Cursor 2.4 Adds Parallel Subagents, Agent Skills, and Built-in Image Generation

Cursor

Jan 22, 2026 · Updated Apr 25, 2026

Cursor's 2.4 update adds parallel subagents for faster multi-file execution, Agent Skills support for domain-specific workflows, and built-in image generation via Google Nano Banana Pro. Agents can also ask clarifying questions without pausing work, making complex prompts less disruptive to your coding flow.

Cursor 2.4 adds parallel subagents that execute tasks simultaneously across your codebase, Agent Skills support via SKILL.md files, built-in image generation with Google Nano Banana Pro, mid-task clarifying questions, and Cursor Blame for AI attribution in git (Enterprise only).

Parallel execution is the headline - refactoring across components or migrating APIs no longer runs sequentially file by file. Skills give agents procedural knowledge for your stack without cluttering global rules, and image generation keeps mockups in your workflow.

Update to Cursor 2.4 - parallel execution kicks in automatically for multi-file operations. Full changelog covers 29 improvements and 14 bug fixes.

View the full update on cursor.com

Cursor

@cursor_aiJan 22

Cursor now uses subagents to complete parts of a task in parallel. Subagents lead to faster overall execution and better context usage. They also let agents work on longer-running tasks. Also new: Cursor can generate images, ask clarifying questions, and more. https://t.co/LTsxuaYuoU

1702k

View on X

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 →

Keep reading

Cursor Launches /orchestrate Skill to Recursively Spawn and Verify AI Agents

Cursor launched the /orchestrate skill for its SDK, allowing AI agents to recursively spawn sub-agents to handle complex engineering tasks. This moves agentic coding into a self-correcting loop where planners spawn workers to write code and verifiers to validate the output.

OpenAIMar 16

OpenAI Codex Now Runs Parallel Subagents for Multi-Agent Workflows

OpenAI's Codex coding agent now supports subagent workflows, letting developers spawn specialized agents that run tasks in parallel. A single prompt can split a PR review across security, quality, and test agents simultaneously, keeping the main context clean.

HiggsfieldMay 22

Higgsfield MCP Connects to Cursor for Agentic Website Asset Generation

Higgsfield AI integrated its Model Context Protocol server with the Cursor code editor, allowing developers to generate production-ready images and videos directly within their IDE. This update enables a unified workflow where coding agents can autonomously create the visual content needed to populate the user interfaces they build.