Feature request for product/service
Cloud Agent (GitHub, Slack, Linear,…)
Describe the request
Hi Cursor Team,
I’m using Self-hosted Cloud Agents with My Machines (private workers). Today, the worker runs tool calls on my machine, but model inference still happens in Cursor’s cloud. That means I can’t use the custom models configured on my remote machine (e.g. OpenAI-compatible endpoints / Override OpenAI Base URL), even when the agent is already running on my own infrastructure.
Request
Please add support for using custom models (or a custom inference endpoint) with Self-hosted / My Machines Cloud Agents, so both tool execution and model inference can run in our own environment.
Why this matters
We want to use internal / self-hosted model services consistently
We have compliance and private-network requirements for the inference path
Execution already runs on our machines, but models are still limited to Cursor’s curated cloud list
Is this on the roadmap? If not yet supported, what’s the recommended alternative (enterprise model access, Bedrock, or another BYOK path)?
Thanks!