Cursor released Cursor 3, a rebuilt interface that puts agents at the center of the development workflow instead of treating them as sidebar assistants.
The release, announced on the Cursor blog, was built from scratch rather than layered on top of the existing VS Code fork. The core bet: developers now spend more time directing agents than editing files, and the interface should reflect that shift.
Parallel Agents Across Environments
The flagship feature is the Agents Window (accessible via Cmd+Shift+P). It lets developers run multiple agents simultaneously across local development environments, cloud VMs, and remote SSH machines. All active agents, whether kicked off from desktop, mobile, web, Slack, GitHub, or Linear, appear in a unified sidebar.
Cloud agents produce demos and screenshots of their work for verification. Sessions can move between local and cloud: push a local agent to the cloud when closing your laptop, or pull a cloud agent local when you need to edit and test directly. Cursor says this handoff is designed to be fast enough that developers use it mid-task rather than planning around it.
Design Mode
Design Mode adds visual annotation to agent interactions. Developers can shift-drag to select a UI area in the integrated browser, then Cmd+L to add that element directly to the agent’s chat context. Instead of typing “the submit button on the checkout page has the wrong padding,” developers point at the element and let the agent see exactly what they mean.
This is a bet that the “show, don’t tell” interaction model scales better than text-only prompting for UI work, where spatial relationships and visual context are hard to describe precisely.
Agent Tabs
Agent Tabs let developers view multiple agent conversations side-by-side or in a grid layout, reducing the context-switching cost of managing parallel workstreams. Combined with the multi-repo workspace support, a single Cursor window can now track agents working across different codebases simultaneously.
From Editor to Orchestration Platform
Cursor explicitly frames this as the “third era of software development,” where fleets of agents work autonomously to ship improvements, according to the blog announcement. The company cites its own internal workflow as evidence: Cursor’s team has moved from manually editing files to working primarily through agents.
The redesign signals where developer tooling is heading. When the IDE rebuilds itself around agent orchestration rather than code editing, it reflects a market judgment that agent-first workflows are the default, not the exception. For developers evaluating their tooling stack, Cursor 3 is the first major IDE to make that bet structurally, not just feature-by-feature.