Feature request for product/service
BugBot
Describe the request
We use BugBot across multiple repositories that have very different sensitivity profiles and testing requirements.
Some repos are highly sensitive — security-critical code, core backend services, infrastructure — where we genuinely want high-effort reviews to maximize bug detection. Other repos are much lower risk: internal tools, simple frontends, low-impact utilities. For those, default effort is more than sufficient.
The problem is that effort configuration is currently global. If we set high effort globally, we pay elevated costs on every repo including ones that don’t need it. If we set default globally, we under-review the repos where Bugbot’s deeper reasoning actually matters most.
Proposed solution: per-repository effort level configuration
Allow teams to assign effort levels (Default / High / Custom) at the individual repository level from the Bugbot dashboard, instead of only globally. This would ideally include:
- Per-repo effort setting (Default or High)
- Per-repo custom rules — e.g., “use high effort only for PRs touching
auth/,billing/, orinfra/; default elsewhere” - Optionally, support this via a repo-level config file (e.g.,
.cursor/bugbot.md) so it’s version-controlled
Why this matters
- Cost control without sacrificing safety — High-effort reviews cost more and take longer. Per-repo effort lets us keep Bugbot spend predictable while still using deep reviews where they reduce real risk.
- Better alignment with actual risk profiles — Orgs have a mix of critical and low-impact repos. Treating all PRs identically is either wasteful or unsafe depending on which global setting you pick.
- Better DX — Devs on low-risk repos get faster, cheaper reviews without over-aggressive noise. Devs on security-critical repos get thorough reviews where Bugbot’s extra reasoning is most valuable.
- Natural evolution of existing effort levels — Cursor already supports Default/High/Custom effort for tuning cost vs depth. Extending that to the repository dimension feels like the next logical step, especially for teams on usage-based billing.