← All docs

Mystery Shopper

10-probe adversarial audit for autonomous AI agents. Tests security, ethics, PII handling, reliability, idempotency, concurrency, error recovery, prompt-injection resistance, data exfiltration, and social engineering.

Run an audit (CLI)

npx @merchantguard/mystery-shopper https://your-agent.example.com

Drop in a probe handler (Node)

npm
npm install @merchantguard/probe-handler
Express example
import express from 'express'
import { handleProbe } from '@merchantguard/probe-handler'

const app = express()
app.post('/probe', handleProbe({ secret: process.env.MG_PROBE_SECRET }))
app.listen(3000)

Pricing

  • Free: 3 probes
  • Starter: 5/$4.99
  • Growth: 15/$9.99
  • Business: 50/$29.99
  • Pro: unlimited / $99/mo (includes continuous monitoring)

Full probe specification: /docs/probe-spec

Nothing on this page is legal advice. Trademarks pending: MerchantGuard™ (Serial 99051215), GuardScore™ (Serial 99030125), AgentGuard™ (Serial 99462472). Patents 63/983,615 / 63/983,621 / 63/983,843 / 63/984,626 (provisional, filed Feb 17 2026). See LEGAL.md, PATENTS.md, DISCLAIMER.md.