News & Updates, Code Editors, AI Agents

Cursor 3.2 Ships Async Subagents, Worktrees, and Multi-repo Support

Cursor 3.2 lands with /multitask for parallel async subagents, redesigned Agents Window worktrees, and multi-root workspace support for cross-repo changes — a significant push toward fleet-based autonomous coding.

3 min read
Cursor 3.2 Ships Async Subagents, Worktrees, and Multi-repo Support

Image by Anysphere

Cursor 3.2 Ships Async Subagents, Worktrees, and Multi-repo Support

Cursor shipped version 3.2 on April 24, 2026, pushing its parallel agent architecture further with three developer-focused upgrades: a new /multitask command for async subagent parallelization, a redesigned worktrees experience inside the Agents Window, and multi-root workspace support that lets a single agent session span multiple repositories simultaneously.

The release continues the trajectory set by Cursor 3.0 in early April — building out the Agents Window as the center of gravity for professional coding workflows, and reducing the friction of coordinating work across multiple agents, branches, and codebases at once.

/multitask: Parallel Async Subagents on Demand

The headline feature is /multitask, a new command that tells Cursor to parallelize incoming work rather than serialize it in a queue. When invoked, Cursor spawns async subagents to handle requests concurrently. For larger tasks, it breaks them into smaller chunks and distributes them across a fleet of subagents simultaneously.

The practical implication is meaningful: if you have queued messages waiting for the current agent run to finish, you can now redirect them to run as parallel subagents instead. The bottleneck of one-agent-at-a-time execution on large tasks has been a recurring friction point for developers running multi-step workflows, and /multitask is a direct answer to it.

Worktrees Upgraded in the Agents Window

The 3.0 release introduced worktrees via the /worktree command in the classic editor chat. Cursor 3.2 brings worktree support into the Agents Window proper, with a redesigned experience. You can now run isolated tasks in the background across different branches, and when a branch is ready to test, bring it into your local foreground with a single click.

This replaces what was previously a more manual branch-switching and context-switching workflow. Agents running on separate branches no longer block your main working context — they run asynchronously until you are ready to pull their output into view.

Multi-root Workspaces: Cross-repo Changes in One Session

The third major addition is multi-root workspace support in the Agents Window. A single agent session can now target a reusable workspace made of multiple folders, allowing Cursor to execute cross-repo changes — spanning frontend, backend, and shared libraries — without requiring the user to retarget the agent each time it moves between repositories.

For monorepo shops this is less transformative, but for teams with separated codebases that share business logic or API contracts, this closes a meaningful gap. Previously, making coordinated changes across a TypeScript frontend and a Python backend required separate agent sessions and manual handoffs. Multi-root workspaces allow that coordination to happen within a single session.

The Bigger Picture

Cursor 3.2 is the third significant update since the 3.0 launch on April 2. The 3.1 release on April 13 brought tiled layout for parallel agent views and upgraded voice input. The April 15 update added interactive canvases. Version 3.2 deepens the parallelism story at the execution layer, not just the display layer.

The release cadence reflects the pressure Cursor faces from Claude Code and OpenAI Codex, both of which are pushing hard into multi-agent workflows. The /multitask command is Cursor's direct answer to the kind of coordinated multi-agent execution that Claude Code's Agent Teams feature has been building toward.

Cursor has over one million daily active users and crossed $2 billion in annualized revenue in early 2026. The company is reportedly in talks to raise at a $50B+ valuation, and SpaceX struck an acquisition rights deal in April. The pace of 3.x updates signals that Anysphere is prioritizing product velocity over fundraising distraction.

What's Unconfirmed

Cursor has not disclosed performance metrics for /multitask subagent runs — specifically how token usage scales when multiple subagents are spawned from a single request. The practical cost implications for Pro and Business plan users running fleet-scale workflows remain unclear. Multi-root workspace support is documented as targeting multiple folders, but the degree to which cross-language tooling and environment setup is handled automatically has not been detailed publicly.

Share:

Other Latest News

Windsurf Becomes Devin Desktop With New Agent Command Center
AI Agents, News & Updates, Code Editors

Windsurf Becomes Devin Desktop With New Agent Command Center

Cognition has relaunched Windsurf as Devin Desktop, a full IDE with an Agent Command Center and open ACP protocol support that lets Codex, Claude Agent, OpenCode, and custom in-house agents run side by side in a shared Kanban workspace.

Jun 3, 2026
OpenAI Codex Lands on Amazon Bedrock for 5 Million Weekly Users
AI Agents, News & Updates, Code Editors

OpenAI Codex Lands on Amazon Bedrock for 5 Million Weekly Users

OpenAI's AI coding agent Codex is now generally available on Amazon Bedrock, giving developers at AWS-native organizations direct access to Codex through existing cloud infrastructure, procurement, and compliance workflows.

Jun 3, 2026
Anthropic Ships Claude Opus 4.8 With Dynamic Workflows and 3x Cheaper Fast Mode
AI Agents, News & Updates, Code Editors

Anthropic Ships Claude Opus 4.8 With Dynamic Workflows and 3x Cheaper Fast Mode

Anthropic releases Claude Opus 4.8 with improved agentic coding benchmarks, a 3x cheaper Fast Mode, and Dynamic Workflows that run hundreds of parallel subagents inside Claude Code.

May 29, 2026
MiniMax Revenue Doubles as M3 Promises 15x Long-Context Speed Boost
News & Updates, AI Assistants, Industry Analysis

MiniMax Revenue Doubles as M3 Promises 15x Long-Context Speed Boost

MiniMax annualized revenue doubled in two months, crossing 1M enterprise users. The company just teased M3's sparse attention architecture claiming 15.6x faster decoding at 1M-token contexts.

May 28, 2026
Anthropic Ships Free Claude Code Security Plugin to Catch Vulnerabilities in Real Time
AI Agents, News & Updates, Code Editors

Anthropic Ships Free Claude Code Security Plugin to Catch Vulnerabilities in Real Time

Anthropic's new Security Guidance plugin for Claude Code automatically scans code changes for injection flaws, unsafe deserialization, and 25+ dangerous patterns before they reach pull requests — free for all users on every plan.

May 27, 2026
Anthropic's Mythos Model Surfaces in Claude Code Ahead of Wider Release
AI Agents, News & Updates, Code Editors

Anthropic's Mythos Model Surfaces in Claude Code Ahead of Wider Release

Anthropic's most powerful restricted model, Claude Mythos 1, briefly appeared in Claude Code and Claude Security interfaces on May 25 — signaling the first commercial rollout of the model that found over 10,000 zero-day vulnerabilities under Project Glasswing.

May 26, 2026
← Scroll for more →