Cursor’s New Tool Lets Users Delegate to a Team of Coding Agents

Cursor has launched Cursor 3, a new environment designed to let developers manage teams of AI coding agents working on their behalf. Positioned as a "unified workspace for building software with agents," the release represents a strategic shift in interface rather than a model overhaul. The tool allows users to delegate coding tasks to multiple AI agents—both local and cloud-based—while working across several repositories simultaneously. This transforms the developer's role from hands-on coder to project manager, providing a high-level overview of the work.

The launch comes at a critical time for Cursor, which faces intense competition from Anthropic's Claude Code (reportedly holding 54% of the AI coding market) and OpenAI's Codex 5.3, which recently set new benchmarking highs. The company also seeks reputational repair after last month's bumpy launch of Composer 2, its underlying coding model that was discovered to be largely a licensed version of Moonshot AI's open-source Kimi 2.5 model. While licensing models is common practice, Cursor's lack of upfront disclosure about Composer 2's origins has made some users wary of the company's transparency.