Bugbot Effort Levels
TL;DR
Teams admins and Individual plan users can now customize Bugbot effort levels for PR reviews with Default, High, or Custom configurations based on natural language instructions. Usage-based billing is required for customization.
What changed
Cursor added effort level controls for Bugbot on May 18 2026. Teams admins and Individual plan users can now choose Default, High, or Custom modes for PR reviews. Custom mode accepts natural language instructions but requires usage-based billing to activate.
The update applies only to PR review workflows inside the Cursor platform. No changes were announced for other AI features such as inline chat or composer.
Why it matters
Granular effort settings let builders match review depth to project risk without rewriting prompts each time. This shifts Cursor from a fixed AI reviewer toward a configurable service that competes with lighter open-source alternatives.
The move bets that users will accept usage billing once they see measurable differences in bug detection rates. It pressures competitors who still offer only one-size-fits-all PR checks.
How to use it
Open the Cursor dashboard and navigate to the Bugbot section under workspace settings. Enable usage-based billing first, then select Custom and enter instructions such as focus on security or limit suggestions to performance issues. Changes take effect on the next PR review.
Individual users follow the same path inside their personal account settings. Test on a single repository before rolling out to a team.
Watch for
Track whether Custom instructions reduce false positives on large codebases. The bet breaks if token costs exceed the value of extra reviews. Expect Cursor to add effort presets for other tools such as test generation next.
Who this matters for
- Vibe Builders: Use Custom mode to align PR reviews with your specific brand voice or project style guides.
- Developers: Toggle High effort for critical security modules to catch edge cases that Default reviews miss.
Harsh’s take
Cursor is moving toward a consumption model by gating the best features behind usage-based billing. This is a smart play for reliability: by letting users define effort levels via natural language, they solve the noise problem inherent in automated PR reviews. Builders should treat this as a specialized agent configuration rather than a generic tool update.
The shift to Custom instructions means you can finally stop fighting with generic AI feedback that ignores your specific architectural patterns. If you are on a team plan, the ROI here depends entirely on your ability to write tight, restrictive instructions that prevent token waste on trivial style nitpicks. It is a clear signal that Cursor wants to be the infrastructure for code quality, not just a text editor.
by Harsh Desai
About Cursor
View the full Cursor page →All Cursor updatesGo deeper
More AI news
- Daily RoundupGLM 5.3 and Vercel agent tools land, Gemini SAT practice, plus fresh models on Hugging Face
Google and Vercel released new agent and model features while Hugging Face and Replicate added trending models; OpenAI tightened safety after agent issues.
- Daily RoundupVercel-Origin beta and GPT-5.6 Sol discount, plus agent tools for builders
Vercel rolled out Origin repository deployments and a month-long GPT-5.6 Sol discount on AI Gateway while Product Hunt featured new agent layers and Hugging Face highlighted an uncensored Qwen model, with industry moves around data use and automation exits.
- Weekly DigestHermes Agent v0.20 voice and A2A protocol, OpenClaw mobile updates, and multi-agent tools to test now
Hermes Agent rolled out v0.20 releases with voice streaming, A2A interoperability, desktop artifacts, and self-recovery tools while OpenClaw added Android switching, iOS caching, and plugin provenance checks across the week.