Dokumentacija

Vodiči za integraciju MCP Server, REST API i SDK za anonym.legal i cloak.business

MCP Server

7 alata preko HTTP-a · Claude Desktop, Cursor, VS Code

REST API

Direktna API integracija · SDK nije potreban

Primeri koda

Python, JavaScript, cURL

Claude Desktop

Dodaj u claude_desktop_config.json:

{
  "mcpServers": {
    "anonym-legal": {
      "type": "http",
      "url": "https://anonym.legal/mcp",
      "headers": { "Authorization": "Bearer YOUR_API_KEY" }
    }
  }
}

Lokacije konfiguracije:
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json

Cursor IDE

Dodaj u Cursor MCP podešavanja (Settings › Features › MCP):

{
  "mcpServers": {
    "anonym-legal": {
      "type": "http",
      "url": "https://anonym.legal/mcp",
      "headers": { "Authorization": "Bearer YOUR_API_KEY" }
    }
  }
}

7 MCP alata

Alat	Opis
`anonymize_text`	Anonimizuje tekst — replace, redact, hash, encrypt, mask ili keep po entitetu
`analyze_text`	Detektuje PII entitete — vraća pozicije, tipove i ocene pouzdanosti
`detokenize_text`	Vraća originalne vrednosti iz sesijskih tokena
`get_balance`	Proverava preostali broj tokena
`estimate_cost`	Procenjuje potrošnju tokena pre poziva anonymize_text
`list_sessions`	Prikazuje listu aktivnih sesija anonimizacije
`delete_session`	Briše sesiju — brisanje po GDPR čl. 17

Preduslovi

Zahtev	Obavezno	Opis
`anonym.legal account`	Da	Pro ili Business plan (MCP Server nije dostupan u Free/Basic planovima)
`API key`	Da	Šalje se kao `Authorization: Bearer YOUR_API_KEY` header — Node.js ili npm instalacija nisu potrebni
`MCP client`	Da	Bilo koji klijent koji podržava HTTP transport (Claude Desktop, Cursor, VS Code, Windsurf, Continue, Cline)

Osnovni URL

https://api.anonym.legal/v1

Autentifikacija

Unesite svoj API ključ u Authorization header:

Authorization: Bearer your-api-key

POST/anonymize

Anonimizuje tekst i detektuje PII entitete.

Telo zahteva

{
  "text": "John Smith's email is john@example.com",
  "method": "replace",
  "language": "en",
  "entities": ["PERSON", "EMAIL_ADDRESS"]
}

Odgovor

{
  "anonymized": "Max Mueller's email is max.mueller@example.de",
  "entities": [{ "type": "PERSON", "original": "John Smith", "start": 0, "end": 10 }],
  "tokens_used": 2
}

POST/detokenize

Vraća originalne vrednosti iz sesijskih tokena.

{ "text": "enc_xK9mP2... email is enc_jL7nQ4...", "session_id": "abc123" }

Python (requests)

import requests

API_KEY = "your-api-key"
BASE_URL = "https://api.anonym.legal/v1"

def anonymize(text, method="replace"):
    response = requests.post(
        f"{BASE_URL}/anonymize",
        headers={"Authorization": f"Bearer {API_KEY}"},
        json={"text": text, "method": method}
    )
    return response.json()

result = anonymize("Contact John at john@example.com")
print(result["anonymized"])
# → "Contact Max Mueller at max.mueller@example.de"

JavaScript / Node.js

const BASE_URL = 'https://api.anonym.legal/v1';
const API_KEY  = 'your-api-key';

const result = await fetch(`${BASE_URL}/anonymize`, {
  method: 'POST',
  headers: { 'Authorization': `Bearer ${API_KEY}`, 'Content-Type': 'application/json' },
  body: JSON.stringify({ text: 'Contact John at john@example.com', method: 'replace' })
}).then(r => r.json());

console.log(result.anonymized);
// → "Contact Max Mueller at max.mueller@example.de"

cURL

curl -X POST https://api.anonym.legal/v1/anonymize \
  -H "Authorization: Bearer your-api-key" \
  -H "Content-Type: application/json" \
  -d '{"text":"Contact John at john@example.com","method":"replace"}'

MCP Server

10 alata · tekst, batch obrada, OCR slika

Python SDK

pip install cloak-business

Node.js SDK

npm i @cloak-business/sdk

Primeri koda

Batch obrada, OCR slika, isečci SDK koda

Claude Desktop

Dodaj u claude_desktop_config.json:

{
  "mcpServers": {
    "cloak": {
      "command": "npx",
      "args": ["-y", "cloak-business-mcp-server"],
      "env": { "CLOAK_API_KEY": "your-api-key" }
    }
  }
}

Lokacije konfiguracije:
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json

Cursor IDE

Dodaj u Cursor MCP podešavanja (Settings › Features › MCP):

{
  "mcpServers": {
    "cloak": {
      "command": "npx",
      "args": ["-y", "cloak-business-mcp-server"],
      "env": { "CLOAK_API_KEY": "your-api-key" }
    }
  }
}

10 dostupnih alata

Alat	Opis
`cloak_analyze_text`	Detektuje PII u tekstu (320+ tipova, 70+ zemalja)
`cloak_anonymize_text`	Anonimizuje tekst pomoću 7 operatora
`cloak_deanonymize_text`	Vraća originalne vrednosti iz sesije
`cloak_batch_analyze`	Analizira od 1 do 100 tekstova u jednom API pozivu
`cloak_analyze_image`	OCR + detekcija PII u slikama (38 jezika)
`cloak_redact_image`	Vizuelno prikriva (redaktuje) PII regione u slikama
`cloak_create_session`	Kreira trajnu sesiju anonimizacije
`cloak_get_session`	Preuzima mapiranja tokena za sesiju
`cloak_list_presets`	Prikazuje listu dostupnih preset konfiguracija entiteta
`cloak_apply_preset`	Anonimizuje koristeći sačuvani preset

