Hi Cursor team,
This follows up on the closed thread “API / SDK - Terms of use - Question”
(May 5), where Kevin explained that embedding Cursor as a backend AI service
in a product’s tech stack is an intended use of the SDK, and that the line is
reselling Cursor itself or “Cursor-powered usage under your account”.
Our setup: we run a SaaS platform for e-commerce merchants. Inside our app,
merchants chat with our own store assistant, built on the Cursor TypeScript SDK
(local agents running on our server, with our own custom tools that read and
edit the merchant’s own Shopify or WooCommerce store). Merchants never get
access to Cursor, the IDE, a Cursor account or an API key, and they cannot use
the agent for anything outside their store. We don’t train models on outputs
and don’t process payment card or health data.
Two details we want to confirm before launch:
-
Model choice: merchants can pick which model the assistant uses
(for example Composer, Grok, Claude, GPT, Gemini) from a list we control. -
Billing: merchants pay us per usage, in credits, based on the tokens their
chats use. Our price per token currently matches Cursor’s list price.
Is this within the Terms of Service? Or would it count as reselling
“Cursor-powered usage”, meaning we should instead price the assistant as our
own feature (a flat plan or our own credit packs) without exposing per-model
token pricing?
And for production at this scale, do you recommend a Teams plan with a service
account key, or a commercial agreement?
Thanks!