rapidapi.com";

async function redact(text) {
  const res = await fetch(`https://${HOST}/redact`, {
    method: "POST",
    headers: {
      "content-type": "application/json",
      "X-RapidAPI-Key": process.env.RAPIDAPI_KEY,
      "X-RapidAPI-Host": HOST,
    },
    body: JSON.stringify({ text }),
  });
  if (!res.ok) throw new Error(`Redaction failed: ${res.status}`);
  const { redacted, entities } = await res.json();
  const originals = {};
  for (const e of entities) originals[e.placeholder] ??= text.slice(e.start, e.end);
  return { redacted, originals };
}

function restore(text, originals) {
  // Longest placeholder first, so [PERSON_1] never touches [PERSON_10].
  return Object.keys(originals)
    .sort((a, b) => b.length - a.length)
    .reduce((out, p) => out.split(p).join(originals[p]), text);
}

const { redacted, originals } = await redact("Hi, my name is Karina Dahl. Refund to DK50 0040 0440 1162 43.");
console.log(redacted); // send this to your LLM
console.log(restore("Hi [PERSON_1], we refunded [IBAN_1].", originals));

5. LangChain

pip install "eu-pii-redact[langchain]"

from eu_pii_redact.langchain import protect

safe_llm = protect(llm)                   # any model or chain that takes a string
safe_llm.invoke("Draft a reply to Karina Dahl about her refund")

6. n8n

In n8n: Settings → Community Nodes → Install and enter n8n-nodes-eu-pii-redact. Put EU PII Redaction → Redact before your AI node and Restore after it.

7. AI assistants (MCP)

claude mcp add eu-pii-redact -e RAPIDAPI_KEY=your-key -- uvx eu-pii-redact

For Claude Desktop, download the one-click bundle. The redact_file tool redacts files without their contents ever entering the conversation.

Options

• "types": ["EMAIL", "IBAN"]: only some kinds of data (PERSON, EMAIL, PHONE, IBAN, CREDIT_CARD, EU_VAT, NATIONAL_ID, IP_ADDRESS).
• "mode": "mask": asterisks instead of placeholders.
• Each entity has start/end offsets in your original text; national IDs also have country and scheme.