Cursor 3.2 Parallelizes Agent Workflows With Async Subagents and Multi-Repo Support

CursorCursor

· Updated

Cursor 3.2 introduces the multitask command to run asynchronous subagents in parallel rather than waiting for sequential queue execution. This update shifts the developer experience from serial task management to simultaneous orchestration across multiple repositories within a single agent session.

Cursor, an AI-native code editor built by Anysphere, released version 3.2 featuring a new /multitask command for its agentic coding (AI that autonomously writes, tests, and iterates on code) interface. The update allows the editor to spawn asynchronous subagents that execute multiple requests in parallel instead of using a sequential queue.

This shift extends the orchestration of multiple autonomous agents for complex engineering. By enabling parallel execution, the system builds on parallel subagent workflows introduced in earlier versions to handle large-scale refactors and multi-file features without stalling the developer's flow.

You can now also target multi-root workspaces, allowing a single agent session to modify code across several repositories simultaneously. This capability follows the integration of frontier models optimized for multi-file tasks. These features are available in the Agents window of the 3.2 desktop application.

View the full update on cursor.com
Cursor
Cursor
@cursor_ai
X

Introducing /multitask in the new Cursor 3 interface. Cursor can now run async subagents to parallelize your requests instead of adding them to the queue. For already queued messages, you can ask Cursor to multitask on them instead of waiting for the current run to finish. https://t.co/gtvOlup2hX

44retweets714likes
View on X

Still wondering? A few quick answers below.

The multitask command is a new feature in the Cursor 3 interface that allows the AI to run asynchronous subagents. Instead of placing your requests into a sequential queue where each task must finish before the next begins, this command parallelizes your requests so multiple coding tasks can execute at the same time.

Async subagents are independent AI processes that can handle different parts of a coding request simultaneously. In the new Agents window, you can trigger these subagents to work in parallel. If you have messages already waiting in a queue, you can specifically ask Cursor to multitask on them instead of waiting for the current run to complete.

Yes, Cursor 3.2 introduces support for multi-root workspaces. This allows a single AI agent session to target a workspace made of multiple folders or repositories. The agent can reason about and apply code changes across these different directories at once, making it easier to manage complex projects that span multiple repos.

To use these new features, you need to download and install Cursor version 3.2. Once updated, the parallel subagent capabilities and the multitask command are accessible within the Agents window. This interface is specifically designed to help developers orchestrate and manage multiple autonomous AI tasks running concurrently across their local codebase.

Every HeadsUpAI update is written based on its original source and reviewed before it's published. Read our editorial standards →

See all AI news & updates from Cursor →

Share this update