Connect AI agents to
real-world APIs

import { Zipe } from '@zipe/sdk';

const zipe = new Zipe({ apiKey: 'zipe_sk_...' });

// List available tools for a user
const tools = await zipe.tools.list({
  userId: 'user_123',
  categories: ['payments', 'communication'],
});

// Execute a tool
const result = await zipe.tools.execute(
  'stripe.create_payment',
  {
    userId: 'user_123',
    arguments: {
      amount: 9900,
      currency: 'usd',
      description: 'Order #1234',
    },
  }
);

# Start an MCP server with your tools
zipe mcp serve --port 3001

# Or configure programmatically
from zipe.mcp import MCPServer

server = MCPServer(
    api_key="zipe_sk_...",
    user_id="user_123",
    toolkits=["payments", "communication"]
)

# Expose tools over MCP protocol
server.start(port=3001)

# Any MCP-compatible client can now connect:
# → Claude Desktop, Cursor, or custom agents

from zipe import Zipe

zipe = Zipe(api_key="zipe_sk_...")

# Create an event trigger
trigger = zipe.triggers.create(
    name="payment_received",
    event="stripe.payment.completed",
    user_id="user_123",
    webhook_url="https://your-app.com/hook",
    filters={"amount_gte": 1000}
)

# List active triggers
active = zipe.triggers.list(user_id="user_123")

# Triggers fire automatically — your agent
# receives structured event payloads

# Authenticate with your API key
$ zipe auth login

# List available toolkits
$ zipe toolkits list
┌─────────────┬──────────┬────────┐
│ Toolkit     │ Tools    │ Status │
├─────────────┼──────────┼────────┤
│ stripe      │ 12       │ active │
│ twilio      │ 8        │ active │
│ github      │ 15       │ active │
└─────────────┴──────────┴────────┘

# Test a tool execution locally
$ zipe tools execute stripe.create_payment \
    --user user_123 \
    --arg amount=9900 \
    --arg currency=usd

# Deploy an MCP server
$ zipe mcp deploy --toolkit payments