Exploring Cursor 1.0: AI-Powered Code Review, Background Agent & New MCP Features

1. Using BugBot for automatic code review

BugBot allows AI to automatically review pull requests, catching potential errors and issues. To enable it, open the Cursor web settings, grant GitHub authorization for the desired repositories, and configure options such as disabling automatic execution or requiring manual @ mentions.

Note that any call to the AI large model is billed per token (approximately 1 CNY per 1,000 characters). You can set a monthly spending limit in the settings.

After enabling BugBot, submit a PR containing buggy Java code. BugBot will comment on the PR, identifying the bugs. You can click the "Fix in Cursor" button to open the editor with a pre‑filled fix suggestion, making the debugging process smoother.

2. Background Agent opened to all users

Background Agent, previously in beta, is now officially available to every user. It provides a remote coding proxy, effectively giving Cursor a cloud server that can execute AI‑driven tasks on a remote machine.

To use it, go to Cursor settings → Beta and enable Background Agent. You must first disable Privacy Mode in the settings.

Open the AI chat window with Cmd/Ctrl+E . The first time you try a task, Cursor will ask you to initialise a Git repository.

Install Git from the official website, then run git init in Cursor’s terminal to initialise a repository.

You also need a remote repository on GitHub. Create a private repo, copy the linking commands, paste them into Cursor’s terminal, and restart Cursor for the changes to take effect.

After restarting, the Background Agent can run tasks without hitting the spending limit warning.

The Agent panel shows AI working on a remote server, saving local resources. Future plans include team‑shared Agents for collaborative AI control.

3. Memory feature (Memories)

The new Memory system lets Cursor remember specific rules from your conversations and reuse them later. Enable it in Settings → Rules. Unlike chat history, Memories store only rule‑like information, such as “prefer Windows commands when generating code”.

After adding a rule, generated code will follow it. Rules can be managed or deleted from the same page.

4. MCP one‑click installation and OAuth support

MCP (Model‑Control‑Protocol) services can now be installed with a single click from Cursor’s tool list. After installation, you only need to adjust the authentication credentials.

Some services, such as Notion, also support OAuth. After installing the service, click the provided link to authorize the app in a browser.

Other improvements

Additional updates include richer chat responses (Mermaid diagrams, Markdown tables), Jupyter Notebook Agent support, a new settings dashboard with usage analytics, and various UI tweaks.

Keyboard shortcut for opening the Background Agent panel:

Press Cmd/Ctrl+E to open the Background Agent control panel.

Feature improvements:

@Link and web search can now parse PDFs and include them in context.

Network diagnostics in Settings to verify connectivity.

Parallel tool calls for faster responses.

Collapsible tool calls in chat.

Account updates:

Enterprise users can only access stable releases.

Team admins can now disable Privacy Mode.

Team access to usage and spend data via management API.

Model update:

Max mode is now available for Gemini 2.5 Flash.

Overall, the most useful additions for many developers appear to be the Background Agent and the streamlined MCP installation, marking a significant step toward cloud‑based AI development.