Anthropic Claude + xLink
Quickstart

# npm
npm install @private.me/anthropic-xlink

# pnpm
pnpm add @private.me/anthropic-xlink

# yarn
yarn add @private.me/anthropic-xlink

import { createAnthropicClient } from '@private.me/anthropic-xlink';

// Connect to Claude with xLink (no API key needed)
const client = await createAnthropicClient({
  identity: 'my-ai-agent',
  // That's it - xLink handles authentication
});

// Use Claude exactly as before
const response = await client.messages.create({
  model: 'claude-opus-4-6',
  max_tokens: 1024,
  messages: [
    { role: 'user', content: 'Explain quantum computing' }
  ]
});

console.log(response.content[0].text);

import { createAnthropicClient, FileTrustRegistry } from '@private.me/anthropic-xlink';

const registry = new FileTrustRegistry('./trust-registry.json');

const client = await createAnthropicClient({
  identity: 'my-ai-agent',
  trustRegistry: registry
});

// Only agents in trust-registry.json can authenticate

// Region 1 (US-East)
const clientUSEast = await createAnthropicClient({
  identity: 'claude-us-east',
  trustRegistry: sharedRegistry  // Same registry, different identity
});

// Region 2 (EU-West)
const clientEUWest = await createAnthropicClient({
  identity: 'claude-eu-west',
  trustRegistry: sharedRegistry  // No credential sync needed
});

// Each region authenticates independently, no shared secrets

const agents = await Promise.all([
  createAnthropicClient({ identity: 'analyzer' }),
  createAnthropicClient({ identity: 'summarizer' }),
  createAnthropicClient({ identity: 'translator' })
]);

// 3 agents, 3 unique identities, 0 API keys to manage
const results = await Promise.all(
  agents.map((client, i) =>
    client.messages.create({ /* ... */ })
  )
);

// Before (with API keys)
import Anthropic from '@anthropic-ai/sdk';

const client = new Anthropic({
  apiKey: process.env.ANTHROPIC_API_KEY  // Vulnerable
});

// After (with xLink)
import { createAnthropicClient } from '@private.me/anthropic-xlink';

const client = await createAnthropicClient({
  identity: 'my-ai-agent'  // Secure, no credentials
});

// Everything else stays exactly the same