Skip to main content
AI Choice Engine
← All AI code review tools

Cursor

Cursor Bugbot

Cursor Bugbot is a Cursor pull-request reviewer on GitHub (including GHES), GitLab (including self-hosted), Bitbucket (including Data Center), Azure DevOps Services (public beta), Cursor Origin (documented autofix modes), billed as usage-based, last verified October 1, 2026. Cursor’s production reviewer: bug-first, not style-first, and wired back into the Cursor editor the team already switched to.

Catalog record checked October 1, 2026; vendor pricing and policies can change.

Open Cursor BugbotPricing source
Cursor Bugbot specification
SpecificationCursor Bugbot
VendorCursor
Git hostsGitHub (including GHES), GitLab (including self-hosted), Bitbucket (including Data Center), Azure DevOps Services (public beta), Cursor Origin (documented autofix modes)
SurfacePR check plus Cursor editor fixes
Billing modelUsage-based
Hourly / throughputUsage-based; API 30–60 req/min
Trial14-day free trial on all plans (product page).
Config / rulesRoot/nested `.cursor/BUGBOT.md` and team rules; each rule is capped at 30,000 characters and the combined rule context at 100,000 characters per review. Cursor `.mdc` rules under `.cursor/rules` do not apply to Bugbot
Inline comment capNot verifiedUnverified
Self-hosted optionNo
CI hostsNot a separate CI product
AutofixSupported with conditions
Merge gateSupported with conditions
Free for qualifying OSSNot verifiedUnverified
Usage-based

Cursor Bugbot

Cursor’s production reviewer: bug-first, not style-first, and wired back into the Cursor editor the team already switched to.

Best for

  • Teams already on Cursor who want a PR gate without a second vendor
  • Repos where logic bugs matter more than nits

Watch out

Not a substitute for choosing Cursor the editor — that comparison lives under /ai-harnesses. Usage billing means a busy repo can outrun a flat-seat mental model; size spend from cursor.com/pricing#bugbot, not from a third-party $/run table. Azure DevOps is currently public beta and limited to Azure DevOps Services, not Azure DevOps Server. Requiring the Bugbot check without fail-on-unresolved-issues still allows merge when findings are posted. Autofix needs storage plus on-demand usage pricing, and its available modes vary by code host (GitHub supports more modes than GitLab/Bitbucket; Azure DevOps autofix is unavailable).

How a review works. Automatic review on PR updates or a manual trigger. Findings default to a neutral status, so requiring the check alone does not block merge; enable fail-on-unresolved-issues where available. Autofix can use a Cloud Agent, but requires on-demand usage pricing and storage to be enabled (it is unavailable in Legacy Privacy Mode). Existing-branch autofix is capped at three attempts per PR. Its default model follows the configured agent model, with team/system fallback; effort levels are documented separately.

Pricing (verified). Usage-based since the May 2026 change (legacy $40/seat plans renew onto usage billing after 2026-06-08). Individuals consume included usage then on-demand; Teams bill from on-demand spend. Cursor's May 2026 blog gives a $1.00–$1.50 average run cost depending on PR size; this is a vendor average, not a fixed list price. Cursor's current pricing page lists Pro at $20/month and Teams at $40/user/month, with Bugbot on usage-based billing. Confirm current run rates at cursor.com/pricing#bugbot. 14-day free trial on the product page.

Hourly / throughput. No published hourly PR-review cap. API limits are endpoint-specific: POST /bugbot/review is limited to 30 requests/minute/team, dry runs to 10/minute/team, and admin API endpoints to 60/minute/team. Dry runs are still billed and persist findings, even though they do not write to the code host. These are API limits, not review allowances. Reviews first consume included usage (Individuals) or on-demand spend (Teams); Autofix uses Cloud Agent credits at plan rates.

Trial. 14-day free trial on all plans (product page).

Watch out. Not a substitute for choosing Cursor the editor — that comparison lives under /ai-harnesses. Usage billing means a busy repo can outrun a flat-seat mental model; size spend from cursor.com/pricing#bugbot, not from a third-party $/run table. Azure DevOps is currently public beta and limited to Azure DevOps Services, not Azure DevOps Server. Requiring the Bugbot check without fail-on-unresolved-issues still allows merge when findings are posted. Autofix needs storage plus on-demand usage pricing, and its available modes vary by code host (GitHub supports more modes than GitLab/Bitbucket; Azure DevOps autofix is unavailable).

Compare Cursor Bugbot

vs

Cursor Bugbot vs Graphite

vs

Cursor Bugbot vs CodeRabbit

vs

Cursor Bugbot vs Gitar

vs

Cursor Bugbot vs Greptile

vs

Cursor Bugbot vs GitHub Copilot code review

Common questions

Questions about Cursor Bugbot

Answered from the verified figures on this page rather than general guidance.

What is Cursor Bugbot?
Cursor’s production reviewer: bug-first, not style-first, and wired back into the Cursor editor the team already switched to. It reviews on GitHub (including GHES), GitLab (including self-hosted), Bitbucket (including Data Center), Azure DevOps Services (public beta), Cursor Origin (documented autofix modes) as pr check plus cursor editor fixes.
How does Cursor Bugbot price reviews?
Usage-based since the May 2026 change (legacy $40/seat plans renew onto usage billing after 2026-06-08). Individuals consume included usage then on-demand; Teams bill from on-demand spend. Cursor's May 2026 blog gives a $1.00–$1.50 average run cost depending on PR size; this is a vendor average, not a fixed list price. Cursor's current pricing page lists Pro at $20/month and Teams at $40/user/month, with Bugbot on usage-based billing. Confirm current run rates at cursor.com/pricing#bugbot. 14-day free trial on the product page.
What is the hourly or throughput limit for Cursor Bugbot?
No published hourly PR-review cap. API limits are endpoint-specific: POST /bugbot/review is limited to 30 requests/minute/team, dry runs to 10/minute/team, and admin API endpoints to 60/minute/team. Dry runs are still billed and persist findings, even though they do not write to the code host. These are API limits, not review allowances. Reviews first consume included usage (Individuals) or on-demand spend (Teams); Autofix uses Cloud Agent credits at plan rates.
Is there a trial for Cursor Bugbot?
14-day free trial on all plans (product page).
Does Cursor Bugbot autofix or block merges?
Autofix: Yes. Merge gate: Yes. Automatic review on PR updates or a manual trigger. Findings default to a neutral status, so requiring the check alone does not block merge; enable fail-on-unresolved-issues where available. Autofix can use a Cloud Agent, but requires on-demand usage pricing and storage to be enabled (it is unavailable in Legacy Privacy Mode). Existing-branch autofix is capped at three attempts per PR. Its default model follows the configured agent model, with team/system fallback; effort levels are documented separately.
What is the catch with Cursor Bugbot?
Not a substitute for choosing Cursor the editor — that comparison lives under /ai-harnesses. Usage billing means a busy repo can outrun a flat-seat mental model; size spend from cursor.com/pricing#bugbot, not from a third-party $/run table. Azure DevOps is currently public beta and limited to Azure DevOps Services, not Azure DevOps Server. Requiring the Bugbot check without fail-on-unresolved-issues still allows merge when findings are posted. Autofix needs storage plus on-demand usage pricing, and its available modes vary by code host (GitHub supports more modes than GitLab/Bitbucket; Azure DevOps autofix is unavailable).