Cursor ships self-hosted cloud agents

Cursor’s latest changelog entry introduces self-hosted cloud agents, giving teams a way to run agent workflows entirely inside their own infrastructure. Instead of sending code, secrets, and tool execution to Cursor-hosted environments, organizations can keep execution on internal machines while still using the cloud-agent model. This is a meaningful release because infrastructure control is one of the biggest blockers to adopting autonomous coding agents in regulated or security-sensitive environments. Cursor is clearly pushing toward enterprise readiness by making the runtime model more flexible.

Key Features or Updates

Cursor says self-hosted cloud agents keep codebases, build outputs, and secrets inside the customer’s own network. The company also says these agents preserve the same core capabilities as Cursor-hosted agents, including isolated environments, full development tooling, multi-model harnesses, plugins, and more. That combination makes the feature more than a compliance checkbox—it changes where the agent actually runs.

Impact on Developers

For developers and platform teams, this lowers one of the biggest barriers to adopting cloud-style coding agents in enterprise settings. Teams that need tighter security controls, internal-only execution, or closer data residency guarantees now have a path to use Cursor’s agent workflows without fully outsourcing runtime control. That should make the product far more viable for larger organizations and security-conscious teams.

How to use it

Teams interested in the feature need to enable self-hosted cloud agents from the Cursor Dashboard and connect the agent runtime to their own infrastructure. The obvious first step is to test it on an internal codebase where secrets, build outputs, and tool execution policies matter. If adoption goes well, it could become the default way to introduce agents into a stricter engineering environment.