← Back to News

Cursor launches self-hosted cloud agents

2026-04-06 AI-CODING

Cursor has added self-hosted cloud agents, letting companies run agent workflows entirely within their own infrastructure. The launch keeps code, build outputs, and tool execution on internal machines while preserving the cloud-agent experience. This is a major enterprise step for agentic development tools. It lowers one of the biggest adoption barriers for regulated or security-sensitive teams by making cloud agents much easier to deploy without giving up control of the runtime.

Key Updates

Cursor says self-hosted cloud agents can run with the same capabilities as Cursor-hosted agents, including isolated VMs, full development environments, multi-model harnesses, and plugins. The difference is where execution happens: locally inside the customer's own network rather than in Cursor-managed infrastructure. That gives teams the same workflow with stricter control over code and secrets.

What Developers Need to Know

This is especially relevant for teams with proprietary code, sensitive data, or compliance requirements. It also shows that agent platforms are maturing into enterprise-grade infrastructure rather than just developer tools. The ability to keep execution local while still using cloud-style orchestration may accelerate adoption inside larger organizations.

How to use it or Next Steps

Teams interested in the feature should enable self-hosted cloud agents in the Cursor Dashboard and connect them to their own infrastructure. Start with an internal project where security boundaries matter, then measure whether the agent fits your review and deployment process. The best test is whether it speeds up work without creating new governance problems.

Read Original Post →