Promenljive okruženja

Promenljiva	Obavezno	Opis
`CLOAK_API_KEY`	Da	API ključ sa cloak.business
`CLOAK_METHOD`	Ne	Podrazumevano: replace. Opcije: replace, redact, hash, encrypt, mask, keep, rsa
`CLOAK_LANGUAGE`	Ne	Podrazumevano: auto. ISO 639-1 kod jezika

Python SDK

Instalacija preko pip-a:

pip install cloak-business

from cloak_business import CloakBusiness

client = CloakBusiness(api_key="your-api-key")

# Anonymize a single text
result = client.anonymize("John Smith emailed john@example.com")
print(result["anonymized"])

# Batch: 1–100 texts per call
results = client.batch_anonymize([
    "Patient: Alice, DOB 1985-04-12",
    "Invoice to Bob Corp, VAT DE123456789",
])
for r in results: print(r["anonymized"])

Kompletna API referenca: cloak.business/docs

Node.js SDK

Instalacija preko npm-a:

npm install @cloak-business/sdk

import { CloakBusiness } from '@cloak-business/sdk';

const client = new CloakBusiness({ apiKey: 'your-api-key' });

// Anonymize a single text
const result = await client.anonymize('John Smith emailed john@example.com');
console.log(result.anonymized);

// Batch: 1–100 texts per call
const batch = await client.batchAnonymize([
  'Patient: Alice, DOB 1985-04-12',
  'Invoice to Bob Corp, VAT DE123456789',
]);
batch.forEach(r => console.log(r.anonymized));

Kompletna API referenca: cloak.business/docs

Batch obrada (Python)

from cloak_business import CloakBusiness

client = CloakBusiness(api_key="your-api-key")

# Anonymize up to 100 documents per call
documents = [
    "HR record: Jane Doe, jane@corp.com, +49-30-12345678",
    "Contract: Acme GmbH, HRB 123456, VAT DE987654321",
    # ... up to 100 texts
]

results = client.batch_anonymize(documents, method="replace")
for i, r in enumerate(results):
    print(f"[{i}] {r['anonymized']}")

# Tokens consumed by this batch
total_tokens = sum(r.get("tokens_used", 0) for r in results)
print(f"{total_tokens} tokens used")

OCR slika + Redakcija (Python)

from cloak_business import CloakBusiness
import base64, pathlib

client = CloakBusiness(api_key="your-api-key")

# Read image and encode to base64
image_b64 = base64.b64encode(
    pathlib.Path("scanned_contract.png").read_bytes()
).decode()

# Detect PII in image (OCR — 38 languages)
analysis = client.analyze_image(image_b64, language="de")
print("Found:", [e["type"] for e in analysis["entities"]])

# Redact visually — returns a new base64-encoded image
redacted = client.redact_image(image_b64, language="de")
pathlib.Path("redacted.png").write_bytes(base64.b64decode(redacted["image"]))

Obostrana deanonimizacija (Node.js)

import { CloakBusiness } from '@cloak-business/sdk';
const client = new CloakBusiness({ apiKey: 'your-api-key' });

// Step 1: anonymize and save the session
const { anonymized, session_id } = await client.anonymize(
  'Call Alice at +49-30-555-1234 to discuss the contract.',
  { method: 'replace', createSession: true }
);
// → "Call Emma at +49-30-555-9876 to discuss the contract."

// Step 2: send anonymized text to AI model ...
const aiResponse = 'Follow up with Emma tomorrow.';

// Step 3: restore original values in AI response
const { deanonymized } = await client.deanonymize(aiResponse, { session_id });
console.log(deanonymized);
// → "Follow up with Alice tomorrow."

replace Podrazumevano

Zamenjuje PII realističnim lažnim podacima. Zadržava čitljivost dokumenta.

"John Smith" → "Max Mueller"

redact

Potpuno uklanja PII. Nepovratno, ali maksimalna privatnost.

"john@example.com" → "[REDACTED]"

hash SHA-256

Jednosmerni kriptografski heš. Dosledna pseudonimizacija.

"555-0123" → "a7f3c9d2..."

encrypt AES-256-GCM

Enkripcija vojnog nivoa. Potpuno povratna uz ispravan ključ.

"DE89370400..." → "enc_xK9mP2..."

mask

Delimična vidljivost sa maskiranim karakterima. Prikazuje strukturu.

"4532-1234-5678-9012" → "4532-****-****-9012"

Treba vam pomoć?

Naš tim je spreman da vam pomogne da integrišete Anonymize.dev u vaš radni tok.

Kontaktirajte podršku Preuzmite API ključ

Dokumentacija

MCP Server

REST API

Primeri koda

Podešavanje MCP Server-a

Claude Desktop

Cursor IDE

7 MCP alata

Preduslovi

REST API

Osnovni URL

Autentifikacija

POST/anonymize

Telo zahteva

Odgovor

POST/detokenize

Primeri koda

Python (requests)

JavaScript / Node.js

cURL

MCP Server

Python SDK

Node.js SDK

Primeri koda

Podešavanje MCP Server-a

Claude Desktop

Cursor IDE

10 dostupnih alata

Promenljive okruženja

SDK referenca

Python SDK

Node.js SDK

Primeri koda

Batch obrada (Python)

OCR slika + Redakcija (Python)

Obostrana deanonimizacija (Node.js)

Metode anonimizacije

Treba vam pomoć?