Cursor Enhances Bugbot with Learned Rules & MCP Access

What's New with Cursor's Bugbot?

Cursor has rolled out significant enhancements to Bugbot, its AI-powered code review assistant, focusing on real-time self-improvement and broader contextual awareness. The highlight is "Learned Rules," a groundbreaking feature allowing Bugbot to automatically learn from feedback on pull requests. By analyzing reactions and replies to its own comments, as well as human reviewer input, Bugbot generates candidate rules that are then automatically promoted or disabled based on their usefulness. This means Bugbot continually gets smarter, tailoring its suggestions to your team's specific coding practices and preferences.

Beyond intelligent self-learning, Bugbot now boasts support for Multi-Cloud Platform (MCP) servers. This crucial update grants Bugbot access to additional context during code reviews, enabling more accurate and insightful feedback, especially for complex, distributed systems. Combined with a robust set of improvements like a new "Fix All" action to apply multiple suggested fixes in a single go, Bugbot is set to streamline your development workflow more efficiently than ever. These advancements have already pushed Bugbot's resolution rate to an impressive 78%.

Why These Updates Matter for Your Team

The introduction of Learned Rules fundamentally transforms Bugbot from a static AI assistant into a dynamic, self-improving team member. Imagine an AI that adapts its understanding of "good code" to your project's evolving needs without manual retraining. This not only reduces the noise often associated with automated reviews but also ensures that Bugbot's suggestions are consistently relevant and actionable. You can explore more about how Bugbot learns in this announcement on Bugbot's learning capabilities.

Furthermore, the integration with MCP servers is a game-changer for teams operating in multi-cloud environments. By providing Bugbot with deeper contextual understanding, reviews become more comprehensive, catching potential issues that might otherwise be missed. Cursor has also refined Bugbot Autofix, ensuring it now only runs when findings are substantial enough to warrant a fix and uses only relevant rules, further minimizing distractions. Reliability of Autofix CI checks on PRs has been significantly improved, and GitHub PR check messages are simplified for clearer progress updates. Bugbot has also addressed infrastructure issues to reduce run times and fixed a bug related to stale privacy mode incorrectly blocking Autofix, ensuring smoother operations.

Getting Started and Taking Control

Ready to leverage these smart new capabilities? The new Learned Rules feature starts working automatically as Bugbot receives feedback. For a deeper dive into how these rules operate and how to manage them, you can consult the documentation on Bugbot Rules. Teams on Cursor's Teams and Enterprise plans can take full advantage of the MCP support by adding tools to Bugbot directly through the Bugbot Dashboard, unlocking that crucial additional context.

Managing Bugbot has also become simpler with a redesigned settings interface that clearly separates personal and team configurations. If you're looking to fine-tune how Bugbot interacts with your repositories, the repository rules management page is your go-to. These updates are designed to give you more control while letting Bugbot do the heavy lifting of intelligent code review, ensuring your team ships high-quality code faster.