Cursor has launched Bugbot Autofix, an AI-powered code review agent designed to automatically identify and resolve issues within pull requests. This move addresses a growing bottleneck in software development, where code review often lags behind code generation by advanced agents.
Leveraging independent cloud agents in virtual machines, Bugbot Autofix tests software and proposes fixes. Currently, over 35% of its suggested changes are merged directly into the base PR, significantly streamlining the development workflow.
Out of beta and available to all Bugbot users, the system has nearly doubled the average number of issues identified per run in the last six months. Its resolution rate—bugs resolved by users before PR merger—has surged from 52% to 76%, indicating improved accuracy and fewer false positives.
This initial deployment of automatic, event-driven agents marks a foundational step. Cursor plans to expand beyond code review, enabling custom automations for various workflows. Future enhancements for Bugbot include self-verification of findings, deep research capabilities for complex issues, and continuous codebase scanning to proactively address bugs, advancing the field of AI code review automation.
