PR CURSOR_SUMMARY not refreshed when Bugbot skips re-run after offline/stored review

Support Bug Reports
,
1

Where does the bug appear (feature/product)?

Cloud Agent (GitHub, Slack, Web, Linear)

Describe the Bug

Skipping a full Bugbot re-run when a matching review already exists on Cursor servers (e.g. after local /review-bugbot, or the newer “don’t re-review every push” behavior) is fine. What regressed is that the PR description block … no longer gets rewritten from that stored report / on subsequent pushes.

The overview freezes on an older commit SHA in the footer (“Reviewed by … for commit …”) while the PR tip moves on. Reviewers see a stale Medium Risk / Overview even though Cursor already has (or could cheaply regenerate) summary text for the latest state.

Distinct from Bugbot summary comment not updated after fixes are pushed (closed; that was the findings comment). This is the PR description overview block going stale when re-run is intentionally skipped.

Steps to Reproduce

Enable Bugbot + PR summaries on a repo.
Run Bugbot offline (/review-bugbot) and/or let Bugbot post a CURSOR_SUMMARY on the PR for commit A.
Push commit B that materially changes behavior (not just fixing prior findings).
Observe Bugbot skip a full re-review (expected this week: reuse stored/offline report / run-once / patch-id skip).
Check the PR description CURSOR_SUMMARY block — it still cites commit A / omits commit B’s behavior.

Expected Behavior

Even when Bugbot skips expensive re-analysis, it should still update (or regenerate) the PR description CURSOR_SUMMARY from the stored report / latest PR tip so Overview + reviewed-commit footer stay current. Skipping bug-finding ≠ freezing the description summary forever.

Operating System

Linux

Version Information

IDE:
Version: 3.13.10
VSCode Version: 1.125.0
Commit: 4f02290ccd9304f0e6bf8ee85f6e9106f02ac1f0
CLI:
CLI Version 2026.03.30-a5d3e17

Additional Information

Related closed thread (findings comment, not description summary): Bugbot summary comment not updated after fixes are pushed
Workaround: manually edit the PR body CURSOR_SUMMARY block.

Does this stop you from using Cursor

No - Cursor works, but with this issue

7

Hi @Patrick_Beeson Thank you very much for the post! I tried to replicate this behavior, but unfortunately, I could not. If there’s any chance you can share a request ID from the local /review-bugbot run that would be helpful!

Appreciate you raising this and we are happy to take a closer look.

8

I’m going to hold off on filing a bug report for now but if you see this behavior again, feel free to point us towards a PR and a request ID and we can investigate it further.