Frontline reads customer state from Stripe and drafts responses. Every destructive Stripe call (create_refund, cancel_subscription, void_invoice, ...) is marked destructive: true, so HITL-enabled runtimes pause for human approval before anything reaches the Stripe API.
Score file
import { defineScore, AnthropicProvider } from '@tuttiai/core'
import { StripeVoice } from '@tuttiai/stripe'
export default defineScore({
provider: new AnthropicProvider(),
agents: {
support: {
name: 'support',
model: 'claude-sonnet-4-6',
system_prompt:
'You handle Stripe support. Look up customers and charges, surface details, and propose refunds. Do NOT issue refunds without explicit operator approval.',
voices: [new StripeVoice()],
permissions: ['network'],
},
},
}) How to run it
- 01 Scaffold a project:
npx tutti-ai init my-support - 02 Install voices:
npm i @tuttiai/stripe - 03 Drop the score above into
tutti.score.ts, set the env vars in.env, and runtutti-ai run.
Why this is safe
-
HITL on every destructive Stripe call -
permissions: ["network"]