claude-code - 💡(How to fix) Fix /ultrareview: counter decremented despite crashed run with no findings [1 comments, 2 participants]

Official PRs (…)
ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

Helpful · Quick feedback

Loading…
GitHub stats
anthropics/claude-code#56054Fetched 2026-05-05 05:59:24
View on GitHub
Comments
1
Participants
2
Timeline
6
Reactions
0
Author
Timeline (top)
labeled ×4commented ×1cross-referenced ×1

Error Message

{"error":"Review crashed before producing findings. See session logs for details."}

  • Output: error JSON above, no findings, no partial results
  • Surface the crash reason to the user (the current error message points at "session logs" the user cannot access).

Fix Action

Workaround

Smaller windows (HEAD~30, ~54 files / ~1.4k lines) completed cleanly in my earlier run, so the crash appears scope-related.

Code Example

{"error":"Review crashed before producing findings. See session logs for details."}
RAW_BUFFERClick to expand / collapse

What happened

I ran /ultrareview HEAD~80 (Run 2 of my 3 free runs). The remote agent crashed before producing any findings:

{"error":"Review crashed before producing findings. See session logs for details."}

Despite the crash producing zero output, my free-run counter still decremented. When I attempted to launch another review, the CLI now reports Free ultrareview 3 of 3. — meaning the failed Run 2 was charged against my quota.

Session details

  • Crashed session ID: r5qhd71nl
  • Trigger: /ultrareview HEAD~80
  • Scope reported by CLI: 148 files changed, 5038 insertions(+), 1499 deletions(-)
  • Output: error JSON above, no findings, no partial results

Expected behavior

Crashed runs that produce no findings should not decrement the free-run counter. The user receives no value from a crashed review, so charging the quota is incorrect.

Suggested fix

  • Refund the counter when the remote agent errors out before producing findings.
  • Surface the crash reason to the user (the current error message points at "session logs" the user cannot access).
  • Consider an upper bound on scope size (or graceful chunking) so large windows like 148 files don't crash silently.

Workaround

Smaller windows (HEAD~30, ~54 files / ~1.4k lines) completed cleanly in my earlier run, so the crash appears scope-related.

extent analysis

TL;DR

The most likely fix is to refund the free-run counter when the remote agent crashes before producing findings, and consider implementing an upper bound on scope size to prevent similar crashes.

Guidance

  • Verify the scope size and complexity of the review by checking the number of files changed and lines inserted/deleted, as reported by the CLI.
  • Test smaller scope sizes (e.g., HEAD~30) to see if the review completes successfully and does not decrement the free-run counter.
  • If possible, investigate the session logs (although currently inaccessible to the user) to determine the root cause of the crash.
  • Consider reporting the issue to the development team to implement a refund mechanism for crashed reviews and improve error handling.

Example

No code snippet is provided as the issue is related to the behavior of the /ultrareview command and the remote agent.

Notes

The suggested fix and workaround are based on the information provided in the issue and may not be comprehensive. The development team's input is necessary to implement a permanent solution.

Recommendation

Apply workaround: use smaller scope sizes (e.g., HEAD~30) to avoid crashes and prevent decrementing the free-run counter until a permanent fix is implemented. This approach allows users to continue using the /ultrareview command while minimizing the impact of the issue.

Vote matrix · Quick signals

Works
Did the solution work? Tap to confirm.
Easy Fix
Was it a quick fix?
Time Saver
Did it save you time?
Blocking
Was it severely blocking?
Common Issue
Are others likely hitting this too?
Flaky / Intermittent
Is it intermittent?
Verified / Reproducible
Can you reproduce it reliably?
Loading…

FAQ

Expected behavior

Crashed runs that produce no findings should not decrement the free-run counter. The user receives no value from a crashed review, so charging the quota is incorrect.

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Back to top recommendations

TRENDING

claude-code - 💡(How to fix) Fix /ultrareview: counter decremented despite crashed run with no findings [1 comments, 2 participants]