Cursor 3 quality‑of‑life updates
Cursor rolled out QoL updates in Cursor 3 including terminal‑style agent splitting for multitasking, a reliable hold‑to‑talk voice input (Ctrl‑M), and event‑based automations for Sentry issues such as automatic investigations and PR creation. The company’s momentum was also highlighted by community figures planning builder events and public endorsements. (x.com) (x.com)
Cursor is adding faster ways to juggle coding agents and automate bug work in Cursor 3, its agent-first interface released on April 2. (cursor.com) Cursor’s new Agents Window lets users run many agents in parallel across repositories and environments, including local machines, cloud sandboxes, worktrees, and remote Secure Shell sessions. Cursor says the window is generally available in Cursor 3 and can be opened with the command palette. (cursor.com 1) (cursor.com 2) The same release added agent tabs in the editor for side-by-side or grid chat layouts, plus an Await tool that lets agents watch long-running shell jobs and wait for output such as “Ready” or “Error.” Cursor also said large-file diffs now render faster and use less memory. (cursor.com) Cursor has been moving its product toward agents that work more like delegated coworkers than autocomplete. Its documentation describes the Agents Window as a unified workspace for managing agents across projects, with review, commit, and pull request controls inside the same interface. (cursor.com) That push expanded again on March 5, when Cursor introduced Automations, a system for cloud agents that run on schedules or after events such as a Slack message, a new Linear issue, a merged GitHub pull request, or a PagerDuty incident. Cursor said those agents spin up a cloud sandbox, use configured tools, and verify their own output. (cursor.com) The Sentry example now making the rounds shows how that event model can turn bug intake into code changes. In a video published April 14, Sentry showed user feedback landing as a Sentry issue, triggering a Cursor Automation on assignment, generating a pull request against the repository, and posting the link into Slack. (opsmatters.com) Cursor had already added built-in voice control in Cursor 2.0, saying users could control Agent with speech-to-text conversion and custom submit keywords. The company’s current keyboard shortcuts page documents shortcut support broadly, though the specific hold-to-talk key combination cited in community posts was not clearly documented in the pages reviewed here. (cursor.com 1) (cursor.com 2) Cursor’s April 2 release also split its product more clearly between an agent-first window and a classic editor window. Cursor says users can switch back to the editor at any time, and that the editor remains the better fit for Visual Studio Code extensions and flexible file-based screen splitting. (cursor.com) Public reaction has tracked the company’s recent rise in developer tools. In the material provided for this story, Cursor pointed to community event plans and praise from Vercel executive Lee Robinson, alongside its own product post about the update. (x.com 1) (x.com 2) For Cursor, the release keeps shifting the product from a code editor with artificial intelligence features into a system for supervising several coding agents at once. The new additions are small on their own, but they all push in the same direction: less manual handoff, more work happening in parallel. (cursor.com) (cursor.com)
