Let verified agents into your platform.

Boot the bad ones. When an AI agent shows up to post, buy, or call your API, verify who it is and what it's allowed to do, then admit it or bounce it. Free, drop-in, runs in your own backend.

$npm install axis-platform-sdk

read the quickstart

Gate a route. One import, one line.

node / express

import { axisGate } from 'axis-platform-sdk/express'

app.post('/comments',
  axisGate({ audience, requireScopes: ['content:comment'] }),
  (req, res) => res.json({ by: req.axis.agent_id }))

cloudflare worker

import { aitGate, denialResponse } from 'axis-platform-sdk'

const gate = aitGate({ audience, requireScopes: ['content:comment'] })
// in fetch(): const v = await gate(request)

Verify identity

Signature, revocation, and delegation chain, checked by the public registry.

Check the permission

Match the action against the agent's proven scope, not what its token claims.

Boot bad actors

Block one agent or a whole operator at runtime, with no redeploy.

Zero infra

One outbound HTTPS call. No database, no key from us, nothing to host.

Self-host, day one. The SDK runs entirely in your own backend — everything you need to launch today. alpha A cloud-hosted version (hosted console, durable history) is in alpha testing, planned for Q3 2026.