|
| 1 | +date-created:: [[2025-10-29 Wed]] |
| 2 | +alias:: [[CursorAI/v/2.0 - New Coding Model and Agent Interface]] |
| 3 | +tags:: [[CursorAI/Changelog]] |
| 4 | + |
| 5 | +- # [New Coding Model and Agent Interface · Cursor](https://cursor.com/changelog/2-0) |
| 6 | + - ### Multi-Agents |
| 7 | + - Manage up to eight agents simultaneously within the new editor interface |
| 8 | + - Each agent operates in its own isolated copy of your codebase |
| 9 | + - Utilizes git worktrees or remote machines to prevent file conflicts |
| 10 | + - ### Composer |
| 11 | + - Introducing Composer, a frontier coding model that is four times faster than similarly intelligent models |
| 12 | + - Enhances coding efficiency |
| 13 | + - ### Browser (General Availability) |
| 14 | + - The Browser feature for Agent, previously in beta, is now generally available |
| 15 | + - Includes embedded in-editor support with tools to select elements and forward DOM information to the agent |
| 16 | + - ### Improved Code Review |
| 17 | + - Easily view all changes made by the Agent across multiple files without needing to navigate between individual files |
| 18 | + - ### Sandboxed Terminals (General Availability) |
| 19 | + - Sandboxed terminals are now generally available for macOS |
| 20 | + - Agent commands run in a secure sandbox by default |
| 21 | + - Read/write access to your workspace with no internet access |
| 22 | + - ### Team Commands |
| 23 | + - Define and manage custom commands and rules for your team via the Cursor dashboard |
| 24 | + - These settings are automatically applied to all team members without local storage requirements |
| 25 | + - ### Voice Mode |
| 26 | + - Control the Agent using voice commands with built-in speech-to-text conversion |
| 27 | + - Custom submit keywords can be defined in settings to trigger the agent |
| 28 | + - ### Improved Performance |
| 29 | + - Enhanced performance of Language Server Protocols (LSPs) for all languages |
| 30 | + - Leads to faster loading and usage, especially noticeable when working with agents and viewing diffs |
| 31 | + - ### Plan Mode in Background |
| 32 | + - Create plans with one model and build them with another |
| 33 | + - Choose to build in the foreground or background |
| 34 | + - Plan with parallel agents for multiple plan reviews |
| 35 | + - ### Shareable Team Commands |
| 36 | + - Share custom rules, commands, and prompts with your entire team |
| 37 | + - Create deeplinks through the Cursor Docs |
| 38 | + - ### Improved Prompt UI |
| 39 | + - Files and directories are now displayed inline as pills |
| 40 | + - Context menu has been streamlined |
| 41 | + - Allows the Agent to self-gather context without manual attachment |
| 42 | + - ### Improved Agent Harness |
| 43 | + - Significant improvements to the underlying harness for working with Agent across all models |
| 44 | + - Results in notable quality enhancements, especially for GPT-5 Codex |
| 45 | + - ### Cloud Agents |
| 46 | + - Cloud agents now offer 99.9% reliability, instant startup, and an improved user interface |
| 47 | + - The experience of sending agents to the cloud from the editor has also been enhanced |
| 48 | + - ### Cursor for Enterprise |
| 49 | + - #### Sandboxed Terminals: Admin Controls |
| 50 | + - Enterprise teams can enforce standard settings for sandboxed terminals across their team |
| 51 | + - Configuring sandbox availability, git access, and network access at the team level |
| 52 | + - #### Hooks: Cloud Distribution |
| 53 | + - Enterprise teams can distribute hooks directly from the web dashboard |
| 54 | + - Admins can add new hooks from the dashboard |
| 55 | + |
0 commit comments