Bugbot now self-improves with learned rules

Summary

At its July 2025 beta exit, Bugbot resolved 52% of identified bugs by the relevant PR merge; that rate is now nearing 80%. In the public-repository comparison, Cursor Bugbot reached 78.13% across 50,310 analyzed PRs, versus Greptile’s 63.49%; an LLM judge checked whether comments were addressed by merge. It now turns reactions and replies to Bugbot comments, along with human-reviewer comments, into live feedback rather than relying exclusively on offline experiments. Bugbot converts those signals into candidate rules, evaluates them on incoming PRs, promotes candidates to active status as signal accumulates, and can disable rules receiving consistent negative feedback. More than 110,000 repositories have enabled learning, generating over 44,000 rules, which users can edit or delete in the Cursor Dashboard.

Takeaways

1. 1

   Three stated feedback signals are reactions to Bugbot comments, developer replies explaining problems or improvements, and human-reviewer comments identifying issues Bugbot missed.

2. 2

   Candidate rules are evaluated against incoming PRs before promotion to active status; active rules can later be disabled when they generate consistent negative feedback.

3. 3

   Since the learned-rules beta began, more than 110,000 repositories have enabled learning and generated more than 44,000 learned rules.