Quickstart

Create an accountCopy Anchor Link

Sign up at leap360.ai. A project and an API key are created for you, and the project starts a 7-day shadow window.

Install the SDKCopy Anchor Link

npm install leap360

Sign inCopy Anchor Link

npx leap360 login

The CLI prints a link and a code. Approve the device in your browser and the CLI saves your API key to ~/.leap360/config.json. On a server or in CI, set the key as an environment variable instead:

export LEAP360_API_KEY="leap360_..."

You can copy the key from Settings in the console.

Wrap a tool your agent can callCopy Anchor Link

Wrap the function that actually does the work. Leap360 checks each call against your policies before it runs, then records the outcome.

import { Leap360, Leap360ActionBlockedError } from 'leap360';

const leap = new Leap360({ apiKey: process.env.LEAP360_API_KEY, projectId: 'my-app', serviceName: 'deploy-bot' });

const deleteNamespace = leap.wrapTool('delete_namespace', async ({ name }: { name: string }) =>
  k8s.deleteNamespace(name),
);

try {
  await deleteNamespace({ name: 'payments' });
} catch (err) {
  if (err instanceof Leap360ActionBlockedError) console.log(err.gate.message); // the function never ran
}

Hand the wrapped function to your agent framework as you normally would.

Run your agent and open the consoleCopy Anchor Link

Run your agent once. Open Live activity and you will see the call, its risk, and what your policies decided. In shadow mode the decision reads "would be blocked" or "would be held" and the call still runs.